LiveLink™ for MATLAB®

User’s Guide
LiveLink™ for MATLAB® User’s Guide
© 2009–2020 COMSOL
Protected by patents listed on www.comsol.com/patents, and U.S. Patents 7,519,518; 7,596,474;
7,623,991; 8,457,932;; 9,098,106; 9,146,652; 9,323,503; 9,372,673; 9,454,625, 10,019,544,
10,650,177; and 10,776,541. Patents pending.
This Documentation and the Programs described herein are furnished under the COMSOL Software License
Agreement (www.comsol.com/comsol-license-agreement) and may be used or copied only under the terms
of the license agreement.
COMSOL, the COMSOL logo, COMSOL Multiphysics, COMSOL Desktop, COMSOL Compiler,
COMSOL Server, and LiveLink are either registered trademarks or trademarks of COMSOL AB. MATLAB
and Simulink are registered trademarks of The MathWorks, Inc.. All other trademarks are the property of
their respective owners, and COMSOL AB and its subsidiaries and products are not affiliated with, endorsed
by, sponsored by, or supported by those or the above non-COMSOL trademark owners. For a list of such
trademark owners, see www.comsol.com/trademarks.
Version: COMSOL 5.6

Contact Information
Visit the Contact COMSOL page at www.comsol.com/contact to submit general
inquiries, contact Technical Support, or search for an address and phone number. You can
also visit the Worldwide Sales Offices page at www.comsol.com/contact/offices for
address and contact information.

If you need to contact Support, an online request form is located at the COMSOL Access
page at www.comsol.com/support/case. Other useful links include:

• Support Center: www.comsol.com/support

• Product Download: www.comsol.com/product-download
• Product Updates: www.comsol.com/support/updates
• COMSOL Blog: www.comsol.com/blogs
• Discussion Forum: www.comsol.com/community
• Events: www.comsol.com/events
• COMSOL Video Gallery: www.comsol.com/video
• Support Knowledge Base: www.comsol.com/support/knowledgebase

Part number: CM020008

C o n t e n t s

Chapter 1: Introduction

About This Product 12

Help and Documentation 14

Getting Help . . . . . . . . . . . . . . . . . . . . . . . . 14
Where Do I Access the Documentation and the Application
Libraries? . . . . . . . . . . . . . . . . . . . . . . . . 18

Chapter 2: Getting Started

The Client/Server Architecture 24

Running COMSOL Models at the Command Line 26

® ® ®
Starting COMSOL with MATLAB on Windows / Mac OSX /
® . . . . . . . . . . . . . . . . . . . . . . . . 26
Linux
®
Connecting a COMSOL Server and MATLAB Manually . . . . . . . 28
Connecting to COMSOL Server™ . . . . . . . . . . . . . . . . 30
®
Changing the MATLAB Version . . . . . . . . . . . . . . . . 32

The COMSOL Apps 34

Installing Apps in the MATLAB Apps Ribbon . . . . . . . . . . . . 34
Removing Apps in the MATLAB Apps Ribbon.. . . . . . . . . . . . 34
The COMSOL Apps . . . . . . . . . . . . . . . . . . . . . 35

Calling External Functions From Within the Model 36

Chapter 3: Building Models

The Model Object 38

Important Notes About the Model Object . . . . . . . . . . . . . 38

CONTENTS |3
 The Model Object Methods. . . . . . . . . . . . . . . . . . . 39
The General Utility Functionality . . . . . . . . . . . . . . . . . 39
The Model History . . . . . . . . . . . . . . . . . . . . . . 40
Loading and Saving a Model . . . . . . . . . . . . . . . . . . . 41
®
Sharing the Model Between the COMSOL Desktop and the
®
MATLAB Prompt . . . . . . . . . . . . . . . . . . . . 46

Working with Geometries 50

The Geometry Sequence Syntax . . . . . . . . . . . . . . . . . 50
Displaying the Geometry . . . . . . . . . . . . . . . . . . . . 51
Working with Geometry Sequences . . . . . . . . . . . . . . . 52
Retrieving Geometry Information . . . . . . . . . . . . . . . . 62
Modeling with a Parameterized Geometry . . . . . . . . . . . . . 64
Images and Interpolation Data . . . . . . . . . . . . . . . . . . 67
Measuring Entities in Geometry . . . . . . . . . . . . . . . . . 75

Working with Meshes 76

The Meshing Sequence Syntax . . . . . . . . . . . . . . . . . . 76
Displaying the Mesh . . . . . . . . . . . . . . . . . . . . . . 77
Mesh Creation Functions . . . . . . . . . . . . . . . . . . . . 78
Importing External Meshes and Mesh Objects . . . . . . . . . . . 103
Measuring Mesh Quality . . . . . . . . . . . . . . . . . . . 105
Getting Mesh Statistics Information . . . . . . . . . . . . . . . 106
Getting and Setting Mesh Data. . . . . . . . . . . . . . . . . 109

Modeling Physics 114

The Physics Interface Syntax . . . . . . . . . . . . . . . . . 114
The Material Syntax . . . . . . . . . . . . . . . . . . . . . 117
Modifying the Equations . . . . . . . . . . . . . . . . . . . 119
Adding Global Equations . . . . . . . . . . . . . . . . . . . 120
Defining Model Settings Using External Data File . . . . . . . . . . 122
Access the User-Defined Physics Interface . . . . . . . . . . . . 123

Creating Selections 125

The Selection Node. . . . . . . . . . . . . . . . . . . . . 125
Coordinate-Based Selections . . . . . . . . . . . . . . . . . 126
Selection Using Adjacent Geometry. . . . . . . . . . . . . . . 130
Displaying Selections . . . . . . . . . . . . . . . . . . . . 131

4 | CONTENTS
 Computing the Solution 134
The Study Node . . . . . . . . . . . . . . . . . . . . . . 134
The Solver Sequence Syntax . . . . . . . . . . . . . . . . . 135
Run the Solver Sequence . . . . . . . . . . . . . . . . . . . 136
Adding a Parametric Sweep . . . . . . . . . . . . . . . . . . 137
Adding a Job Sequence. . . . . . . . . . . . . . . . . . . . 137
Plot While Solving . . . . . . . . . . . . . . . . . . . . . 138

Analyzing the Results 140

The Plot Group Syntax . . . . . . . . . . . . . . . . . . . 140
Displaying The Results . . . . . . . . . . . . . . . . . . . . 141
The Dataset Syntax . . . . . . . . . . . . . . . . . . . . . 148
The Numerical Node Syntax . . . . . . . . . . . . . . . . . 149
Exporting Data . . . . . . . . . . . . . . . . . . . . . . 150
Generating Report . . . . . . . . . . . . . . . . . . . . . 152

Chapter 4: Working With Models

Using Workspace Variables in Model Settings 156

The Set and SetIndex Methods . . . . . . . . . . . . . . . . 156
®
Using a MATLAB Function to Define Model Properties . . . . . . 157

Extracting Results 159

Extracting Data at Arbitrary Points . . . . . . . . . . . . . . . 159
Evaluating a Minimum of Expression. . . . . . . . . . . . . . . 163
Evaluating a Maximum of Expression . . . . . . . . . . . . . . 165
Evaluating an Integral . . . . . . . . . . . . . . . . . . . . 168
Evaluating an Expression Average . . . . . . . . . . . . . . . 170
Extracting Data at Node Points . . . . . . . . . . . . . . . . 173
Evaluating an Expression at Geometry Vertices . . . . . . . . . . 176
Evaluating Expressions on Particle/Ray Trajectories . . . . . . . . . 179
Evaluating a Global Expression . . . . . . . . . . . . . . . . . 180
Evaluating a Matrix Expression at Points . . . . . . . . . . . . . 182
Evaluating a Global Matrix . . . . . . . . . . . . . . . . . . 184
Extracting Data From Tables . . . . . . . . . . . . . . . . . 186

CONTENTS |5
 Running Models in a Loop 187
The Parametric Sweep Node . . . . . . . . . . . . . . . . . 187
®
Running Model in a Loop Using the MATLAB Tools . . . . . . . . 187

Running Models in Batch Mode 190

The Batch Node . . . . . . . . . . . . . . . . . . . . . . 190
Running an M-File in Batch Mode . . . . . . . . . . . . . . . . 190
Running an M-File in Batch Mode Without Display . . . . . . . . . 191

Working with Matrices 192

Extracting System Matrices . . . . . . . . . . . . . . . . . . 192
Set System Matrices in the Model . . . . . . . . . . . . . . . 198
Extracting State-Space Matrices . . . . . . . . . . . . . . . . 204
Extracting Reduced Order State-Space Matrices . . . . . . . . . . 212

Extracting Solution Information and Solution Vectors 220

Obtaining Solution Information . . . . . . . . . . . . . . . . 220
Retrieving Solution Information and Solution Datasets Based on
Parameter Values . . . . . . . . . . . . . . . . . . . . 222
Extracting Solution Vector . . . . . . . . . . . . . . . . . . 225

Retrieving Xmesh Information 228

The Extended Mesh (Xmesh) . . . . . . . . . . . . . . . . . 228
Extracting Xmesh Information . . . . . . . . . . . . . . . . . 228

Navigating the Model 231

Navigating the Model Object Using a GUI . . . . . . . . . . . . 231
Navigating The Model Object At The Command Line . . . . . . . . 237
Finding Model Expressions . . . . . . . . . . . . . . . . . . 237
Evaluating the Model Parameters . . . . . . . . . . . . . . . . 239
Getting Feature Model Properties . . . . . . . . . . . . . . . 240
Getting Parameter and Variable Definitions . . . . . . . . . . . . 241
Getting Selection Information . . . . . . . . . . . . . . . . . 241

Handling Errors and Warnings 242

Errors and Warnings . . . . . . . . . . . . . . . . . . . . 242
Using MATLAB® Tools to Handle COMSOL® Exceptions . . . . . . 242
Displaying Warnings and Errors in the Model . . . . . . . . . . . 242

6 | CONTENTS
 Improving Performance for Large Models 244
Allocating Memory . . . . . . . . . . . . . . . . . . . . . 244
Disabling Model Feature Update . . . . . . . . . . . . . . . . 245
Disabling The Model History . . . . . . . . . . . . . . . . . 246

Creating a Custom GUI 247

Chapter 5: Calling External Functions

Running External Function 250

Allowing External MATLAB Functions . . . . . . . . . . . . . . 250
®
Disabling MATLAB Splash Screen at Start-Up . . . . . . . . . . 251
®
Running a MATLAB Function in Applications . . . . . . . . . . 251

The MATLAB® Function Feature Node 252

® ®
Defining a MATLAB Function in the COMSOL Model . . . . . . 252
® . . . . . . . . 257
Setting the Function Directory Path in MATLAB
® ®
Adding a MATLAB Function with the COMSOL API Syntax . . . . 257
Function Input/Output Considerations . . . . . . . . . . . . . 258
Updating Functions . . . . . . . . . . . . . . . . . . . . . 259
Defining Function Derivatives . . . . . . . . . . . . . . . . . 259

Chapter 6: Command Reference

Summary of Commands 262

Commands Grouped by Function 263

colortable . . . . . . . . . . . . . . . . . . . . . . . . 266
mphaddplotdata . . . . . . . . . . . . . . . . . . . . . . 268
mphapplicationlibraries . . . . . . . . . . . . . . . . . . . 269
mphcd . . . . . . . . . . . . . . . . . . . . . . . . . . 270
mphdoc . . . . . . . . . . . . . . . . . . . . . . . . . 271
mpheval . . . . . . . . . . . . . . . . . . . . . . . . . 271
mphevalglobalmatrix . . . . . . . . . . . . . . . . . . . . 275
mphevalpoint . . . . . . . . . . . . . . . . . . . . . . . 277

CONTENTS |7
 mphevalpointmatrix . . . . . . . . . . . . . . . . . . . . . 279
mphevaluate . . . . . . . . . . . . . . . . . . . . . . . 280
mphgeom. . . . . . . . . . . . . . . . . . . . . . . . . 281
mphgetadj . . . . . . . . . . . . . . . . . . . . . . . . 284
mphgetcoords . . . . . . . . . . . . . . . . . . . . . . . 285
mphgetexpressions . . . . . . . . . . . . . . . . . . . . . 286
mphgetproperties . . . . . . . . . . . . . . . . . . . . . 286
mphgetselection . . . . . . . . . . . . . . . . . . . . . . 287
mphgetu . . . . . . . . . . . . . . . . . . . . . . . . . 288
mphglobal . . . . . . . . . . . . . . . . . . . . . . . . 290
mphimage2geom . . . . . . . . . . . . . . . . . . . . . . 292
mphinputmatrix . . . . . . . . . . . . . . . . . . . . . . 294
mphint2 . . . . . . . . . . . . . . . . . . . . . . . . . 296
mphinterp . . . . . . . . . . . . . . . . . . . . . . . . 298
mphinterpolationfile. . . . . . . . . . . . . . . . . . . . . 304
mphlaunch . . . . . . . . . . . . . . . . . . . . . . . . 305
mphload . . . . . . . . . . . . . . . . . . . . . . . . . 306
mphmatrix . . . . . . . . . . . . . . . . . . . . . . . . 307
mphmax . . . . . . . . . . . . . . . . . . . . . . . . . 311
mphmean. . . . . . . . . . . . . . . . . . . . . . . . . 314
mphmeasure . . . . . . . . . . . . . . . . . . . . . . . 316
mphmesh . . . . . . . . . . . . . . . . . . . . . . . . . 317
mphmeshstats . . . . . . . . . . . . . . . . . . . . . . . 319
mphmin . . . . . . . . . . . . . . . . . . . . . . . . . 321
mphmodel . . . . . . . . . . . . . . . . . . . . . . . . 324
mphnavigator . . . . . . . . . . . . . . . . . . . . . . . 325
mphopen . . . . . . . . . . . . . . . . . . . . . . . . . 327
mphparticle . . . . . . . . . . . . . . . . . . . . . . . . 328
mphplot . . . . . . . . . . . . . . . . . . . . . . . . . 330
mphquad2tri . . . . . . . . . . . . . . . . . . . . . . . 332
mphray . . . . . . . . . . . . . . . . . . . . . . . . . 333
mphreadstl . . . . . . . . . . . . . . . . . . . . . . . . 334
mphreduction . . . . . . . . . . . . . . . . . . . . . . . 335
mphreport . . . . . . . . . . . . . . . . . . . . . . . . 337
mphsave . . . . . . . . . . . . . . . . . . . . . . . . . 338
mphsearch . . . . . . . . . . . . . . . . . . . . . . . . 339
mphselectbox . . . . . . . . . . . . . . . . . . . . . . . 339
mphselectcoords . . . . . . . . . . . . . . . . . . . . . . 341

8 | CONTENTS
mphshowerrors . . . . . . . . . . . . . . . . . . . . . . 343
mphsolinfo . . . . . . . . . . . . . . . . . . . . . . . . 344
mphsolutioninfo . . . . . . . . . . . . . . . . . . . . . . 346
mphstart . . . . . . . . . . . . . . . . . . . . . . . . . 348
mphstate . . . . . . . . . . . . . . . . . . . . . . . . . 349
mphsurf . . . . . . . . . . . . . . . . . . . . . . . . . 352
mphtable . . . . . . . . . . . . . . . . . . . . . . . . . 353
mphtags . . . . . . . . . . . . . . . . . . . . . . . . . 354
mphthumbnail . . . . . . . . . . . . . . . . . . . . . . . 355
mphversion . . . . . . . . . . . . . . . . . . . . . . . . 356
mphviewselection . . . . . . . . . . . . . . . . . . . . . 357
mphwritestl . . . . . . . . . . . . . . . . . . . . . . . . 359
mphxmeshinfo . . . . . . . . . . . . . . . . . . . . . . . 360

CONTENTS |9
10 | C O N T E N T S
 1

Introduction

This guide introduces you to LiveLink™ for MATLAB®, which extends your
COMSOL modeling environment with an interface between COMSOL
Multiphysics® and MATLAB®. The COMSOL Multiphysics Programming
Reference Manual provides additional documentation of the COMSOL API.

In this chapter:

• About This Product

• Help and Documentation

11
 About This Product
LiveLink™ for MATLAB® connects COMSOL Multiphysics to the MATLAB scripting
environment. Using this functionality you can do the following:

Set Up Models from a Script

LiveLink™ for MATLAB® includes the COMSOL API, which has all the necessary
functions and methods to implement models from scratch. For each operation done in
the COMSOL Desktop there is a corresponding command that is entered at the
MATLAB prompt. It is a simplified syntax based on Java® and does not require any
Java knowledge. The easiest way to learn this syntax is to save the model as an M-file
directly from the COMSOL Desktop. Read more about building a model using the
command line in the section Building Models.

Use MATLAB Functions in Model Settings

Use LiveLink™ for MATLAB® to set model properties with a MATLAB function. For
example, define material properties or boundary conditions as a MATLAB routine that
is evaluated while the model is solved. Read more in Calling External Functions.

Leverage MATLAB Functionality for Program Flow

Use the API syntax together with MATLAB functionality to control the flow of your
programs. For example, implement nested loops using for or while commands,
implement conditional model settings with if or switch statements, or handle
exceptions using try and catch. Some of these operations are described in Running
Models in a Loop and Handling Errors and Warnings.

Analyze Results in MATLAB

The API wrapper functions included make it easy to extract data at the command line.
Functions are available to access results at node points or arbitrary locations. You can
also get low level information about the extended mesh, such as finite element mesh
coordinates, and connection information between the elements and nodes. Extracted
data are available as MATLAB variables ready to be used with any MATLAB function.
See Extracting Results and Retrieving Xmesh Information.

Create Custom Interfaces for Models

Use the MATLAB Guide functionality to create a user-defined graphical interface that
is combined with a COMSOL Multiphysics model. Make your models available for
others by creating graphical user interfaces tailored to expose settings and parameters
of your choice.

12 | CHAPTER 1: INTRODUCTION
Connect to COMSOL Server ™
LiveLink™ for MATLAB® has the ability to connect to COMSOL Server™ as well as
COMSOL Multiphysics Server. This means that MATLAB scripts and GUIs that
utilize COMSOL functionality can be distributed to and used by any user that have
access to COMSOL Server™.

ABOUT THIS PRODUCT | 13

 Help and Documentation
In this section:

• Getting Help
• Where Do I Access the Documentation and the Application Libraries?

Getting Help
COMSOL Multiphysics and LiveLink™ for MATLAB® have several sources of help and
information.

THE INTRODUCTION TO LIVELINK™ FOR MATLAB®

To get started with LiveLink™, it is recommended that you read the Introduction to
LiveLink™ for MATLAB®. It contains detailed examples about how to get you started
with the product.

ONLINE DOCUMENTATION AND OTHER RESOURCES

• Read this user’s guide to get detailed information about the different parts of the
model object and how these are accessed from MATLAB. In the section Command
Reference the function available for use with LiveLink™ for MATLAB® are
described.
• The COMSOL Multiphysics Programming Reference Manual contains reference
documentation that describes the methods in the model object.

You can access the online documentation directly at the MATLAB prompt using the
function mphdoc.

To open the COMSOL Documentation Help Desk enter:

mphdoc

To open the help window for a specific entry node enter:

mphdoc(node)

where node is the model object node (for instance, model.geom).

To view the help of a specific feature of a node enter:

mphdoc(node,<fname>)

14 | CHAPTER 1: INTRODUCTION
where <fname> is a string defining the feature name in the COMSOL API, for
example, mphdoc(model,'Rectangle').

It is of course also possible to use MATLAB’s own help function such as

help mphinterp

and

doc mphinterp

M-FILES
You can save COMSOL Multiphysics models as M-files. Use the COMSOL Desktop
to get your first model implemented using the COMSOL API.

Set up the model using the graphical user interface, then save the model as an M-file.
Next go to the File menu and select Save, in the save window locate Save as type list
and select Model File for MATLAB (*.m). This generates an M-function that can be run
using COMSOL with MATLAB.

Since version 5.3 a new syntax closer to the structure in the Model Builder is available.
This new syntax includes the component node as in the example below:

model.component(<cTag>).geom(<geomTag>)

In the previous syntax to access the geometry node you need to enter:

model.geom(<geomTag>)

Both versions are fully supported, and the new syntax is used by default when saving a
model in the M-file format. To save a model M-file using the old syntax, you need to
change the preferences settings as described in the steps below:

1 In the COMSOL Desktop, go to the File menu and click Preferences.

2 In the Preferences window, click Methods. In the Code generation settings clear the
option Use component syntax.
3 Click OK.

At the MATLAB prompt use mphsave to save the model object model in the *.m
format as in the command below:

mphsave(model,<filename.m>)

HELP AND DOCUMENTATION | 15

 where <filename.m> is the name of the file with the *.m extension.

The component syntax is not used by default.

To save the model object in the *.m format using the component syntax enter:

mphsave(model,<filename.m>,'component','on');

Models created with older versions than COMSOL 5.3 cannot be saved
using the component syntax unless File>Compact History is used.

16 | CHAPTER 1: INTRODUCTION
THE APPLICATION LIBRARIES WINDOW

Study the LiveLink™ for MATLAB® Application Library

LiveLink™ for MATLAB® includes an Application Library with detailed example
models. Use the function mphapplicationlibraries at the command line to start a
GUI for viewing the Application Libraries.

If you have installed the COMSOL apps in the MATLAB Apps ribbon,
click the COMSOL Application Libraries icon ( ).

User defined library can also be listed in this window. Such a library has to be added
using the COMSOL Desktop.

The following are some models that can help you get started.

Model Examples
• Learn how to activate and deactivate domains alternatively during a transient
analysis. See the model Domain Activation and Deactivation (model name
domain_activation_llmatlab).

HELP AND DOCUMENTATION | 17

 • Homogenization in a Chemical Reactor (model name
homogenization_llmatlab) shows how to simulate a periodic homogenization
process in a space-dependent chemical reactor model. This homogenization
removes concentration gradients in the reactor at a set time interval.
• Convective Heat Transfer with Pseudo-Periodicity (model name
pseudoperiodicity_llmatlab) simulates convective heat transfer in a channel
filled with water. To reduce memory requirements, the model is solved repeatedly
on a pseudo-periodic section of the channel. Each solution corresponds to a
different section, and before each solution step the temperature at the outlet
boundary from the previous solution is mapped to the inlet boundary.
• Temperature Distribution in a Vacuum Flask (model name
vacuum_flask_llmatlab) shows how to use the MATLAB function callback. This
example solves for the temperature distribution inside a vacuum flask with hot
coffee.
• Electrical Heating of a Busbar Solved with LiveLink™ for SOLIDWORKS® and
LiveLink™ for MATLAB® (model name busbar_llsw_llmatlab) performs
geometry optimization using COMSOL Multiphysics, MATLAB, and
SOLIDWORKS®.

Where Do I Access the Documentation and the Application Libraries?

A number of internet resources have more information about COMSOL, including
licensing and technical information. The electronic documentation, topic-based (or
context-based) help, and the application libraries are all accessed through the
COMSOL Desktop.

If you are reading the documentation as a PDF file on your computer, the
blue links do not work to open an application or content referenced in a
different guide. However, if you are using the Help system in COMSOL
Multiphysics, these links work to other modules (as long as you have a
license), application examples, and documentation sets.

THE DOCUMENTATION AND ONLINE HELP

The COMSOL Multiphysics Reference Manual describes all core physics interfaces
and functionality included with the COMSOL Multiphysics license. This book also has
instructions about how to use COMSOL Multiphysics and how to access the
electronic Documentation and Help content.

18 | CHAPTER 1: INTRODUCTION
Opening Topic-Based Help
The Help window is useful as it is connected to many of the features on the GUI. To
learn more about a node in the Model Builder, or a window on the Desktop, click to
highlight a node or window, then press F1 to open the Help window, which then
displays information about that feature (or click a node in the Model Builder followed
by the Help button ( ). This is called topic-based (or context) help.

To open the Help window:

• In the Model Builder, Application Builder, or Physics Builder click a node or

window and then press F1.
• On any toolbar (for example, Home, Definitions, or Geometry), hover the
mouse over a button (for example, Add Physics or Build All) and then
press F1.
• From the File menu, click Help ( ).
• In the upper-right corner of the COMSOL Desktop, click the Help( )
button.

To open the Help window:

• In the Model Builder or Physics Builder click a node or window and then
press F1.
• On the main toolbar, click the Help ( ) button.
• From the main menu, select Help>Help.

Opening the Documentation Window

To open the Documentation window:

• Press Ctrl+F1.
• From the File menu select Help>Documentation ( ).

HELP AND DOCUMENTATION | 19

 To open the Documentation window:

• Press Ctrl+F1.
• On the main toolbar, click the Documentation ( ) button.
• From the main menu, select Help>Documentation.

THE APPLICATION LIBRARIES WINDOW

Each application includes documentation with the theoretical background and
step-by-step instructions to create a model application. The applications are available
in COMSOL as MPH-files that you can open for further investigation. You can use the
step-by-step instructions and the actual applications as a template for your own
modeling and applications. In most models, SI units are used to describe the relevant
properties, parameters, and dimensions in most examples, but other unit systems are
available.

Once the Application Libraries window is opened, you can search by name or browse
under a module folder name. Click to view a summary of the application and its
properties, including options to open it or a PDF document.

The Application Libraries Window in the COMSOL Multiphysics

Reference Manual.

Opening the Application Libraries Window

To open the Application Libraries window ( ):

• From the Home toolbar, Windows menu, click ( ) Applications

Libraries.
• From the File menu select Application Libraries.

To include the latest versions of model examples, from the File>Help

menu, select ( ) Update COMSOL Application Library.

Select Application Libraries from the main File> or Windows> menus.

To include the latest versions of model examples, from the Help menu
select ( ) Update COMSOL Application Library.

20 | CHAPTER 1: INTRODUCTION
CONTACTING COMSOL BY EMAIL
For general product information, contact COMSOL at [email protected].

To receive technical support from COMSOL for the COMSOL products, please
contact your local COMSOL representative or send your questions to
[email protected]. An automatic notification and case number is sent to you by
email.

COMSOL WEBSITES

COMSOL website www.comsol.com

Contact COMSOL www.comsol.com/contact
COMSOL Access www.comsol.com/access
Support Center www.comsol.com/support
Product Download www.comsol.com/product-download
Product Updates www.comsol.com/support/updates
COMSOL Blog www.comsol.com/blogs
Discussion Forum www.comsol.com/community
Events www.comsol.com/events
COMSOL Video Gallery www.comsol.com/video
Support Knowledge Base www.comsol.com/support/knowledgebase

HELP AND DOCUMENTATION | 21

22 | CHAPTER 1: INTRODUCTION
 2

Getting Started

In this chapter:
• The Client/Server Architecture
• Running COMSOL Models at the Command Line
• The COMSOL Apps
• Calling External Functions From Within the Model

23
 The Client/Server Architecture
LiveLink™ for MATLAB® uses the client/server mode to connect a COMSOL server
and MATLAB.

The term COMSOL server designates either the COMSOL Multiphysics

Server or the COMSOL Server™.

When starting COMSOL with MATLAB, two processes are started — a COMSOL
Multiphysics Server and the MATLAB desktop. The COMSOL Desktop does not have
to be started, but it possible to have both MATLAB and COMSOL Desktop
connected to the same COMSOL Multiphysics Server at the same time.

COMSOL Desktop cannot connect to COMSOL Server™.

The communication between the two processes is based on a TCP /IP communication
protocol. You provide login information the first time COMSOL is started with
MATLAB. The login information is not related to the system’s username and
password. This information is stored in the user preferences file and is not required
again when using COMSOL with MATLAB. The same login information can be used
when exchanging the model object between the COMSOL server and the COMSOL
Desktop.

The communication between the COMSOL server and MATLAB is established by

default using port number 2036. If this port is in use, port number 2037 is used
instead, and so on.

You can manually specify the port number. See the COMSOL
Multiphysics Installation Guide for more information on the COMSOL
server start-up properties.

24 | CHAPTER 2: GETTING STARTED

 You can manually specify the port number. See the COMSOL
Multiphysics Installation Guide for more information on the COMSOL
server start-up properties.

A connection can be local (on the same computer), which is the common case, or
remote to a COMSOL server located on a different computer, in the later case you to
connect manually MATLAB to the COMSOL server as described in the section
Connecting a COMSOL Server and MATLAB® Manually.

THE CLIENT/SERVER ARCHITECTURE | 25

 Running COMSOL Models at the
Command Line
The command to run COMSOL with MATLAB® automatically connects a COMSOL
process with MATLAB. You can also connect the process manually. This section
describes this process as well as how to change the MATLAB path in the COMSOL
settings.

The System Requirements section in the COMSOL Multiphysics

Installation Guide lists the versions of MATLAB supported by
LiveLink™ for MATLAB®.

In this section:

• Starting COMSOL® with MATLAB® on Windows ®/ Mac OSX / Linux®

• Connecting a COMSOL Server and MATLAB® Manually
• Changing the MATLAB® Version

Starting COMSOL® with MATLAB® on Windows ®/ Mac OSX /

Linux®
To run a COMSOL Multiphysics model at the MATLAB® prompt, start COMSOL
with MATLAB:

• On Windows® use the COMSOL Multiphysics with MATLAB shortcut icon that is
created on the desktop after the automatic installation. A link is also available in the
Windows start menu under COMSOL Multiphysics 5.6>COMSOL Multiphysics 5.6 with
MATLAB.
• On Mac OS X, use the COMSOL Multiphysics 5.6 with MATLAB application available in
the Application folder.
• On Linux®, enter the command comsol mphserver matlab at a terminal window.

See the COMSOL Multiphysics Installation Guide for a complete

description about how to start COMSOL with MATLAB on these
supported platforms.

26 | CHAPTER 2: GETTING STARTED

 The first time COMSOL Multiphysics with MATLAB is started, login and
password information is requested to establish the client/server
connection. The information is saved in the user preference file and is not
required again.

Launching COMSOL Multiphysics with MATLAB enables LiveLink™ for

Simulink® as long as it is installed on the machine.

To reset the login information, add the flag -login force to the icon
target path.

To reset the login information, enter the command comsol mphserver

matlab -login force at a system command prompt.

RUNNING A MODEL M-FILE FROM TERMINAL PROMPT

If you want to run an M-file directly at a terminal prompt immediately after having
started COMSOL Multiphysics with MATLAB enter the startup command as described
below:

comsolmphserver.exe matlab filename

comsol mphserver matlab filename

RUNNING COMSOL MODELS AT THE COMMAND LINE | 27

 where filename.m is the file containing both MATLAB and COMSOL API
command to be executed using COMSOL with MATLAB.

RUNNING WITHOUT DISPLAY

If you need to run COMSOL with MATLAB on a machine without support for
graphics display, add the flags nodesktop and mlnosplash to the startup command as
described below:

comsolmphserver.exe matlab -nodesktop -mlnosplash

comsol mphserver matlab -nodesktop -mlnosplash

The above command starts MATLAB without splash screen and without the MATLAB
desktop.

To avoid the splash screen on the COMSOL server you need to create the
environment variable COMSOL_MATLAB_INIT and set it to matlab -nosplash.

COMSOL_MATLAB_INIT=’matlab -nosplash’

Connecting a COMSOL Server and MATLAB® Manually

Manually connecting MATLAB® to a COMSOL server can be useful if you want to
start a MATLAB standalone and then connect to a COMSOL server, or if you need to
connect MATLAB and a COMSOL server running on different computers.
LiveLink™ for MATLAB® provides the function mphstart to operate the client/server
connection. This section contains the instruction to follow to connect MATLAB to
either a COMSOL Multiphysics Server or the COMSOL Server™.

28 | CHAPTER 2: GETTING STARTED

CONNECTING MATLAB TO A COMSOL MULTIPHYSICS SERVER

Starting a COMSOL Multiphysics Server

• On Windows, click COMSOL Multiphysics Server in the COMSOL Launchers folder
underneath your COMSOL Multiphysics folder on the Windows Start menu.
• On Mac OS X or Linux enter comsol mphserver at a terminal window.

Connecting MATLAB to the COMSOL Multiphysics Server

1 In MATLAB, add the path of the COMSOL5.6/mli directory.
2 Enter this command at the MATLAB prompt:
mphstart

If the COMSOL Multiphysics Server started listening to a different port than the
default one (which is 2036) use the mphstart function as in the command below:

mphstart(<portnumber>)

where <portnumber> is an integer corresponding to the port used by the COMSOL

server.

CONNECTING MATLAB AND A SERVER ON DIFFERENT COMPUTERS

Connecting MATLAB and a COMSOL Multiphysics Server requires a

Floating Network License (FNL).

To connect MATLAB and a COMSOL server that are running on different

computers, specify the IP address of the computer where the COMSOL server is
running in the function mphstart:

mphstart(<ipaddress>, <portnumber>)

<ipaddress> can also be defined with the COMSOL server domain name.

The command above assume that the same user login information are set on the server
and client machine. In case the login information are not accessible from the client
machine, specify manually the user name and password to the COMSOL server with
the command:

mphstart(<ipaddress>, <portnumber>, <username>, <password>)

If the COMSOL Multiphysics installation folder cannot be found automatically, you

can specify its location manually as in the command below:

mphstart(<ipaddress>, <portnumber>, <comsolpath>)

RUNNING COMSOL MODELS AT THE COMMAND LINE | 29

 where <comsolpath> is the path of the COMSOL installation folder.

You can also specify all the information to connect a COMSOL server within the same
command, use the following command:

mphstart(<ipaddress>, <portnumber>, <comsolpath>, ...

<username>, <password>)

MEMORY SETTINGS
To be able to manipulate the model object and extract data at the MATLAB prompt,
you may need to modify the Java® heap size in MATLAB. See Improving Performance
for Large Models.

IMPORTING THE COMSOL CLASS

Once MATLAB and the COMSOL server are manually connected, import the
COMSOL class by entering the following command at the MATLAB prompt:

import com.comsol.model.util.*

Disconnecting MATLAB and the COMSOL server

To disconnect MATLAB and the COMSOL server, run this command at the
MATLAB prompt:

ModelUtil.disconnect;

Connecting to COMSOL Server™

When using a COMSOL Multiphysics installation with LiveLink for MATLAB the
connected is made between MATLAB and a COMSOL Multiphysics Server. It is also
possible to connect MATLAB to a COMSOL Server™ if a COMSOL Server™ is
available with a LiveLink for MATLAB License.

When using LiveLink for MATLAB with COMSOL Server™ this way MATLAB is
installed on the local computer where the user is executing commands and COMSOL
Server™ will most often be on another computer that is handled by an IT-department.

In order to be able to connect from MATLAB to a COMSOL Server™ without having

COMSOL Multiphysics installed some files must be installed on the computer where
MATLAB is installed. This is handled by using the COMSOL Server™ Client installer.

30 | CHAPTER 2: GETTING STARTED

When installing the COMSOL Server™ Client it is important to select LiveLink™ for
MATLAB© in the installation window.

It is recommended to install Desktop and Start menu shortcuts. If these are not
installed, the only way to create a link between MATLAB and COMSOL Server™ is
to use the mphstart command as previously described for use with COMSOL
Multiphysics Server.

After the installation a shortcut is available.

When clicking on this shortcut MATLAB will start and a dialog box appears that makes
it possible to connect to the COMSOL Server™ that is already assumed running.

Fill in the missing information and click OK to connect.

RUNNING COMSOL MODELS AT THE COMMAND LINE | 31

 After a connection has been made the LiveLink work in the same way as it does when
connected to a COMSOL Multiphysics Server with these exceptions:

• Some graphical user interfaces that are included in a regular COMSOL Multiphysics
license do not work
• Plot on the server is not supported, use mphplot to plot on a local MATLAB figure.

Changing the MATLAB® Version

The path of the MATLAB® version connected to COMSOL Multiphysics is defined
during the initial COMSOL installation. The MATLAB root path can be changed
using the Preferences dialog box:

1 From the File (Windows users) or Options menu (Mac and Linux users), select
Preferences ( ).
2 In the Preferences dialog box, click LiveLink Connections.
3 Set the MATLAB root directory path in the MATLAB® installation folder field.
4 Windows OS users also need to click the Register MATLAB® as COM Server button;
otherwise, the specified MATLAB version does not start when calling external
MATLAB functions from the COMSOL model.
5 Click OK.

6 To update the preferences file, close and reopen the COMSOL Desktop.

32 | CHAPTER 2: GETTING STARTED

On Mac OS X, select the COMSOL with MATLAB application available in the
Application folder. The correct path includes the .app extension.

RUNNING COMSOL MODELS AT THE COMMAND LINE | 33

 The COMSOL Apps
Install apps in the MATLAB desktop for an easy access to COMSOL information and
navigation functions that use Graphical User Interfaces.

In the section the term COMSOL apps designates COMSOL wrapper

function using a Graphical User Interface. It does not refer to application
created using the COMSOL Application Builder.

Installing Apps in the MATLAB Apps Ribbon

The automatic COMSOL installation does not include installation of the COMSOL
apps in the MATLAB Apps Ribbon. To install the apps follow the steps below:

1 In the MATLAB Desktop, go to the Apps Tab and select Install App.
2 Browse to the COMSOL Installation directory and go to the folder:
COMSOL56/Multiphysics/mli/toolbox
3 Change File name extension to All Files (*.*) and select the file
LiveLink for MATLAB.mltbx
4 Click Open. This opens the Install LiveLink for MATLAB window.
5 The installed apps are now listed in the Add-On Manager window.

Removing Apps in the MATLAB Apps Ribbon.

To remove Apps from the MATLAB Apps Ribbon, right-click on the apps icon and
select Uninstall. The operation is individual for each apps and need to be repeated for
every apps to be removed.

34 | CHAPTER 2: GETTING STARTED

The COMSOL Apps
The available COMSOL apps that can be installed in the MATLAB Apps ribbon are
listed below:

• COMSOL Model Library, opens a GUI for viewing the Model Library, see also The
Application Libraries Window.
• COMSOL Model Navigator, opens a GUI for viewing the COMSOL model object
defined as model in MATLAB. You can get more information in the section
Navigating the Model Object Using a GUI.
• COMSOL Open, opens a GUI for opening recent files, see also Loading a Model
from a List of Existing files.
• COMSOL Search, opens a GUI for searching for expressions in the COMSOL
model object defined as model in MATLAB. See also Finding Model Expressions.

To run these apps you need a connection between MATLAB and a

COMSOL server, either using COMSOL with MATLAB or using
manual connection.

THE COMSOL APPS | 35

 Calling External Functions From
Within the Model
Use LiveLink™ for MATLAB® to call MATLAB functions from within the model — for
instance, when working in the COMSOL Desktop. The procedure is different than
implementing a model using a script as you do not need to run COMSOL with
MATLAB.

Start COMSOL Multiphysics as a standalone application. The external MATLAB

function needs to be defined in the COMSOL model so that a MATLAB process can
automatically start when the function needs to be evaluated. The result of the function
evaluation in MATLAB is then sent back to the COMSOL environment.

Calling External Functions

To run a MATLAB function, enable Allow external MATLAB® functions in

the Preferences window; see Allowing External MATLAB Functions for
more information.

36 | CHAPTER 2: GETTING STARTED

 3

Building Models

This chapter gives an overview of the model object and provides an introduction
to building models using the LiveLink™ interface.

In this chapter:

• The Model Object

• Working with Geometries
• Working with Meshes
• Modeling Physics
• Creating Selections
• Computing the Solution
• Analyzing the Results

37
 The Model Object
While working with the LiveLink™ interface in MATLAB® you work with models
through the model object. Use methods to create, modify, and access models.

In this section:

• Important Notes About the Model Object

• The Model Object Methods
• The General Utility Functionality
• The Model History
• Loading and Saving a Model
• Sharing the Model Between the COMSOL Desktop® and the MATLAB® Prompt

Detailed documentation about model object methods is in About

General Commands in the COMSOL Multiphysics Programming
Reference Manual.

Important Notes About the Model Object

Consider the following information regarding the model object:

• All algorithms and data structures for the model are integrated in the model object.
• The model object is used by the COMSOL Desktop to represent your model. This
means that the model object and the COMSOL Desktop behavior are virtually
identical.
• The model object includes methods to set up and run sequences of operations to
create geometry, meshes, and to solve your model.

LiveLink™ for MATLAB® includes the COMSOL API, which is a programming

interface based on Java®. In addition, the product includes a number of M-file utility
functions that wrap API functionality for greater ease of use.

The Model Object in the COMSOL Multiphysics Programming

Reference Manual.

38 | CHAPTER 3: BUILDING MODELS

The Model Object Methods
The model object has a large number of methods. The methods are structured in a
tree-like way, very similar to the nodes in the model tree in the Model Builder window
on the COMSOL Desktop. The top-level methods just return references that support
further methods. At a certain level the methods perform actions, such as adding data
to the model object, performing computations, or returning data.

Detailed documentation about model object methods is in About

General Commands in the COMSOL Multiphysics Programming
Reference Manual.

The General Utility Functionality

The model object utility methods are available with the ModelUtil object. These
methods can be used, for example, to create or remove a new model object, but also
to enable the progress bar or list the model object available in the COMSOL server.

MANAGING THE COMSOL MODEL OBJECT

Use the method ModelUtil.create to create a new model object in the COMSOL
server:

model = ModelUtil.create(<ModelTag>);

This command creates a model object Model on the COMSOL server and a MATLAB
object model that is linked to the model object <ModelTag> in the COMSOL server.

It is possible to have several model objects on the COMSOL server, each with a
different name. To access each model object requires different MATLAB variables
linked to them and each MATLAB variable must have a different name.

Create a MATLAB variable linked to an existing model object with the function
ModelUtil.model. For example, to create a MATLAB variable model that is linked to
the existing model object <ModelTag> on the COMSOL server, enter the command:

model = ModelUtil.model(<ModelTag>);

Alternatively you can use the function mphload as in the command below:

model = mphload(<ModelTag>);

THE MODEL OBJECT | 39

 To remove a specific model object use the method ModelUtil.remove. For example,
to remove the model object <ModelTag> from the COMSOL server enter the
command:

ModelUtil.remove(<ModelTag>);

Alternatively remove all the COMSOL objects stored in the COMSOL server with the
command:

ModelUtil.clear

List the names of the model objects available on the COMSOL server with the
command:

mphtags -show

ACTIVATING THE PROGRESS BAR

By default no progress information is displayed while running COMSOL with
MATLAB. To manually enable a progress bar and visualize the progress of operations
(such as loading a model, creating a mesh, assembling matrices, or computing the
solution), enter the command:

ModelUtil.showProgress(true)

To deactivate the progress bar enter:

ModelUtil.showProgress(false)

Mac OS X does not support the progress bar.

The Model History

The model contains its entire modeling history corresponding to every settings added
once to the model. When you save a model as an M-file, you get all the operations
performed to the model, including settings that are no longer part of the model.

Using the model history is a convenient way to learn the COMSOL API.
The latest settings enter in the command Desktop being listed at the end
of the M-file.

40 | CHAPTER 3: BUILDING MODELS

The model history is automatically enabled when the model is created in the
COMSOL Desktop. It is however possible to manually disable the model history
recording from the MATLAB prompt with the command:

model.hist.disable

The functions mphload and mphopen automatically disable model history

when loading a model.

To enable the model history, enter the command:

model.hist.enable

COMPACTING THE MODEL HISTORY

To clean the M-file for the model so that it contains only the settings that are part of
the current model you need to compact the model history before saving the model as
an M-file.

To compact the model history in the COMSOL Desktop, from File menu (Windows
users) or from the toolbar (Mac and Linux users), select Compact History ( ).

To compact the model history at the MATLAB prompt enter the command:

model.resetHist

Loading and Saving a Model

LOADING A MODEL AT THE MATLAB PROMPT

To load an existing model saved as an MPH-file use the function mphopen. To load the
model with the name <filename> enter:

model = mphopen(<filename>)

where <filename> is a string. This creates a model object Model on the COMSOL
server that is accessible using the MATLAB variable model.

A shorter form is to simply use

mphopen <filename>

THE MODEL OBJECT | 41

 that will load the model with the given filename and use the variable name model for
accessing the model later. Any existing variable model will be overwritten without
warning.

The function mphload can also be used with the same property. In the
following documentation the commands also work with mphload.

mphload does not store the filename in the recent file list as mphopen does
by default.

Once the model is loaded, the file name and its associated model object tag are
displayed in the COMSOL server window.

If there is already a model object Model in the COMSOL server, mphopen overrides
the existing model object unless the model is also open in a COMSOL Multiphysics
Client. In the later case, an index number is appended to the new model object name,
for instance Model1.

mphopen and mphload do not look for lock file when opening a model in
the COMSOL server.

If you want to manually specify the model object in the COMSOL server. use the
command:

model = mphopen(<filename>, <ModelTag>);

where <ModelTag> is a string defining the tag that defines the loaded model in the
COMSOL server.

When using the function mphopen, the model history is automatically disabled to
prevent large history information when running a model in a loop. To turn model
history on, use the function mphopen:

model = mphopen(<filename>,'-history');

The history recording can be useful when using the COMSOL Desktop.
All the operations are then stored in the saved M-file.

42 | CHAPTER 3: BUILDING MODELS

If you do not want to update the recent opened file list with the model you are about
to open, use the -nostore flag with the function mphopen as in the command below:

model = mphopen(<filename>, <ModelTag>, '-nostore')

If the model mph-file is protected using a password, use mphload as in the command
below:

model = mphopen(<filename>, <ModelTag>, <password>)

where <password> is a string defining the password protecting the file.

If you want to get the full filename of the loaded file, add a second output as in the
command below:

[model, filenameloaded]= mphopen(<filename>, ...)

LOADING A MODEL FROM A LIST OF EXISTING FILES

You can use a GUI where to load the model from a list files corresponding to the recent
opened file or the files in a specified directory.

At the MATLAB prompt enter the command:

mphopen

THE MODEL OBJECT | 43

 This starts a GUI with a list of the recent opened files.

If you have installed the COMSOL apps in the MATLAB Apps ribbon,
click the COMSOL Open icon ( ).

For each selected files, the model information is available in the File Info section.

Click the Recent button to get the list of the recent opened file. Click the Search button
to search for a file using file pattern. Click the Browse button to browse the directory
where to get filename list.

To clear the recent opened file list enter the command:

mphopen -clear

To open the GUI with the list of files in a specific directory (<dirpath>), enter the
command:

mphopen -dir <dirpath>

SAVING A MODEL
Use the function mphsave to save the model object linked to the MATLAB object
model:

44 | CHAPTER 3: BUILDING MODELS

 mphsave(model,<filename>)

where <filename> is a string. If the filename specified <filename> does not provide
a path, the file is saved relative to the current MATLAB path. The file extension
determines the format to use (*.mph, *.m, *.java, or *.vba).

Alternatively, use the save method:

model.save(<filename>)

If <filename> does not provide a path, the file is saved relative to the local server path.

Any files saved in the MPH format can be loaded by COMSOL Desktop. In addition,
the model can be saved as an M-file:

model.save(<filename>,'m')

When saving the model as an M-file mphsave does not automatically use
the component syntax to save model using the COMSOL API, to enable
this syntax run the command below:

mphsave(model,<filename>,'component','on')

The models are not automatically saved between MATLAB sessions.

SET A MODEL THUMBNAIL

Before saving your model, you may want to include a model thumbnail to quickly
identify your model in your own Application Library or when using mphopen. To set
the model thumbnail enter the command:

mphthumbnail(model,<filename>)

where <filename> is the image file name.

You can also use a MATLAB figure to set the thumbnail. The following command will
set the thumbnail to the image of the current figure:

mphthumbnail(model,gcf)

Note that the thumbnail is stored in memory. In order to save the thumbnail in the
model file the model must be saved.

You can extract the image and image filename for the thumbnail stored in model, enter
the command:

THE MODEL OBJECT | 45

 [image, imagefilename] = mphthumbnail(model)

Example
The code below shows how to get the model thumbnail as MATLAB image data, show
the image in a MATLAB figure and store the new image as thumbnail in the model.

mphopen model_tutorial_llmatlab
im = mphthumbnail(model);
imshow(im)
mphthumbnail(model, gcf)

Sharing the Model Between the COMSOL Desktop® and the

MATLAB® Prompt
It is possible to connect a COMSOL Desktop to the COMSOL Multiphysics Server
that is already connected with MATLAB and then access the model from both client
(the COMSOL Desktop and MATLAB). The change performed from either client are
directly accessible from the other one; for instance, type a command at the MATLAB
prompt and see the resulting modification in the Model Builder window, or extract
data at the MATLAB prompt from a model set up in the COMSOL Desktop.

CONNECT THE COMSOL DESKTOP TO THE COMSOL MULTIPHYSICS

SERVER FROM THE PROMPT
At the prompt call mphlaunch to start a COMSOL Desktop, connect it to the same
COMSOL Multiphysics Server to which MATLAB is connected to, and import a
model.

Run the command below:

mphlaunch

This starts a new COMSOL Desktop, connect it to the COMSOL Multiphysics Server
that is already connected with MATLAB, and import the model available in the server.
In case several model are available in the server you can specify which one to import
by running the command below:

mphlaunch ModelTag

where ModelTag is the tag of the model to import.

List the tags of the application loaded in the server with the command
mphtags.

46 | CHAPTER 3: BUILDING MODELS

You can also specify the MATLAB object name that is link to the application to be
imported in the COMSOL Desktop, enter the command:

mphlaunch(model)

If a COMSOL Multiphysics client is already connected to the COMSOL Multiphysics

Server you will be asked to disconnect the connected client and connect the new one
or cancel the operation.

mphlaunch sets automatically a timeout to make MATLAB wait 0.5 second until the
COMSOL server is free again. If you need to increase the timeout run the command
below:

mphlaunch(model,<timeout>)

where <timeout> is the time in milliseconds to wait for the server to be free again.

CONNECT THE COMSOL DESKTOP TO THE COMSOL MULTIPHYSICS

SERVER
Connect the COMSOL Desktop to a COMSOL Multiphysics Server manually using
the Connect to Server dialog box:

1 From the File (Windows users) or Options menu (Mac and Linux users), select
COMSOL Multiphysics Server>Connect to Server ( ).
2 In the Connect to Server window, you specify the Server configuration and the user
settings. In the Server section enter the COMSOL Multiphysics Server name (the
default name is localhost) and the Port number (the default is 2036). This number
corresponds to the port that the COMSOL Multiphysics Server is listening to, the
number is displayed at the COMSOL Multiphysics Server window.
3 In the User section enter a Username and a Password (if they are empty); these are
defined the first time you are connected to the COMSOL Multiphysics Server.

THE MODEL OBJECT | 47

 4 Click OK.

The first time you connect the COMSOL Desktop to the COMSOL
Multiphysics Server no model is loaded to the GUI. See Import An
application from the COMSOL Multiphysics Server to the COMSOL
Desktop to know how connect the GUI to a model loaded in the
COMSOL Multiphysics Server.

IMPORT AN APPLICATION FROM THE COMSOL MULTIPHYSICS SERVER

TO THE COMSOL DESKTOP
Once you have the COMSOL Desktop connected to the COMSOL Multiphysics
Server you can import the model in the GUI:

1 From the File (Windows users) or Options menu (Mac and Linux users), select
COMSOL Multiphysics Server>Import Application from Server ( ).
2 In the Import Application from Server window, specify the application you want to
import.

IMPORT A MODEL FROM THE COMSOL MULTIPHYSICS SERVER TO

MATLAB
To access a model stored in the COMSOL Multiphysics Server from the MATLAB
prompt enter the command:

model = mphload(<ModelTag>)

where model is the variable in MATLAB used to access the model stored on the
COMSOL server and <ModelTag> is the tag of the COMSOL Model.

You can get the list of the models stored in the COMSOL Multiphysics Server with
the command:

mphtags -show

48 | CHAPTER 3: BUILDING MODELS

Set up a time-out in MATLAB
To prevent MATLAB sending command to the COMSOL Multiphysics Server while
it is busy to update the COMSOL Desktop, you need to set up a time-out in MATLAB
and specify how long to wait the COMSOL Multiphysics Server to be free again. Enter
the command:

ModelUtil.setServerBusyHandler(ServerBusyHandler(<timeout>))

Where <timeout> is the time in millisecond to wait the server to be free again.

THE MODEL OBJECT | 49

 Working with Geometries
This section describes how to set up and run a geometry sequence. In this section:

• The Geometry Sequence Syntax

• Displaying the Geometry
• Working with Geometry Sequences
• Retrieving Geometry Information
• Modeling with a Parameterized Geometry
• Images and Interpolation Data
• Measuring Entities in Geometry

• Geometry Modeling and CAD Tools in the COMSOL Multiphysics

Reference Manual
• Geometry in the COMSOL Multiphysics Programming Reference
Manual

The Geometry Sequence Syntax

In the COMSOL Multiphysics Programming Reference Manual:

• For a list of geometry operations, see About Geometry Commands.

• For a property list available for the geometry features see Geometry.

Create a geometry node using the syntax:

model.component(<ctag>).geom.create(<geomtag>, sdim)

where <geomtag> is a string used to refer to the geometry and <ctag> is the string
defined when the component is created. The integer sdim specifies the space
dimension of the geometry and it can be either 0, 1, 2, or 3.

To add an operation to a geometry sequence, use the syntax:

geometry.feature.create(<ftag>, operation)

where geometry is a link to the geometry node. The string <ftag> is used to refer to
the operation.

50 | CHAPTER 3: BUILDING MODELS

To set the feature property with different values than the default, use the set method:

geometry.feature(<ftag>).set(property, <value>)

where <ftag> is the string defined when creating the operation.

To build the geometry sequence, enter:

geometry.run

Alternatively, to build the geometry sequence up to and including a given feature ftag
enter:

geometry.run(<ftag>)

Displaying the Geometry

Use the function mphgeom to display the geometry in a MATLAB figure:

mphgeom(model)

To specify the geometry to display, enter:

mphgeom(model, <geomtag>)

where <geomtag> is the tag of the geometry node to display. If the model only
contains a single geometry the tag <geomtag> can be left empty. When specifying a
property the geometry tag is required.

Adding a view property will add some view settings from the COMSOL model such
as axes labels (units) and grid and supports hiding of geometric entities. Usually it is
sufficient to use the auto value for the view property:

mphgeom(model, <geomtag>, 'view', 'auto')

When running mphgeom the geometry node is automatically built. Set the build
property to specify how the geometry node is supposed to be built before displaying
it. Enter:

mphgeom(model, <geomtag>, 'build', build)

where build is a string with the value: 'off', 'current', or the geometry feature tag
<ftag>, which, respectively, does not build the geometry (off), builds the geometry
up to the current feature (current), or builds the geometry up to the specified
geometry feature node (ftag).

If the geometry contains workplane, you can plot the geometry entities inside a
specified workplane as a 2D geometry. Enter:

WORKING WITH GEOMETRIES | 51

 mphgeom(model, <geomtag>, 'workplane', <wptag>)

where <wptag> is the tag of the workplane to use. It is also possible to combine the
workplane geometry display with the build property to display the geometry built up
to a certain feature.

Use the parent property to specify the axes handle where to display the plot:

mphgeom(model, <geomtag>, 'parent', <axes>)

The following properties are also available to specify the vertex, edge, or face
rendering:

• vertexmode • edgecolor
• edgemode • facecolor
• facemode • vertexlabelscolor
• vertexlabels • edgelabelscolor
• edgelabels • facelabelscolor
• facelabels • domainlabelscolor
• domainlabels • facealpha

Use mphgeom to display a specified geometric entity. To set the geometric entity, enter
the entity property and set the geometric entity index in the selection property to:

mphgeom(model, <geomtag>, 'entity', entity, 'selection', <idx>)

where entity can be either 'point', 'edge', 'boundary', or 'domain', and <idx>
is a positive integer array that contains the list of the geometric entity indices.

You can get the handle of the plotted geometry entities with the command:

h = mphgeom(model, <geomtag>, ...)

where h is a Patch array of the plotted entities.

Working with Geometry Sequences

This section shows how to create geometry sequences using the syntax outlined in The
Geometry Sequence Syntax. This section has these examples:

• Creating a 1D Geometry
• Creating a 2D Geometry Using Primitive Geometry Objects

52 | CHAPTER 3: BUILDING MODELS

• Creating a 2D Geometry Using Boundary Modeling
• Creating a 3D Geometry Using Solid Modeling

For more information about geometry modeling, see the Geometry

chapter in the COMSOL Multiphysics Reference Manual.

CREATING A 1D GEOMETRY
From the MATLAB command prompt, create a 1D geometry model by adding a
geometry sequence and then adding geometry features. The last step is to run the
sequence using the run method.

First create a model object:

model = ModelUtil.create('Model');

Then continue with the commands:

model.component.create('comp1',true);

geom1 = model.component('comp1').geom.create('geom1',1);

i1 = geom1.feature.create('i1','Interval');
i1.set('intervals','many');
i1.set('p','0,1,2');

geom1.run;

This creates a geometry sequence with a 1D solid object consisting of vertices at x = 0,

1, and 2, and edges joining the vertices adjacent in the coordinate list.

Then enter:

p1 = geom1.feature.create('p1','Point');
p1.set('p',0.5);

geom1.run;

to add a point object located at x = 0.5 to the geometry.

To plot the result, enter:

WORKING WITH GEOMETRIES | 53

 mphgeom(model,'geom1','vertexmode','on');

Code for Use with MATLAB ®

model = ModelUtil.create('Model');
model.component.create('comp1',true);
geom1 = model.component('comp1').geom.create('geom1',1);
i1 = geom1.feature.create('i1','Interval');
i1.set('intervals','many');
i1.set('p','0,1,2');
geom1.run;
p1 = geom1.feature.create('p1','Point');
p1.set('p',0.5);
geom1.run;
mphgeom(model,'geom1','vertexmode','on')

CREATING A 2D GEOMETRY USING PRIMITIVE GEOMETRY OBJECTS

Creating Composite Objects

Use a model object with a 2D geometry. Enter:

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1',true);
geom2 = comp1.geom.create('geom2',2);

Continue by creating a rectangle with side length of 2 and centered at the origin:

sq1 = geom2.feature.create('sq1','Square');
sq1.set('size',2);
sq1.set('base','center');

The property size describes the side lengths of the rectangle and the property pos
describes the positioning. The default is to position the rectangle about its lower-left
corner. Use the property base to control the positioning.

54 | CHAPTER 3: BUILDING MODELS

Create a circular hole with a radius of 0.5 centered at (0, 0):

c1 = geom2.feature.create('c1','Circle');
c1.set('r',0.5);
c1.set('pos',[0 0]);

The property r describes the radius of the circle, and the property pos describes the
positioning.

The property pos could have been excluded because the default position
is the origin. The default is to position the circle about its center.

Drill a hole in the rectangle by subtracting the circle from it:

co1 = geom2.feature.create('co1','Compose');
co1.selection('input').set({'c1' 'sq1'});
co1.set('formula','sq1-c1');

A selection object is used to refer to the input object. The operators +, *, and -
correspond to the set operations union, intersection, and difference, respectively.

The Compose operation allows you to work with a formula. Alternatively use the
Difference operation instead of Compose. The following sequence of commands
starts with disabling the Compose operation:

co1.active(false);

dif1 = geom2.feature.create('dif1','Difference');
dif1.selection('input').set({'sq1'});
dif1.selection('input2').set({'c1'});

Run the geometry sequence to create the geometry and plot the result:

geom2.run;

WORKING WITH GEOMETRIES | 55

 mphgeom(model,'geom2');

Trimming Solids
Continue with rounding the corners of the rectangle with the Fillet operation:

fil1 = geom2.feature.create('fil1','Fillet');
fil1.selection('point').set('dif1', [1 2 7 8]);
fil1.set('radius','0.5');

Run the sequence again:

geom2.run;

The geometry sequence is updated with rounded corners. To view the result, enter:

mphgeom(model,'geom2');

Code for use with MATLAB ®

model = ModelUtil.create('Model');

56 | CHAPTER 3: BUILDING MODELS

 comp1 = model.component.create('comp1',true);
geom2 = comp1.geom.create('geom2',2);
sq1 = geom2.feature.create('sq1','Square');
sq1.set('size',2);
sq1.set('base','center');
c1 = geom2.feature.create('c1','Circle');
c1.set('r',0.5);
c1.set('pos',[0 0]);
co1 = geom2.feature.create('co1','Compose');
co1.selection('input').set({'c1' 'sq1'});
co1.set('formula','sq1-c1');
co1.active(false)
dif1 = geom2.feature.create('dif1','Difference');
dif1.selection('input').set({'sq1'});
dif1.selection('input2').set({'c1'});
geom2.run;
mphgeom(model,'geom2');
fil1 = geom2.feature.create('fil1','Fillet');
fil1.selection('point').set('dif1', [1 2 7 8]);
fil1.set('radius','0.5');
geom2.run;
mphgeom(model,'geom2');

CREATING A 2D GEOMETRY USING BOUNDARY MODELING

Use the following commands to create six open curve segments that together form a
closed curve:

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1',true);
g1 = comp1.geom.create('g1',2);
w=1/sqrt(2);
c1 = g1.feature.create('c1','BezierPolygon');
c1.set('type','open');
c1.set('degree',2);
c1.set('p',[-0.5 -1 -1;-0.5 -0.5 0]);
c1.set('w',[1 w 1]);
c2 = g1.feature.create('c2','BezierPolygon');
c2.set('type','open');
c2.set('degree',2);
c2.set('p',[-1 -1 -0.5;0 0.5 0.5]);
c2.set('w',[1 w 1]);
c3 = g1.feature.create('c3','BezierPolygon');
c3.set('type','open');
c3.set('degree',1);
c3.set('p',[-0.5 0.5; 0.5 0.5]);
c4 = g1.feature.create('c4','BezierPolygon');
c4.set('type','open');
c4.set('degree',2);
c4.set('p',[0.5 1 1; 0.5 0.5 0]);

WORKING WITH GEOMETRIES | 57

 c4.set('w',[1 w 1]);
c5 = g1.feature.create('c5','BezierPolygon');
c5.set('type','open');
c5.set('degree',2);
c5.set('p',[1 1 0.5; 0 -0.5 -0.5]);
c5.set('w',[1 w 1]);
c6 = g1.feature.create('c6','BezierPolygon');
c6.set('type','open');
c6.set('degree',1);
c6.set('p',[0.5 -0.5; -0.5 -0.5]);

The objects c1, c2, c3, c4, c5, and c6 are all curve2 objects. The vector [1 w 1]
specifies the weights for a rational Bézier curve that is equivalent to a quarter-circle arc.
The weights can be adjusted to create elliptical or circular arcs.

Convert the curve segments to a solid with the following conversion command:

csol1 = g1.feature.create('csol1','ConvertToSolid');
csol1.selection('input').set({'c1' 'c2' 'c3' 'c4' 'c5' 'c6'});

Then issue a final run command:

g1.run;
mphgeom(model,'g1');

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1',true);
g1 = comp1.geom.create('g1',2);
w=1/sqrt(2);
c1 = g1.feature.create('c1','BezierPolygon');
c1.set('type','open');
c1.set('degree',2);
c1.set('p',[-0.5 -1 -1;-0.5 -0.5 0]);

58 | CHAPTER 3: BUILDING MODELS

 c1.set('w',[1 w 1]);
c2 = g1.feature.create('c2','BezierPolygon');
c2.set('type','open');
c2.set('degree',2);
c2.set('p',[-1 -1 -0.5;0 0.5 0.5]);
c2.set('w',[1 w 1]);
c3 = g1.feature.create('c3','BezierPolygon');
c3.set('type','open');
c3.set('degree',1);
c3.set('p',[-0.5 0.5; 0.5 0.5]);
c4 = g1.feature.create('c4','BezierPolygon');
c4.set('type','open');
c4.set('degree',2);
c4.set('p',[0.5 1 1; 0.5 0.5 0]);
c4.set('w',[1 w 1]);
c5 = g1.feature.create('c5','BezierPolygon');
c5.set('type','open');
c5.set('degree',2);
c5.set('p',[1 1 0.5; 0 -0.5 -0.5]);
c5.set('w',[1 w 1]);
c6 = g1.feature.create('c6','BezierPolygon');
c6.set('type','open');
c6.set('degree',1);
c6.set('p',[0.5 -0.5; -0.5 -0.5]);
csol1 = g1.feature.create('csol1','ConvertToSolid');
csol1.selection('input').set({'c1' 'c2' 'c3' 'c4' 'c5' 'c6'});
g1.run;
mphgeom(model,'g1');

CREATING A 3D GEOMETRY USING SOLID MODELING

This section shows how to create 3D solids using work planes and Boolean operations.

Create a 3D geometry with an xy work plane at z = 0:

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1',true);

geom1 = comp1.geom.create('geom1', 3);

wp1 = geom1.feature.create('wp1', 'WorkPlane');

wp1.set('planetype', 'quick');
wp1.set('quickplane', 'xy');

Add a rectangle to the work plane, then add fillet to its corners:

r1 = wp1.geom.feature.create('r1', 'Rectangle');
r1.set('size',[1 2]);

geom1.run;

WORKING WITH GEOMETRIES | 59

 fil1 = wp1.geom.feature.create('fil1', 'Fillet');
fil1.selection('point').set('r1', [1 2 3 4]);
fil1.set('radius', '0.125');

geom1.runCurrent;

ext1 = geom1.feature.create('ext1', 'Extrude');

ext1.set('distance', '0.1');

Add another yz work plane, at x = 0.5:

wp2 = geom1.feature.create('wp2', 'WorkPlane');

wp2.set('planetype', 'quick');
wp2.set('quickplane', 'yz');
wp2.set('quickx', '0.5');

b1 = wp2.geom.feature.create('b1', 'BezierPolygon');
b1.set('type', 'open');
b1.set('degree', [1 1 1 1]);
b1.set('p',
{'0.75','1','1','0.8','0.75';'0.1','0.1','0.05','0.05','0.1'});
b1.set('w', {'1','1','1','1','1','1','1','1'});

wp2.geom.feature.create('csol1', 'ConvertToSolid');
wp2.geom.feature('csol1').selection('input').set({'b1'});

Revolve the triangle from the yz work plane:

rev1 = geom1.feature.create('rev1', 'Revolve');

rev1.selection('input').set({'wp2'});
rev1.setIndex('pos', '1', 0);

Add the difference operation that computes the final 3D geometry:

dif1 = geom1.feature.create('dif1', 'Difference');

dif1.selection('input').set({'ext1'});
dif1.selection('input2').set({'rev1'});

To run the sequence, enter:

geom1.run;

To view the geometry enter:

60 | CHAPTER 3: BUILDING MODELS

 mphgeom(model);

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1',true);
geom1 = comp1.geom.create('geom1', 3);
wp1 = geom1.feature.create('wp1', 'WorkPlane');
wp1.set('planetype', 'quick');
wp1.set('quickplane', 'xy');
r1 = wp1.geom.feature.create('r1', 'Rectangle');
r1.set('size',[1 2]);
geom1.run
fil1 = wp1.geom.feature.create('fil1', 'Fillet');
fil1.selection('point').set('r1', [1 2 3 4]);
fil1.set('radius', '0.125');
geom1.runCurrent;
ext1 = geom1.feature.create('ext1', 'Extrude');
ext1.set('distance', '0.1');
wp2 = geom1.feature.create('wp2', 'WorkPlane');
wp2.set('planetype', 'quick');
wp2.set('quickplane', 'yz');
wp2.set('quickx', '0.5');
b1 = wp2.geom.feature.create('b1', 'BezierPolygon');
b1.set('type', 'open');
b1.set('degree', [1 1 1 1]);
b1.set('p',
{'0.75','1','1','0.8','0.75';'0.1','0.1','0.05','0.05','0.1'});
b1.set('w', {'1','1','1','1','1','1','1','1'});
wp2.geom.feature.create('csol1', 'ConvertToSolid');
wp2.geom.feature('csol1').selection('input').set({'b1'});
rev1 = geom1.feature.create('rev1', 'Revolve');
rev1.selection('input').set({'wp2'});
rev1.setIndex('pos', '1', 0);
dif1 = geom1.feature.create('dif1', 'Difference');

WORKING WITH GEOMETRIES | 61

 dif1.selection('input').set({'ext1'});
dif1.selection('input2').set({'rev1'});
geom1.run;
mphgeom(model);

Retrieving Geometry Information

To retrieve the detailed information about the geometry in a model, see

Geometry Object Information Methods in the COMSOL Multiphysics
Programming Reference Manual.

First create a simple 3D geometry:

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1',true);

geom1 = comp1.geom.create('geom1', 3);

geom1.feature.create('blk1','Block');
geom1.feature.create('con1','Cone');
geom1.run;

To visualize the geometry in a MATLAB figure window enter:

mphgeom(model)

The model object contains general geometry information methods. For example, to
determine the space dimension of the geometry, enter:

geom1.getSDim

62 | CHAPTER 3: BUILDING MODELS

There are also methods to determine the number of geometrical entities. For example,
to inquire about the number of domains and the number of boundaries:

geom1.getNDomains
geom1.getNBoundaries

Another group of geometry information methods concern adjacency properties of the

geometric entities. For example, the number of up and down domain information on
each boundary:

geom1.getUpDown

There are also methods for evaluating properties such as coordinate values and
curvatures on faces and edges. To get the range of the surface parameters of a given
entity, use the method faceParamRange(N), where N is the face number. For example:

geom1.faceParamRange(1)

returns a 4x1 array following the given format: [s1min; s1max; s2min; s2max] where
s1min and s1max are the minimum, and maximum respectively, of the first surface
parameter. s2min and s2max are the minimum, and maximum respectively, of the
second surface parameter.

To evaluate coordinates, in the global coordinate system, on face 1 for the face
parameters (2, 50), enter:

geom1.faceX(1,[2,50])

To get the parameter range of an edge, use the edgeParamRange(N) method. For
example, to get the length of edge number 3, enter:

edgeParamRange = geom1.edgeParamRange(3)

To get the coordinates and the curvature data at the middle of edge number 3 enter:

geom1.edgeX(3,edgeParam(2)/2)
geom1.edgeCurvature(2,edgeParam(2)/2)

There are also methods for getting information about the internal representation of
the geometry. For example, the coordinates of the geometry vertices:

geom1.getVertexCoord

To fetch geometry information from elements in the geometry sequence, enter:

geom1.obj('blk1').getNBoundaries

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1',true);

WORKING WITH GEOMETRIES | 63

 geom1 = comp1.geom.create('geom1', 3);
geom1.feature.create('blk1','Block');
geom1.feature.create('con1','Cone');
geom1.run;
mphgeom(model)
SDim = geom1.getSDim
NDomains = geom1.getNDomains
NBoundaries = geom1.getNBoundaries
UpDown = geom1.getUpDown
faceParamRange = geom1.faceParamRange(1)
faceX = geom1.faceX(1,[2,50])
edgeParamRange = geom1.edgeParamRange(3)
edgeX = geom1.edgeX(3,edgeParamRange(2)/2)
edgeCurvature = geom1.edgeCurvature(2,edgeParamRange(2)/2)
VertexCoord = geom1.getVertexCoord
objNBoundaries = geom1.obj('blk1').getNBoundaries

Modeling with a Parameterized Geometry

COMSOL Multiphysics has built-in support for parameterized geometries. Parameters
can be used in most geometry operations. To exemplify parameterizing a geometry,
the following script studies the movement of a circular source through two adjacent
rectangular domains:

model = ModelUtil.create('Model');
model.param.set('a',0.2);
comp1 = model.component.create('comp1',true);

geom1 = comp1.geom.create('geom1',2);

r1 = geom1.feature.create('r1','Rectangle');
r1.set('size',[0.5 1]);
r1.set('pos',[0 0]);

r2 = geom1.feature.create('r2','Rectangle');
r2.set('size',[0.6 1]);
r2.set('pos',[0.5 0]);

c1 = geom1.feature.create('c1','Circle');
c1.set('r',0.1);
c1.set('pos',{'a','0.5'});

64 | CHAPTER 3: BUILDING MODELS

 mphgeom(model);

Change the position of the circle by changing the value of parameter a:

model.param.set('a',0.5);

mphgeom(model);

Create a loop that changes the position of the circle in increments:

for a=0.2:0.1:0.5
model.param.set('a',a);
geom1.run;
end

Create a mesh:

comp1.mesh.create('mesh1');

WORKING WITH GEOMETRIES | 65

 Add a Weak Form PDE interface:

w = comp1.physics.create('w', 'WeakFormPDE', 'geom1');

w.feature('wfeq1').set('weak', 1, '-test(ux)*ux-test(uy)*uy');

dir1 = w.feature.create('dir1', 'DirichletBoundary', 1);

dir1.selection.set([1 2 3 6 7]);

src1 = w.feature.create('src1', 'SourceTerm', 2);

src1.set('f', 1, '1');
src1.selection.set([3]);

Then, create a stationary study step:

std1 = model.study.create('std1');

stat1 = std1.feature.create('stat1', 'Stationary');

Create a parametric sweep feature:

p1 = model.batch.create('p1','Parametric');
p1.set('pname', 'a');
p1.set('plist','range(0.2,0.1,0.8)');
p1.run;

Alternatively, you can run the parametric sweep using a MATLAB for loop:

for a=0.2:0.1:0.8
model.param.set('a',a);
std1.run;
end

After updating a parameter that affects the geometry, COMSOL detects

this change and automatically updates the geometry and mesh before
starting the solver. The geometry is associative, which means that physics
settings are preserved as the geometry changes.

Code for use with MATLAB ®

model = ModelUtil.create('Model');
model.param.set('a',0.2);
comp1 = model.component.create('comp1',true);
geom1 = comp1.geom.create('geom1',2);
r1 = geom1.feature.create('r1','Rectangle');
r1.set('size',[0.5 1]);
r1.set('pos',[0 0]);
r2 = geom1.feature.create('r2','Rectangle');
r2.set('size',[0.6 1]);
r2.set('pos',[0.5 0]);
c1 = geom1.feature.create('c1','Circle');

66 | CHAPTER 3: BUILDING MODELS

 c1.set('r',0.1);
c1.set('pos',{'a','0.5'});
mphgeom(model);
model.param.set('a',0.5);
mphgeom(model);
for a=0.2:0.1:0.5
model.param.set('a',a);
geom1.run;
end
comp1.mesh.create('mesh1');
w = comp1.physics.create('w', 'WeakFormPDE', 'geom1');
w.feature('wfeq1').set('weak', 1, '-test(ux)*ux-test(uy)*uy');
dir1 = w.feature.create('dir1', 'DirichletBoundary', 1);
dir1.selection.set([1 2 3 6 7]);
src1 = w.feature.create('src1', 'SourceTerm', 2);
src1.set('f', 1, '1');
src1.selection.set([3]);
std1 = model.study.create('std1');
stat1 = std1.feature.create('stat1', 'Stationary');
p1 = model.batch.create('p1','Parametric');
p1.set('pname', 'a');
p1.set('plist','range(0.2,0.1,0.8)');
p1.run;
for a=0.2:0.1:0.8
model.param.set('a',a);
std1.run;
end

Images and Interpolation Data

This section describes how to generate geometry from a set of data points by using
interpolation curves and how to create geometry from image data.

• Creating a Geometry Using Curve Interpolation

• Creating Geometry from Image Data

CREATING A GEOMETRY USING CURVE INTERPOLATION

Use the interpolation spline feature to import a set of data points that describe a 2D
geometry. To create an interpolation spline feature, enter:

model.component(<ctag>).geom(<geomtag>).feature.create(<ftag>,
'InterpolationCurve')

Then specify data points in a table:

ftag.set('table', <data>)

WORKING WITH GEOMETRIES | 67

 where ftag is the curve interpolation node and <data> can either be a 2-by-N cell
array or a 2-by-N array.

Control the type of geometry generated by the operation with the command:

ftag.set('type', type)

where type can either be 'solid' to generate a solid object, 'closed' to generate a
closed curve or 'open' to generate an open curve.

Example: Curve Interpolation

Create a set of data points in MATLAB, then use these to construct a 2D geometry.

1 Create data points that describe a circle, sorted by the angle, and remove some of
the points:
phi = 0:0.2:2*pi;
phi([1 3 6 7 10 20 21 25 28 32]) = [];
p = [cos(phi);sin(phi)];

2 Add some noise to the data points:

randn('state',17)
p = p+0.02*randn(size(p));

3 Create a 2D geometry with a square:

model = ModelUtil.create('Model');

4 Add a square geometry:

comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 2);

sq1 = geom1.feature.create('sq1', 'Square');

sq1.set('base', 'center');
sq1.set('size', '3');

5 Add an interpolation curve feature:

ic1 = geom1.feature.create('ic1', 'InterpolationCurve');

6 Use the variable p for the data points:

ic1.set('table', p');

7 Specify a closed curve:

ic1.set('type', 'closed');

8 Plot the geometry with the mphgeom command:

68 | CHAPTER 3: BUILDING MODELS

 mphgeom(model)

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 2);
sq1 = geom1.feature.create('sq1', 'Square');
sq1.set('base', 'center');
sq1.set('size', '3');
phi = 0:0.2:2*pi;
phi([1 3 6 7 10 20 21 25 28 32]) = [];
p = [cos(phi);sin(phi)];
randn('state',17)
p = p+0.02*randn(size(p));
ic1 = geom1.feature.create('ic1', 'InterpolationCurve');
ic1.set('table', p');
ic1.set('type', 'closed');
mphgeom(model);

CREATING GEOMETRY FROM IMAGE DATA

Use the function mphimage2geom to create geometry from image data. The image data
format can be M-by-N array for a grayscale image or M-by-N-by-3 array for a true
color image. This section also includes an example (see Example: Convert Image Data
to Geometry).

See the MATLAB function imread to convert an image file to image data.

WORKING WITH GEOMETRIES | 69

 If you specify the image data and the level value that represents the geometry contour
you want to extract, the function mphimage2geom returns a model object with the
desired geometry:

model = mphimage2geom(<imagedata>, <level>)

where imagedata is a C array containing the image data, and level is the contour
level value used to generate the geometry contour.

Specify the type of geometry object generated:

model = mphimage2geom(<imagedata>, <level>, 'type', type)

where type is 'solid' and generates a solid object, 'closed' generates a closed
curve object, or 'open' generates an open curve geometry object.

Use the property curvetype to specify the type of curve used to generate the
geometry object:

model = mphimage2geom(<imagedata>, <level>, 'curvetype', curvetype)

where curvetype can be set to 'polygon' to use a polygon curve. The default curve
type creates a geometry with the best suited geometrical primitives. For interior curves
it uses interpolation curves, while for curves that are touching the perimeter of the
image a polygon curve is used.

To scale the geometry use the scale property where scale is a double value:

model = mphimage2geom(<imagedata>, <level>, 'scale', scale)

Set the minimum distance (in pixels) between coordinates in curve with the mindist
property where mindist is a double value:

model = mphimage2geom(<imagedata>, <level>, 'mindist', mindist)

Set the minimum area (in square pixels) for interior curves where minarea is a double
value:

model = mphimage2geom(<imagedata>, <level>, 'minarea', minarea)

In case of overlapping solids, the function mphimage2geom automatically creates a

Compose node in the model object. If you do not want this geometry feature, set the
property compose to off:

model = mphimage2geom(<imagedata>, <level>, 'compose', 'off')

To create a rectangle domain surrounding the object generated use the property
rectangle:

model = mphimage2geom(<imagedata>, <level>, 'rectangle', 'on')

70 | CHAPTER 3: BUILDING MODELS

mphimage2geom returns a model object with the created geometry stored in a
geometry node. The default geometry node has the tag geom1, to specify manually
the geometry tag use the function as below:

model = mphimage2geom(<imagedata>, <level>, 'geom', <geomtag>)

where <geomtag> is a string corresponding to the tag of the geometry node.

It is also possible to create a geometry object and include it in an existing model object,
to proceed use the command below:

mphimage2geom(<imagedata>, <level>, 'geom', <geomnode>)

where <geomnode> is the geometry node object where to include the newly generated
geometry.

To manually specify the tag of the model object created in the COMSOL server use
the command below:

model = mphimage2geom(<imagedata>, <level>, 'modeltag', <Modeltag>)

where <Modeltag> is a string defining the tag of the model object in the COMSOL
server.

Example: Convert Image Data to Geometry

This example shows how to create geometry based on gray scale image data. First
generate the image data in MATLAB and display the contour in a figure. Then, create
a model object including the geometry represented by the contour value 40.

At the MATLAB prompt enter these commands:

p = (peaks+7)*5;
[c,h] = contourf(p);
clabel(c, h);
model = mphimage2geom(p, 40);
figure(2)
mphgeom(model)

WORKING WITH GEOMETRIES | 71

 Use the property type to create closed or open curves. For example, to create a
geometry following contour 40 with closed curves, enter:

model = mphimage2geom(p, 40, 'type', 'closed');

mphgeom(model)

To scale the geometry, use the scale property. Using the current model scale the
geometry with a factor of 0.001 (1e-3):

model = mphimage2geom(p, 40, 'scale', 1e-3);

72 | CHAPTER 3: BUILDING MODELS

 mphgeom(model)

To insert a rectangle in the geometry that has an outer domain surrounding the
created contour, set the property rectangle to on:

model = mphimage2geom(p, 40, 'rectangle', 'on');

mphgeom(model)

Only include the interior curves with an area larger than 100 square pixels:

model = mphimage2geom(p, 40, 'minarea', 100);

WORKING WITH GEOMETRIES | 73

 mphgeom(model)

Insert the geometry in an existing geometry object:

model = mphopen('model_tutorial_llmatlab');
geom1 = model.component('comp1').geom('geom1');
wp1 = geom1.feature.create('wp1', 'WorkPlane');
wp1.set('quickz',1e-2);
mphimage2geom(p, 50, 'scale', 1e-3, 'geom', wp1.geom);
mphgeom(model)

Code for use with MATLAB ®

p = (peaks+7)*5;
[c,h] = contourf(p);
clabel(c, h);
model = mphimage2geom(p, 40);
figure(2)
mphgeom(model)

74 | CHAPTER 3: BUILDING MODELS

 model = mphimage2geom(p, 40, 'type', 'closed');
mphgeom(model)
model = mphimage2geom(p, 40, 'scale', 1e-3);
mphgeom(model)
model = mphimage2geom(p, 40, 'rectangle', 'on');
mphgeom(model)
model = mphimage2geom(p, 40, 'minarea', 100);
mphgeom(model)
model = mphopen('model_tutorial_llmatlab');
geom1 = model.component('comp1').geom('geom1');
wp1 = geom1.feature.create('wp1', 'WorkPlane');
wp1.set('quickz',1e-2);
mphimage2geom(p, 50, 'scale', 1e-3, 'geom', wp1.geom);
mphgeom(model)

Measuring Entities in Geometry

Use the function mphmeasure to measure the geometry entities in the model.

Enter the command:

measure = mphmeasure(model, <geomtag>, entity, …)

to get the measurement of the entity type entity in the geometry <geomtag>. entity
can be one of 'point', 'edge', 'boundary', or 'domain'.

The output measure return the value of a coordinates, a length, an area or volume,
respectively. For point entities, if you select two or more points, the output correspond
to the midpoint coordinates.

To specify the entity selection to measure enter:

measure = mphmeasure(model, <geomtag>, entity, 'selection', sel, …)

where sel is an integer array that contains the selection number of the entities to
measure.

When you select several entities you can get another measurement value with the
command:

[msr1, msr2]= mphmeasure(model, <geomtag>, entity, …)

where msr2 corresponds to a surface area when the entities are volumes and a distance
when the input entities are two points.

WORKING WITH GEOMETRIES | 75

 Working with Meshes
This section describes how to set up and run meshing sequences in a model.

• The Meshing Sequence Syntax

• Displaying the Mesh
• Mesh Creation Functions
• Importing External Meshes and Mesh Objects
• Measuring Mesh Quality
• Getting Mesh Statistics Information
• Getting and Setting Mesh Data

• Meshing in the COMSOL Multiphysics Reference Manual

• Mesh in the COMSOL Multiphysics Programming Reference
Manual

The Meshing Sequence Syntax

Create a meshing sequence by using the syntax:

model.component(<ctag>).mesh.create(<meshtag>, <geomtag>)

where <meshtag> is a string that you use to refer to the meshing sequence. The tag
geomtag specifies the geometry to use for this mesh node.

To add an operation to a sequence, use the syntax:

mesh.feature.create(<ftag>, operation)

where mesh is a link to a mesh node and the string <ftag> is a string that you use to
refer to the operation.

About Mesh Commands in the COMSOL Multiphysics Programming

Reference Manual

To set a property to a value in a operation, enter:

mesh.feature(<ftag>).set(property, <value>)

76 | CHAPTER 3: BUILDING MODELS

To build the mesh sequence, enter:

mesh.run

To run the mesh node up to a specified feature node <ftag>, enter:

mesh.run(ftag)

For more details on available operations and properties in the sequence,

see Mesh in the COMSOL Multiphysics Programming Reference
Manual.

Displaying the Mesh

To display the mesh in a MATLAB figure, use the function mphmesh. Make sure that
the mesh is built before calling this command:

mphmesh(model)

If there are several meshes in a model, specify the mesh to display using the command:

mphmesh(model, <meshtag>)

Adding a view property will add some view settings from the COMSOL model such
as axes labels (units) and grid and supports hiding of mesh entities. Usually it is
sufficient to use the auto value for the view property:

mphmesh(model, <meshtag>, ‘view’, ‘auto’)

If the model only contains one mesh then the <meshtag> may be left empty.

Use the parent property to specify the axes handle where to display the plot:

mphmesh(model, <meshtag>, 'parent', <axes>)

WORKING WITH MESHES | 77

 The following properties are also available to specify the vertex, edge, or face
rendering:

• vertexmode • meshcolor
• edgemode • edgecolor
• facemode • vertexcolor
• vertexlabels • edgelabelscolor
• edgelabels • vertexlabelscolor
• facelabels • facelabelscolor
• facealpha

Mesh Creation Functions

Several mesh features are discussed, with examples in this section:

• Mesh Sizing Properties

• Creating a 2D Mesh with Triangular Elements
• The Free Meshing Method
• Creating a 2D Mesh with Quadrilateral Elements
• Creating Structured Meshes
• Creating a Structured Quadrilateral Mesh
• Building a Mesh Incrementally
• Revolving a Mesh by Sweeping
• Extruding a Mesh by Sweeping
• Combining Unstructured and Structured Meshes
• Creating Boundary Layer Meshes
• Refining Meshes
• Copying Boundary Meshes
• Converting Mesh Elements

MESH SIZING PROPERTIES

The Size attribute provides a number of input properties that can control the mesh
element size, such as the following properties:

• Maximum and minimum element size

78 | CHAPTER 3: BUILDING MODELS

• Element growth rate
• Curvature factor
• Resolution of narrow regions

These properties are available both globally and locally. The following examples are
included: Creating a 2D Mesh with Triangular Elements and Creating a 2D Mesh with
Quadrilateral Elements. Also discussed is The Free Meshing Method.

There are several predefined settings that can be used to set a suitable combination of
values for many properties. To select one of these settings, use the property hauto and
pass an integer from 1 to 9 as its value to describe the mesh resolution:

• Extremely fine (1) • Coarse (6)

• Extra fine (2) • Coarser (7)
• Finer (3) • Extra coarse (8)
• Fine (4) • Extremely coarse (9)
• Normal (5) (the default)

For details about predefined mesh size settings and mesh element size
parameters, see Size in the COMSOL Multiphysics Programming
Reference Manual.

CREATING A 2D MESH WITH TRIANGULAR ELEMENTS

Generate a triangular mesh of a unit square:

model = ModelUtil.create('Model');

comp1 = model.component.create('comp1', true);

geom1 = comp1.geom.create('geom1',2);
geom1.feature.create('r1','Rectangle');

mesh1 = comp1.mesh.create('mesh1');
ftri1 = mesh1.feature.create('ftri1','FreeTri');
mesh1.run;

WORKING WITH MESHES | 79

 mphmesh(model)

Figure 3-1: Default mesh on a unit square.

The default size feature is generated with the property hauto set to 5, that is:

mesh1.feature('size').set('hauto',5);

To override this behavior, set hauto to another integer. Override this by setting
specific size properties, for example, making the mesh finer than the default by
specifying a maximum element size of 0.02:

mesh1.feature('size').set('hmax',0.02);
mesh1.run;

mphmesh(model)

80 | CHAPTER 3: BUILDING MODELS

This value corresponds to 1/50 of the largest axis-parallel distance, whereas the default
value is 1/15.

Figure 3-2: Fine mesh (maximum element size = 0.02).

Sometimes a nonuniform mesh is desirable. Make a mesh that is denser on the left side
by specifying a smaller maximum element size only on the edge segment to the left
(edge number 1):

mesh1.feature('size').set('hauto',5);

size1 = ftri1.feature.create('size1','Size');
size1.set('hmax',0.02);
size1.selection.geom('geom1',1);
size1.selection.set(1);

mesh1.run

mphmesh(model)

WORKING WITH MESHES | 81

 .

Figure 3-3: Refined mesh on boundary 1(maximum element size = 0.02).

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1',2);
geom1.feature.create('r1','Rectangle');
mesh1 = comp1.mesh.create('mesh1');
ftri1 = mesh1.feature.create('ftri1','FreeTri');
mesh1.run;
mphmesh(model)
mesh1.feature('size').set('hauto',5);
mesh1.feature('size').set('hmax',0.02);
mesh1.run;
mphmesh(model)
mesh1.feature('size').set('hauto',5);
size1 = ftri1.feature.create('size1','Size');
size1.set('hmax',0.02);
size1.selection.geom('geom1',1);
size1.selection.set(1);
mesh1.run
mphmesh(model)

THE FREE MESHING METHOD

The default method to generate free triangle meshes in 2D is based on an advancing
front algorithm. To switch to a Delaunay algorithm use the value del for the method
property. Start by creating a geometry:

model = ModelUtil.create('Model');

comp1 = model.component.create('comp1', true);

82 | CHAPTER 3: BUILDING MODELS

 geom1 = comp1.geom.create('geom1',2);
geom1.feature.create('r1','Rectangle');

c1 = geom1.feature.create('c1','Circle');
c1.set('r','0.5');

co1 = geom1.feature.create('co1','Compose');
co1.selection('input').set({'c1' 'r1'});
co1.set('formula','r1-c1');

geom1.runAll;

mesh1 = comp1.mesh.create('mesh1');

ftri1 = mesh1.feature.create('ftri1','FreeTri');
ftri1.set('method','del');

mesh1.run;

mphmesh(model,'mesh1')

Figure 3-4: Mesh created with the Delaunay method.

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1',2);
geom1.feature.create('r1','Rectangle');
c1 = geom1.feature.create('c1','Circle');
c1.set('r','0.5');
co1 = geom1.feature.create('co1','Compose');
co1.selection('input').set({'c1' 'r1'});
co1.set('formula','r1-c1');

WORKING WITH MESHES | 83

 geom1.runAll;
mesh1 = comp1.mesh.create('mesh1');
ftri1 = mesh1.feature.create('ftri1','FreeTri');
ftri1.set('method','del');
mesh1.run;
mphmesh(model,'mesh1')

CREATING A 2D MESH WITH QUADRILATERAL ELEMENTS

To create an unstructured quadrilateral mesh on a unit circle, enter:

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);

geom1 = comp1.geom.create('geom1',2);

geom1.feature.create('c1','Circle');

mesh1 = comp1.mesh.create('mesh1');

mesh1.feature.create('ftri1','FreeQuad');

mesh1.run;

mphmesh(model)

Figure 3-5: Free quad mesh.

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1',2);
geom1.feature.create('c1','Circle');
mesh1 = comp1.mesh.create('mesh1');

84 | CHAPTER 3: BUILDING MODELS

 mesh1.feature.create('ftri1','FreeQuad');
mesh1.run;
mphmesh(model)

CREATING STRUCTURED MESHES

To create a structured quadrilateral mesh in 2D, use the Map operation. This operation
uses a mapping technique to create the quadrilateral mesh.

Map in the COMSOL Multiphysics Programming Reference Manual

Use the EdgeGroup attribute to group the edges (boundaries) into four edge groups,
one for each edge of the logical mesh. To control the edge element distribution use
the Distribution attribute, which determines the overall mesh density.

CREATING A STRUCTURED QUADRILATERAL MESH

Create a structured quadrilateral mesh on a geometry where the domains are bounded
by more than four edges:

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1',2);
geom1.feature.create('r1','Rectangle');
r2 = geom1.feature.create('r2','Rectangle');
r2.set('pos',[1 0]);
c1 = geom1.feature.create('c1','Circle');
c1.set('r','0.5');
c1.set('pos',[1.1 -0.1]);
dif1 = geom1.feature.create('dif1', 'Difference');
dif1.selection('input').set({'r1' 'r2'});
dif1.selection('input2').set({'c1'});
geom1.run('dif1');

mesh1 = comp1.mesh.create('mesh1');

map1 = mesh1.feature.create('map1','Map');

eg1 = map1.feature.create('eg1', 'EdgeGroup');

eg1.selection.set(1);
eg1.selection('edge1').set([1 3]);
eg1.selection('edge2').set(2);
eg1.selection('edge3').set(8);
eg1.selection('edge4').set(4);

eg2 = map1.feature.create('eg2', 'EdgeGroup');

WORKING WITH MESHES | 85

 eg2.selection.set(2);
eg2.selection('edge1').set(4);
eg2.selection('edge2').set([6 9 10]);
eg2.selection('edge3').set(7);
eg2.selection('edge4').set(5);

mesh1.run;
mphmesh(model);

Figure 3-6: Structured quadrilateral mesh (right) and its underlying geometry.

The left-hand side plot in Figure 3-6 is obtained with this command:

mphgeom(model, 'geom1', 'edgelabels','on')

The EdgeGroup attributes specify that the four edges enclosing domain 1 are
boundaries 1 and 3; boundary 2; boundary 8; and boundary 4. For domain 2 the four
edges are boundary 4; boundary 5; boundary 7; and boundaries 9, 10, and 6.

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1',2);
geom1.feature.create('r1','Rectangle');
r2 = geom1.feature.create('r2','Rectangle');
r2.set('pos',[1 0]);
c1 = geom1.feature.create('c1','Circle');
c1.set('r','0.5');
c1.set('pos',[1.1 -0.1]);
dif1 = geom1.feature.create('dif1', 'Difference');
dif1.selection('input').set({'r1' 'r2'});
dif1.selection('input2').set({'c1'});
geom1.run('dif1');
mesh1 = comp1.mesh.create('mesh1');
map1 = mesh1.feature.create('map1','Map');
eg1 = map1.feature.create('eg1', 'EdgeGroup');

86 | CHAPTER 3: BUILDING MODELS

 eg1.selection.set(1);
eg1.selection('edge1').set([1 3]);
eg1.selection('edge2').set(2);
eg1.selection('edge3').set(8);
eg1.selection('edge4').set(4);
eg2 = map1.feature.create('eg2', 'EdgeGroup');
eg2.selection.set(2);
eg2.selection('edge1').set(4);
eg2.selection('edge2').set([6 9 10]);
eg2.selection('edge3').set(7);
eg2.selection('edge4').set(5);
mesh1.run;
mphmesh(model);
mphgeom(model, 'geom1', 'edgelabels','on')

BUILDING A MESH INCREMENTALLY

To build meshes in a step-by-step fashion, create selections for the parts of the
geometry that you want to mesh in each step, as in this example:

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);

geom1 = comp1.geom.create('geom1',2);
geom1.feature.create('r1','Rectangle');
geom1.feature.create('c1','Circle');
uni1 = geom1.feature.create('uni1', 'Union');
uni1.selection('input').set({'c1' 'r1'});
geom1.runCurrent;
del1 = geom1.feature.create('del1', 'Delete');
del1.selection('input').init(1);
del1.selection('input').set('uni1', 8);
geom1.run('del1');

mesh1 = comp1.mesh.create('mesh1');

dis1 = mesh1.feature.create('dis1', 'Distribution');

dis1.selection.set([2 4]);
dis1.set('type', 'predefined');
dis1.set('method', 'geometric');
dis1.set('elemcount', 20);
dis1.set('reverse', 'on');
dis1.set('elemratio', 20);

dis2 = mesh1.feature.create('dis2', 'Distribution');

dis2.selection.set([1 3]);
dis2.set('type', 'predefined');
dis2.set('method', 'geometric');
dis2.set('elemcount', 20);
dis2.set('elemratio', 20);

WORKING WITH MESHES | 87

 map1 = mesh1.feature.create('map1','Map');
map1.selection.geom('geom1', 2);
map1.selection.set(2);
mesh1.feature.create('frt1','FreeTri');

mesh1.run;

mphmesh(model)

The final mesh is in Figure 3-7. Note the effect of the Distribution feature, with
which the distribution of vertex elements along geometry edges can be controlled.

Figure 3-7: Incrementally generated mesh (right).

The left-hand side plot in Figure 3-7 is obtained with this command:

mphgeom(model, 'geom1', 'edgelabels','on')

To replace the structured quad mesh by an unstructured quad mesh, delete the Map
feature and replace it by a FreeQuad feature:

mesh1.feature.remove('map1');
mesh1.run('dis1');
fq1 = mesh1.feature.create('fq1', 'FreeQuad');
fq1.selection.geom('geom1', 2).set(2);
mesh1.run;

Analogous to working with the meshing sequence in the Model Builder

in the COMSOL Desktop, new features are always inserted after the
current feature.

Thus, to get the FreeQuad feature before the FreeTri feature, the dis1 feature needs
to be made the current feature by building it with the run method. Alternatively, parts

88 | CHAPTER 3: BUILDING MODELS

of a mesh can be selectively removed by using the Delete feature. For example, to
remove the structured mesh from domain 2 (along with the adjacent edge mesh on
edges 3 and 4), and replace it with an unstructured quad mesh, enter these commands:

del1 = mesh1.feature.create('del1','Delete');
del1.selection.geom('geom1', 2).set(2);
del1.set('deladj','on');
frq1 = mesh1.feature.create('frq1','FreeQuad');
frq1.selection.geom('geom1', 2).set(2);
mesh1.run;

For further details on the various commands and their properties see the
COMSOL Multiphysics Programming Reference Manual.

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1',2);
geom1.feature.create('r1','Rectangle');
geom1.feature.create('c1','Circle');
uni1 = geom1.feature.create('uni1', 'Union');
uni1.selection('input').set({'c1' 'r1'});
geom1.runCurrent;
del1 = geom1.feature.create('del1', 'Delete');
del1.selection('input').init(1);
del1.selection('input').set('uni1', 8);
geom1.run('del1');
mesh1 = comp1.mesh.create('mesh1');
dis1 = mesh1.feature.create('dis1', 'Distribution');

WORKING WITH MESHES | 89

 dis1.selection.set([2 4]);
dis1.set('type', 'predefined');
dis1.set('method', 'geometric');
dis1.set('elemcount', 20);
dis1.set('reverse', 'on');
dis1.set('elemratio', 20);
dis2 = mesh1.feature.create('dis2', 'Distribution');
dis2.selection.set([1 3]);
dis2.set('type', 'predefined');
dis2.set('method', 'geometric');
dis2.set('elemcount', 20);
dis2.set('elemratio', 20);
map1 = mesh1.feature.create('map1','Map');
map1.selection.geom('geom1', 2);
map1.selection.set(2);
mesh1.feature.create('frt1','FreeTri');
mesh1.run;
mphmesh(model);
mphgeom(model, 'geom1', 'edgelabels','on')
mesh1.feature.remove('map1');
mesh1.run('dis1');
fq1 = mesh1.feature.create('fq1', 'FreeQuad');
fq1.selection.geom('geom1', 2).set(2);
mesh1.run;
del1 = mesh1.feature.create('del1','Delete');
del1.selection.geom('geom1', 2).set(2);
del1.set('deladj','on');
frq1 = mesh1.feature.create('frq1','FreeQuad');
frq1.selection.geom('geom1', 2).set(2);
mesh1.run;
mphmesh(model);

REVOLVING A MESH BY SWEEPING

Create 3D volume meshes by extruding and revolving face meshes with the Sweep
feature. Depending on the 2D mesh type, the 3D meshes can be hexahedral (brick)
meshes or prism meshes.

Create and visualize a revolved prism mesh as follows:

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);

geom1 = comp1.geom.create('geom1', 3);

wp1 = geom1.feature.create('wp1', 'WorkPlane');
wp1.set('planetype', 'quick');
wp1.set('quickplane', 'xy');
c1 = wp1.geom.feature.create('c1', 'Circle');
c1.set('pos', [2, 0]);
rev1 = geom1.feature.create('rev1', 'Revolve');

90 | CHAPTER 3: BUILDING MODELS

 rev1.set('angle2', '60').set('angle1', -60);
rev1.selection('input').set({'wp1'});
geom1.run('rev1');

mesh1 = comp1.mesh.create('mesh1');
mesh1.feature.create('ftri1', 'FreeTri');
mesh1.feature('ftri1').selection.geom(2);
mesh1.feature('ftri1').selection.set(2);
mesh1.runCurrent;

swe1 = mesh1.feature.create('swe1', 'Sweep');

swe1.selection.geom(3);
swe1.selection.add(1);

mesh1.run;
mphmesh(model)

To obtain a torus, leave the angles property unspecified; the default value gives a
complete revolution.

Figure 3-8: 3D prism mesh created with the Sweep feature.

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 3);
wp1 = geom1.feature.create('wp1', 'WorkPlane');
wp1.set('planetype', 'quick');
wp1.set('quickplane', 'xy');
c1 = wp1.geom.feature.create('c1', 'Circle');
c1.set('pos', [2, 0]);
rev1 = geom1.feature.create('rev1', 'Revolve');
rev1.set('angle2', '60').set('angle1', -60);
rev1.selection('input').set({'wp1'});

WORKING WITH MESHES | 91

 geom1.run('rev1');
mesh1 = comp1.mesh.create('mesh1');
mesh1.feature.create('ftri1', 'FreeTri');
mesh1.feature('ftri1').selection.geom(2);
mesh1.feature('ftri1').selection.set(2);
mesh1.runCurrent;
swe1 = mesh1.feature.create('swe1', 'Sweep');
swe1.selection.geom(3);
swe1.selection.add(1);
mesh1.run;
mphmesh(model)

EXTRUDING A MESH BY SWEEPING

To generate a 3D prism mesh from the same 2D mesh by extrusion and then to plot
it, enter these commands:

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 3);
wp1 = geom1.feature.create('wp1', 'WorkPlane');
wp1.set('planetype', 'quick');
wp1.set('quickplane', 'xy');
c1 = wp1.geom.feature.create('c1', 'Circle');
c1.set('pos', [2, 0]);
ext1 = geom1.feature.create('ext1', 'Extrude');
ext1.selection('input').set({'wp1'});
geom1.runAll;

mesh1 = comp1.mesh.create('mesh1');

ftri1 = mesh1.feature.create('ftri1', 'FreeTri');

ftri1.selection.geom('geom1', 2);
ftri1.selection.set(3);

dis1 = mesh1.feature.create('dis1', 'Distribution');

dis1.selection.set(1);
dis1.set('type', 'predefined');
dis1.set('elemcount', 20);
dis1.set('elemratio', 100);

swe1 = mesh1.feature.create('swe1', 'Sweep');

swe1.selection('sourceface').geom('geom1', 2);
swe1.selection('targetface').geom('geom1', 2);

mesh1.run;
mphmesh(model);

92 | CHAPTER 3: BUILDING MODELS

The result is shown in Figure 3-9. With the properties elemcount and elemratio the
number and distribution of mesh element layers is controlled in the extruded direction.

Distribution in the COMSOL Multiphysics Programming Reference

Manual or at the MATLAB prompt:
mphdoc(model.mesh,'Distribution')

Figure 3-9: Extruded 3D prism mesh.

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 3);
wp1 = geom1.feature.create('wp1', 'WorkPlane');
wp1.set('planetype', 'quick');
wp1.set('quickplane', 'xy');
c1 = wp1.geom.feature.create('c1', 'Circle');
c1.set('pos', [2, 0]);
ext1 = geom1.feature.create('ext1', 'Extrude');
ext1.selection('input').set({'wp1'});
geom1.runAll;
mesh1 = comp1.mesh.create('mesh1');
ftri1 = mesh1.feature.create('ftri1', 'FreeTri');
ftri1.selection.geom('geom1', 2);
ftri1.selection.set(3);
dis1 = mesh1.feature.create('dis1', 'Distribution');
dis1.selection.set(1);
dis1.set('type', 'predefined');
dis1.set('elemcount', 20);

WORKING WITH MESHES | 93

 dis1.set('elemratio', 100);
swe1 = mesh1.feature.create('swe1', 'Sweep');
swe1.selection('sourceface').geom('geom1', 2);
swe1.selection('targetface').geom('geom1', 2);
mesh1.run;
mphmesh(model);

COMBINING UNSTRUCTURED AND STRUCTURED MESHES

By specifying selections for the meshing operations, swept meshing can also be
combined with free meshing. In this case, start by free meshing domain 2, then sweep
the resulting surface mesh through domain 1, as in this example:

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 3);
cone1 = geom1.feature.create('cone1', 'Cone');
cone1.set('r', 0.3).set('h', 1).set('ang', 9);
cone1.set('pos', [ 0 0.5 0.5]).set('axis', [-1 0 0]);
geom1.feature.create('blk1', 'Block');

mesh1 = comp1.mesh.create('mesh1');

ftet1 = mesh1.feature.create('ftet1', 'FreeTet');

ftet1.selection.geom('geom1', 3);
ftet1.selection.set(2);

swe1 = mesh1.feature.create('swe1', 'Sweep');

swe1.selection('sourceface').geom('geom1', 2);
swe1.selection('targetface').geom('geom1', 2);

mesh1.run;
mphmesh(model);

Figure 3-10: Combined structured/unstructured mesh.

94 | CHAPTER 3: BUILDING MODELS

The left-hand side plot in Figure 3-10 is obtained with this command:

mphgeom(model,'geom1','facemode','off','facelabels','on')

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 3);
cone1 = geom1.feature.create('cone1', 'Cone');
cone1.set('r', 0.3).set('h', 1).set('ang', 9);
cone1.set('pos', [ 0 0.5 0.5]).set('axis', [-1 0 0]);
geom1.feature.create('blk1', 'Block');
mesh1 = comp1.mesh.create('mesh1');
ftet1 = mesh1.feature.create('ftet1', 'FreeTet');
ftet1.selection.geom('geom1', 3);
ftet1.selection.set(2);
swe1 = mesh1.feature.create('swe1', 'Sweep');
swe1.selection('sourceface').geom('geom1', 2);
swe1.selection('targetface').geom('geom1', 2);
mesh1.run;
mphmesh(model);
mphgeom(model,'geom1','facemode','off','facelabels','on')

CREATING BOUNDARY LAYER MESHES

For 2D and 3D geometries it is also possible to create boundary layer meshes using the
BndLayer feature. A boundary layer mesh is a mesh with dense element distribution
in the normal direction along specific boundaries. This type of mesh is typically used
for fluid flow problems to resolve the thin boundary layers along the no-slip
boundaries. In 2D, a layered quadrilateral mesh is used along the specified no-slip
boundaries. In 3D, a layered prism mesh or hexahedral mesh is used depending on
whether the corresponding boundary layer boundaries contain a triangular or a
quadrilateral mesh.

If starting with an empty mesh, the boundary-layer mesh uses free meshing to create
the initial mesh before inserting boundary layers into the mesh. This generates a mesh
with triangular and quadrilateral elements in 2D and tetrahedral and prism elements in
3D. The following example illustrates this in 2D:

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);

geom1 = comp1.geom.create('geom1', 2);

r1 = geom1.feature.create('r1', 'Rectangle');
r1.set('size', [10, 5]);
c1 = geom1.feature.create('c1', 'Circle');
c1.set('pos', [3.5 2.5]);
dif1 = geom1.feature.create('dif1', 'Difference');

WORKING WITH MESHES | 95

 dif1.selection('input').set({'r1'});
dif1.selection('input2').set({'c1'});
geom1.runAll;

mesh1 = comp1.mesh.create('mesh1');

bl1 = mesh1.feature.create('bl1', 'BndLayer');

bl1.feature.create('blp1', 'BndLayerProp');
bl1.feature('blp1').selection.set([2 3 5 6 7 8]);

mesh1.run;
mphmesh(model);

Figure 3-11: Boundary layer mesh based on an unstructured triangular mesh.

It is also possible to insert boundary layers in an existing mesh. Use the following
meshing sequence with the geometry sequence from the previous example:

bl1.active(false);

fq1 = mesh1.feature.create('fq1', 'FreeQuad');

fq1.selection.set(1);
mesh1.run;
mphmesh(model)

bl1 = mesh1.feature.create('bl2', 'BndLayer');

bl1.feature.create('blp2', 'BndLayerProp');
bl1.feature('blp2').selection.set([2 3 5 6 7 8]);
mesh1.run;

96 | CHAPTER 3: BUILDING MODELS

 mphmesh(model);

Figure 3-12: Initial unstructured quad mesh (left) and resulting boundary layer mesh
(right).

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 2);
r1 = geom1.feature.create('r1', 'Rectangle');
r1.set('size', [10, 5]);
c1 = geom1.feature.create('c1', 'Circle');
c1.set('pos', [3.5 2.5]);
dif1 = geom1.feature.create('dif1', 'Difference');
dif1.selection('input').set({'r1'});
dif1.selection('input2').set({'c1'});
geom1.runAll;
mesh1 = comp1.mesh.create('mesh1');
bl1 = mesh1.feature.create('bl1', 'BndLayer');
bl1.feature.create('blp1', 'BndLayerProp');
bl1.feature('blp1').selection.set([2 3 5 6 7 8]);
mesh1.run;
mphmesh(model);
bl1.active(false);
fq1 = mesh1.feature.create('fq1', 'FreeQuad');
fq1.selection.set(1);
mesh1.run;
mphmesh(model)
bl1 = mesh1.feature.create('bl2', 'BndLayer');
bl1.feature.create('blp2', 'BndLayerProp');
bl1.feature('blp2').selection.set([2 3 5 6 7 8]);
mesh1.run;
mphmesh(model);

WORKING WITH MESHES | 97

 REFINING MESHES
Given a mesh consisting only of simplex elements (lines, triangles, and tetrahedra) you
can create a finer mesh using the feature Refine. Enter this command to refine the
mesh:

mesh1.feature.create('ref1', 'Refine');

By specifying the property tri, either as a row vector of element numbers or a 2-row
matrix, the elements to be refined can be controlled. In the latter case, the second row
of the matrix specifies the number of refinements for the corresponding element.

The refinement method is controlled by the property rmethod. In 2D, its default value
is regular, corresponding to regular refinement, in which each specified triangular
element is divided into four triangles of the same shape. Setting rmethod to longest
gives longest edge refinement, where the longest edge of a triangle is bisected. Some
triangles outside the specified set might also be refined in order to preserve the
triangulation and its quality.

In 3D, the default refinement method is longest, while regular refinement is only
implemented for uniform refinements. In 1D, the function always uses regular
refinement, where each element is divided into two elements of the same shape.

For stationary or eigenvalue PDE problems you can use adaptive mesh
refinement at the solver stage with the adaption solver step. See
Adaption in the COMSOL Multiphysics Programming Reference
Manual.

COPYING BOUNDARY MESHES

Use the CopyEdge feature in 2D and the CopyFace feature in 3D to copy a mesh
between boundaries.

It is only possible to copy meshes between boundaries that have the same
shape. However, a scaling factor between the boundaries is allowed.

The following example demonstrates how to copy a mesh between two boundaries in
3D and then create a swept mesh on the domain:

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);

geom1 = comp1.geom.create('geom1', 3);

98 | CHAPTER 3: BUILDING MODELS

wp1 = geom1.feature.create('wp1', 'WorkPlane');
wp1.set('planetype', 'quick');
wp1.set('quickplane', 'xy');
c1 = wp1.geom.feature.create('c1', 'Circle');
c1.set('r', 0.5).set('pos', [1, 0]);
rev1 = geom1.feature.create('rev1', 'Revolve');
rev1.set('angle1', 0).set('angle2', 180);
rev1.selection('input').set({'wp1'});
geom1.run('wp1');

mesh1 = comp1.mesh.create('mesh1');

size1 = mesh1.feature.create('size1', 'Size');

size1.selection.geom('geom1', 1);
size1.selection.set(18);
size1.set('hmax', '0.06');

ftri1 = mesh1.feature.create('ftri1', 'FreeTri');

ftri1.selection.geom('geom1', 2);
ftri1.selection.set(10);

cpf1 = mesh1.feature.create('cpf1', 'CopyFace');

cpf1.selection('source').geom('geom1', 2);
cpf1.selection('destination').geom('geom1', 2);
cpf1.selection('source').set(10);
cpf1.selection('destination').set(1);

sw1 = mesh1.feature.create('sw1', 'Sweep');

sw1.selection('sourceface').geom('geom1', 2);
sw1.selection('targetface').geom('geom1', 2);

mesh1.run;
mphmesh(model);

WORKING WITH MESHES | 99

 The algorithm automatically determines how to orient the source mesh on the target
boundary. The result is shown in Figure 3-13.

Figure 3-13: Prism element obtained with the CopyFace and Sweep features.

To explicitly control the orientation of the copied mesh, use the EdgeMap attribute.
The command sequence:

em1 = cpf1.feature.create('em1', 'EdgeMap');

em1.selection('srcedge').set(18);
em1.selection('dstedge').set(2);
mesh1.feature.remove('sw1');
mesh1.feature.create('ftet1', 'FreeTet');

mesh1.run;
mphmesh(model);

copies the mesh between the same boundaries as in the previous example, but now the
orientation of the source mesh on the target boundary is different. The domain is then
meshed by the free mesh, resulting in the mesh in Figure 3-14. In this case it is not
possible to create a swept mesh on the domain because the boundary meshes do not
match in the sweeping direction.

100 | CHAPTER 3: BUILDING MODELS

Figure 3-14: Free tetrahedral mesh after the use of the CopyFace feature.

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 3);
wp1 = geom1.feature.create('wp1', 'WorkPlane');
wp1.set('planetype', 'quick');
wp1.set('quickplane', 'xy');
c1 = wp1.geom.feature.create('c1', 'Circle');
c1.set('r', 0.5).set('pos', [1, 0]);
rev1 = geom1.feature.create('rev1', 'Revolve');
rev1.set('angle1', 0).set('angle2', 180);
rev1.selection('input').set({'wp1'});
geom1.run('wp1');
mesh1 = comp1.mesh.create('mesh1');
size1 = mesh1.feature.create('size1', 'Size');
size1.selection.geom('geom1', 1);
size1.selection.set(18);
size1.set('hmax', '0.06');
ftri1 = mesh1.feature.create('ftri1', 'FreeTri');
ftri1.selection.geom('geom1', 2);
ftri1.selection.set(10);
cpf1 = mesh1.feature.create('cpf1', 'CopyFace');
cpf1.selection('source').geom('geom1', 2);
cpf1.selection('destination').geom('geom1', 2);
cpf1.selection('source').set(10);
cpf1.selection('destination').set(1);
sw1 = mesh1.feature.create('sw1', 'Sweep');
sw1.selection('sourceface').geom('geom1', 2);
sw1.selection('targetface').geom('geom1', 2);

WORKING WITH MESHES | 101

 mesh1.run;
mphmesh(model);
em1 = cpf1.feature.create('em1', 'EdgeMap');
em1.selection('srcedge').set(18);
em1.selection('dstedge').set(2);
mesh1.feature.remove('sw1');
mesh1.feature.create('ftet1', 'FreeTet');
mesh1.run;
mphmesh(model);

CONVERTING MESH ELEMENTS

Use the Convert feature to convert meshes containing quadrilateral, hexahedral, or
prism elements into triangular meshes and tetrahedral meshes. In 2D, the function
splits each quadrilateral element into either two or four triangles. In 3D, it converts
each prism into three tetrahedral elements and each hexahedral element into five, six,
or 28 tetrahedral elements. To control the method used to convert the elements, use
the property splitmethod.

For additional properties supported, see Convert in the COMSOL

Multiphysics Programming Reference Manual or at the MATLAB
prompt: mphdoc(model.mesh,'Convert')

This example demonstrates how to convert a quad mesh into a triangle mesh:

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 2);
geom1.feature.create('c1', 'Circle');
geom1.feature.create('r1', 'Rectangle');
int1 = geom1.feature.create('int1', 'Intersection');
int1.selection('input').set({'c1' 'r1'});
mesh1 = comp1.mesh.create('mesh1', 'geom1');
mesh1.feature.create('fq1', 'FreeQuad');
mesh1.runCurrent;
mphmesh(model);
mesh1.feature.create('conv1', 'Convert');
mesh1.run;
mphmesh(model);

102 | CHAPTER 3: BUILDING MODELS

The result is illustrated in the Figure 3-15:

Figure 3-15: Mesh using free quad elements (left) and converted mesh from quad to
triangle (right).

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 2);
geom1.feature.create('c1', 'Circle');
geom1.feature.create('r1', 'Rectangle');
int1 = geom1.feature.create('int1', 'Intersection');
int1.selection('input').set({'c1' 'r1'});
mesh1 = comp1.mesh.create('mesh1', 'geom1');
mesh1.feature.create('fq1', 'FreeQuad');
mesh1.runCurrent;
mphmesh(model);
mesh1.feature.create('conv1', 'Convert');
mesh1.run;
mphmesh(model);

Importing External Meshes and Mesh Objects

It is possible to import meshes to COMSOL Multiphysics using the following formats:

• COMSOL Multiphysics text files (extension .mphtxt),

• COMSOL Multiphysics binary files (extension .mphbin), and
• NASTRAN® files (extension .nas or .bdf).

IMPORTING MESHES
To import a mesh stored in a supported format use the Import feature. The following
commands import and plot a mesh defined in a COMSOL Multiphysics text file:

WORKING WITH MESHES | 103

 model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
comp1.geom.create('geom1', 3);

mesh1 = comp1.mesh.create('mesh1', 'geom1');

imp1 = mesh1.feature.create('imp1', 'Import');
filenamepath = fullfile(COMSOL, 'applications', ...
'COMSOL_Multiphysics','Meshing_Tutorials');
model.modelPath(filenamepath);
imp1.set('filename','mesh_example_1.mphtxt');
mesh1.feature('imp1').importData;

mesh1.run;
mphmesh(model);

Where COMSOL is the path of root directory where COMSOL Multiphysics is installed.
The above command sequence results in Figure 3-16.

Figure 3-16: Imported mesh.

For additional properties supported, see Import in the COMSOL

Multiphysics Programming Reference Manual.

For a description of the text file format see the COMSOL Multiphysics
Reference Manual.

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
comp1.geom.create('geom1', 3);
mesh1 = comp1.mesh.create('mesh1');
imp1 = mesh1.feature.create('imp1', 'Import');

104 | CHAPTER 3: BUILDING MODELS

 filenamepath = fullfile(COMSOL,'applications',...
'COMSOL_Multiphysics','Meshing_Tutorials');
model.modelPath(filenamepath);
imp1.set('filename','mesh_example_1.mphtxt');
mesh1.feature('imp1').importData;
mesh1.run;
mphmesh(model);

Measuring Mesh Quality

Use the stat method on the meshing sequence to get information on the mesh
quality. The quality measure is a scalar quantity, defined for each mesh element, where
0 represents the lowest quality and 1 represents the highest quality.

The following commands show how to visualize the mesh quality for a mesh on the
unit circle:

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 2);
geom1.feature.create('c1', 'Circle');
geom1.runAll;

mesh1 = comp1.mesh.create('mesh1');
mesh1.feature.create('ftri1', 'FreeTri');
mesh1.run;

meshdset1 = model.result.dataset.create('mesh1', 'Mesh');

meshdset1.set('mesh', 'mesh1');

pg1 = model.result.create('pg1', 2);

meshplot1 = pg1.feature.create('mesh1', 'Mesh');

meshplot1.set('data', 'mesh1');
meshplot1.set('filteractive', 'on');
meshplot1.set('elemfilter', 'quality');
meshplot1.set('tetkeep', 0.25);
mphplot(model,'pg1','rangenum',1);
meshplot1.set('elemfilter','qualityrev');
mphplot(model,'pg1','rangenum',1);

These commands display the worst 25% and the best 25% elements in terms of mesh
element quality. In Figure 3-17, the triangular mesh elements in the right-hand side
plot are more regular than those in the left-hand side plot; this reflects the fact that a

WORKING WITH MESHES | 105

 quality measure of 1 corresponds to a uniform triangle, while 0 means that the triangle
has degenerated into a line.

Figure 3-17: Visualizations of the mesh quality: worst 25% (left) and best 25% (right).

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 2);
geom1.feature.create('c1', 'Circle');
geom1.runAll;
mesh1 = comp1.mesh.create('mesh1', 'geom1');
mesh1.feature.create('ftri1', 'FreeTri');
mesh1.run;
meshdset1 = model.result.dataset.create('mesh1', 'Mesh');
meshdset1.set('mesh', 'mesh1');
pg1 = model.result.create('pg1', 2);
meshplot1 = pg1.feature.create('mesh1', 'Mesh');
meshplot1.set('data', 'mesh1');
meshplot1.set('filteractive', 'on');
meshplot1.set('elemfilter', 'quality');
meshplot1.set('tetkeep', 0.25);
mphplot(model,'pg1','rangenum',1);
meshplot1.set('elemfilter','qualityrev');
mphplot(model,'pg1','rangenum',1);

Getting Mesh Statistics Information

Use the function mphmeshstats to get mesh statistics and mesh information where
stats is a structure containing the mesh statistics information. Enter:

stats = mphmeshstats(model)

106 | CHAPTER 3: BUILDING MODELS

The statistics structure has the following fields:

• meshtag, the tag of the mesh sequence;

• geomtag, the tag of the associated geometry;
• component, the tag of the component the mesh belongs to;
• isactive, boolean variable that indicates if the mesh feature is active (1) or not (0);
• hasproblems, boolean variable that indicates if the mesh feature contains error or
warning nodes (1) or not (0);
• iscomplete, Boolean variable that indicates if the mesh feature is built (1) or
not(0);
• sdim, the space dimension of the mesh feature;
• contributing, the contributing physics and multiphysics feature for
physics-controlled mesh;
• secondorderelements, boolean variable that indicates if the mesh has second
order elements;
• types, the element types present in the mesh. The element type can be vertex (vtx),
edge (edg), triangle (tri), quadrilateral (quad), tetrahedra (tet), pyramid (pyr),
prism (prism), hexahedra (hex). The type can also be of all elements of maximal
dimension in the selection (all);
• numelem, the number of elements for each element type;
• qualitymeasure, the quality measure used to evaluate the quality of the mesh.
• minquality, the minimum element quality;
• meanquality, the mean element quality;
• qualitydistr, the distribution of the element quality (20 values);
• minvolume, the minimum element volume/area;
• maxvolume, the maximum element volume/area;
• volume, the total volume/area of the mesh;
• maxgrowthrate, the maximal growth rate value for the entire selection, regardless
of the element type property;
• meangrowthrate, the average growth rate value for the entire selection, regardless
of the element type property.

If several mesh cases are available in the model object, specify the mesh tag:

stats = mphmeshstats(model, <meshtag>)

WORKING WITH MESHES | 107

 Set the number of bins in the quality distribution histogram (qualitydistr) with the
property qualityhistogram:

stats = mphmeshstats(model,<meshtag>,'qualityhistogram',<num>)

where <num> is an integer corresponding to the desired number of bins.

Set the mesh quality measure from for the mesh statistics with the property
qualitymeasure:

stats = mphmeshstats(model,<meshtag>,'qualitymeasure',<quality>)

where <quality> is one of 'condition' (condition number), 'growth' (growth

rate), 'maxangle' (maximum number), 'skewness' (skewness), 'volcircum'
(volume versus circumradius) or 'vollength' (volume versus length).

Mesh Element Quality and Size in the COMSOL Multiphysics

Programming Reference Manual.

GET MESH STATISTICS ON SPECIFIED ENTITY OR SELECTION

Specify the entity where to evaluate mesh quality with the property entity as in the
command below:

stats = mphmeshstats(model,<meshtag>,'entity',entity)

where entity is one of 'domain', 'boundary', 'edge', or 'point'.

Use the selection property to specify the entity number where to get the mesh
statistics:

stats = mphmeshstats(model,<meshtag>,'entity',entity,...
'selecti on',<selnum>)

where <selnum> is an integer array defining the selection number.

It is also possible to use the selection property to define a selection node defined in the
model:

stats = mphmeshstats(model,<meshtag>,'selection',<seltag>)

where <seltag> is the tag of the desired selection node.

The properties entity and selection cannot be set if the mesh data
information is returned.

108 | CHAPTER 3: BUILDING MODELS

RETRIEVE MESH STATISTICS OF A SPECIFIC ELEMENT TYPE
Restrict the mesh statistics information structure a specific element type with the
property type:

stats = mphmeshstats(model,<meshtag>,'type',type)

where type is one of the mesh element type available: vertex ('vtx'), edge ('edg'),
triangle ('tri'), quad ('quad'), tetrahedron ('tet'), pyramid ('pyr'), prism
('prism') or hexahedron ('hex'). type can also be a cell array to return the mesh
statistics of several mesh element type.

Getting and Setting Mesh Data

The function mphmeshstats also returns the mesh data such as element coordinates.
Use the function with two output variables to get the mesh data. Enter:

[meshstats,meshdata] = mphmeshstats(model)

where meshdata is a MATLAB structure with the following fields:

• types, which contains the element type;

• vertex, which contains the mesh vertex coordinates;
• elem, which contains the element data information;
• elementity, which contains the element entity information for each element type.

Selection and/or Entity properties cannot be set if the mesh data

structure is returned.

EXTRACT AND CREATE MESH INFORMATION

A mesh can be manually created based on a grid generated in MATLAB. However,
before inserting this mesh into the model, a default coarse mesh is generated to get the
mesh information, which enables you to understand the requested mesh structure to
use with the createMesh method. Then a complete mesh can be constructed and
stored in the meshing sequence. If the geometry is not empty, the new mesh is checked
to ensure that it matches the geometry. In other words, to create an arbitrary mesh, an
empty geometry sequence and a corresponding empty meshing sequence need to be
created and the mesh is then constructed on the empty meshing sequence.

Start by creating a 2D model containing a square, and mesh it with triangles:

model = ModelUtil.create('Model');

WORKING WITH MESHES | 109

 comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 2);
geom1.feature.create('sq1', 'Square');
geom1.run;
mesh1 = comp1.mesh.create('mesh1');
mesh1.feature.create('ftri1', 'FreeTri');
mesh1.feature('size').set('hmax', 0.5);
mesh1.run('ftri1');
mphmesh(model);

To get the mesh data information, enter:

[meshstats,meshdata] = mphmeshstats(model);

meshdata =
types: {'edg' 'tri' 'vtx'}
vertex: [2×12 double]
elem: {[2×8 int32] [3×14 int32] [0 5 7 11]}
elementity: {[8×1 int32] [14×1 int32] [4×1 int32]}

The mesh node coordinates are stored in the vertex field:

vtx = meshdata.vertex

vtx =
0 0 0.302 0.5 0.351 0 0.631 1 0.673 0.5 1 1
0 0.5 0.302 0 0.639 1 0.363 0 0.672 1 0.5 1

In the elem field the element information is retrieved, such as the node indices (using
a 0 based) connected to the elements:

tri = meshdata.elem{2}
tri =

110 | CHAPTER 3: BUILDING MODELS

 2 2 2 1 2 2 3 8 9 8 10 8 8 8
1 0 4 4 3 6 7 4 5 9 6 6 10 11
0 3 1 5 6 4 6 6 4 4 7 10 11 9

In the above command, notice that element number 1 is connected to nodes 0, 1, and
2, and element number 2 is connected to nodes 2, 0, and 3.

Then create manually a mesh using a data distribution generated in MATLAB. Enter
the command:

[x,y] = meshgrid([0 0.5 1], [0 0.5 1]);

coord = [x(:) y(:)]';

The node distribution obtained with this command corresponds to the mesh in
Figure 3-18.

Figure 3-18: Mesh with elements (bold) and nodes (italic) indices.

Table 3-1 lists the nodes and element connectivity in the mesh.
TABLE 3-1: ELEMENT AND NODES CONNECTIVITY

ELEMENT NODES

1 0, 3, 4
2 0, 1, 4
3 1, 4, 5
4 1, 2, 5
5 3, 6, 7
6 3, 4, 7
7 4, 7, 8
8 4, 5, 8

WORKING WITH MESHES | 111

 To create the elements and nodes connectivity information use the command:

new_tri(:,1)=[0;3;4];
new_tri(:,2)=[0;1;4];
new_tri(:,3)=[1;4;5];
new_tri(:,4)=[1;2;5];
new_tri(:,5)=[3;6;7];
new_tri(:,6)=[3;4;7];
new_tri(:,7)=[4;7;8];
new_tri(:,8)=[4;5;8];

Assign the element information, node coordinates, and elements connectivity

information, into a new mesh. Use the method createMesh to create the new mesh:

comp2 = model.component.create('comp2', true);

comp2.geom.create('geom2',2);
mesh2 = comp2.mesh.create('mesh2');
mesh2.data.setElem('tri',new_tri)
mesh2.data.setVertex(coord)
mesh2.data.createMesh
mphmesh(model,'mesh2');

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 2);
geom1.feature.create('sq1', 'Square');
geom1.run;
mesh1 = comp1.mesh.create('mesh1');
mesh1.feature.create('ftri1', 'FreeTri');
mesh1.feature('size').set('hmax', 0.5);
mesh1.run('ftri1');
mphmesh(model);
[meshstats,meshdata] = mphmeshstats(model);

112 | CHAPTER 3: BUILDING MODELS

vtx = meshdata.vertex
tri = meshdata.elem{2}
[x,y] = meshgrid([0 0.5 1], [0 0.5 1]);
coord = [x(:) y(:)]';
new_tri(:,1)=[0;3;4];
new_tri(:,2)=[0;1;4];
new_tri(:,3)=[1;4;5];
new_tri(:,4)=[1;2;5];
new_tri(:,5)=[3;6;7];
new_tri(:,6)=[3;4;7];
new_tri(:,7)=[4;7;8];
new_tri(:,8)=[4;5;8];
comp2 = model.component.create('comp2', true);
comp2.geom.create('geom2',2);
mesh2 = comp2.mesh.create('mesh2');
mesh2.data.setElem('tri',new_tri);
mesh2.data.setVertex(coord);
mesh2.data.createMesh;
mphmesh(model,'mesh2');

WORKING WITH MESHES | 113

 Modeling Physics
This section describes how to set up physics interfaces in a model. The physics interface
defines the equations that COMSOL solves.

• The Physics Interface Syntax

• The Material Syntax
• Modifying the Equations
• Adding Global Equations
• Defining Model Settings Using External Data File
• Access the User-Defined Physics Interface

The Physics Interfaces in the COMSOL Multiphysics Reference Manual

The Physics Interface Syntax

Create a physics interface instance using the syntax:

comp = model.component(<ctag>);
phys = comp.physics.create(<phystag>, physint, <geomtag>)

where <phystag> is a string that identifies the physics interface node. Once defined,
you can always refer to a physics interface, or any other feature, by its tag. The string
physint is the constructor name of the physics interface. To get the constructor
name, the best way is to create a model using the desired physics interface in the GUI
and save the model as an M-file. The string <geomtag> refers to the geometry where
you want to specify the interface.

To add a feature to a physics interface, use the syntax:

phys.feature.create(<ftag>, operation)

where <ftag> is a string that you use to refer to the operation. To set a property to a
value in a operation, enter:

phys.feature(<ftag>).set(property, <value>)

114 | CHAPTER 3: BUILDING MODELS

where <ftag> is the string that identifies the feature.

There are alternative syntaxes available. See model.physics() in the

COMSOL Multiphysics Programming Reference Manual or type at the
MATLAB prompt: mphdoc(model.physics).

To disable or remove a feature node, use the methods active or remove, respectively.

The command:

phys.feature(<ftag>).active(false)

disables the feature <ftag>.

To activate the feature node you can set the active method to true:

phys.feature(<ftag>).active(true)

To remove a feature from the model, use the method remove:

phys.feature.remove(<ftag>)

EXAMPLE: IMPLEMENT AND SOLVE A HEAT TRANSFER PROBLEM

This example shows how to add a physics interface and set the boundary conditions in
the model object.

Start to create a model object including a 3D geometry. The geometry consists in a

block with default settings. Enter the following commands at the MATLAB prompt:

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);

geom1 = comp1.geom.create('geom1', 3);

geom1.feature.create('blk1', 'Block');
geom1.run;

Add a Heat Transfer in Solids interface to the model:

phys = comp1.physics.create('ht', 'HeatTransfer', 'geom1');

The tag of the interface is ht. The interface constructor is HeatTransfer. The physics
is defined on geometry geom1.

The physics interface automatically creates a number of default features. To examine

these, enter:

comp1.physics('ht')
ans =

MODELING PHYSICS | 115

 Type: Heat Transfer in Solids
Tag: ht
Identifier: ht
Operation: HeatTransfer
Child nodes: solid1, init1, ins1, idi1, os1, cib1

The physics method has the following child nodes: solid1, init1, ins1, idi1, os1,
and cib1. These are the default features that come with the Heat Transfer in Solids
interface. The first feature, solid1, consists of the heat balance equation. Confirm this
by entering:

solid = phys.feature('solid1')
ans =
Type: Solid
Tag: solid1
Operation: SolidHeatTransferModel

The settings of the solid1 feature node can be modified, for example, to manually set
the material property. To change the thermal conductivity to 400 W/(m*K) enter:

solid.set('k_mat', 1, 'userdef');
solid.set('k', 400);

The Heat Transfer in Solids interface has features you can use to specify domain or
boundary settings. For example, to add a heat source of 105 W/m3 in the study
domain, enter the commands:

hs = phys.feature.create('hs1', 'HeatSource', 3);

hs.selection.set(1);
hs.set('Q', 1, 1e5);

To create a temperature boundary condition on boundaries 3, 5, and 6, enter:

temp = phys.feature.create('temp1', 'TemperatureBoundary', 2);

temp.selection.set([3 5 6]);
temp.set('T0', 1, '300[K]');

Then add a mesh and a study feature and compute the solution:

comp1.mesh.create('mesh1');
std = model.study.create('std1');
std.feature.create('stat', 'Stationary');
std.run

To visualize the solution, create a 3D surface plot group, which is displayed in a

MATLAB figure with the function mphplot:

pg = model.result.create('pg1', 'PlotGroup3D');
pg.feature.create('surf1', 'Surface');
mphplot(model,'pg1','rangenum',1)

116 | CHAPTER 3: BUILDING MODELS

Code for use with MATLAB ®
model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 3);
geom1.feature.create('blk1', 'Block');
geom1.run;
phys = comp1.physics.create('ht', 'HeatTransfer', 'geom1');
comp1.physics('ht')
solid = phys.feature('solid1')
solid.set('k_mat', 1, 'userdef');
solid.set('k', 400);
hs = phys.feature.create('hs1', 'HeatSource', 3);
hs.selection.set(1);
hs.set('Q', 1, 1e5);
temp = phys.feature.create('temp1', 'TemperatureBoundary', 2);
temp.selection.set([3 5 6]);
temp.set('T0', 1, '300[K]');
comp1.mesh.create('mesh1');
std = model.study.create('std1');
std.feature.create('stat', 'Stationary');
std.run
pg = model.result.create('pg1', 'PlotGroup3D');
pg.feature.create('surf1', 'Surface');
mphplot(model,'pg1','rangenum',1)

The Material Syntax

In addition to changing material properties directly inside the physics interfaces,
materials available in the entire model can also be created. Such a material can be used
by all physics interfaces in the model.

MODELING PHYSICS | 117

 Create a material using the syntax:

model.component(<ctag>).material.create(<mattag>)

where <mattag> is a string that you use to refer to a material definition.

A Material is a collection of material models, where each material model defines a set
of material properties, material functions, and model inputs. To add a material model,
use the syntax:

mat.materialmodel.create(<mtag>)

where mat is a link to a material node. The string <mtag> refers to the material model.

To define material properties for the model, set the property value pairs by entering:

mat.materialmodel(<mtag>).set(property, <value>)

COPYING MATERIAL DATA FROM ANOTHER MODEL

Instead of creating the material model from scratch you can copy the material node
from another model object, to proceed use the command:

comp.material.insert(modelMat, <mattag>, '')

where comp is the link to the component, modelMat the model MPH-file name that
contains the source material node and <mattag> the tag of the material node to
import.

EXAMPLE: CREATE A MATERIAL NODE

The section Example: Implement and Solve a Heat Transfer Problem shows how to
change a material property inside a physics interface. This example shows how to
define a material available globally in the model. These steps assume that the previous
example has been followed. Enter:

mat = model.component('comp1').material.create('mat1');

The material automatically creates a material model, def, which can be used to set up
basic properties. For example, use it to define the density and the heat capacity:

mat.materialmodel('def').set('density', 400);
mat.materialmodel('def').set('heatcapacity', 2e3);

To use the defined material in a model, set the solid1 feature to use the material node.
Enter:

118 | CHAPTER 3: BUILDING MODELS

 solid.set('k_mat',1,'from_mat');

model.material() in the COMSOL Multiphysics Programming

Reference Manual or type at the MATLAB prompt:
mphdoc(model.material).

Modifying the Equations

The equation defining the physics node can be edited with the method
featureInfo('info') applied to a feature of the physics node
physics(<phystag>).feature(<ftag>), where <phystag> and <ftag> identify
the physics interface and the feature, respectively:

info = phystag.feature(<ftag>).featureInfo('info');

Use the method getInfoTable(type) to return the tables available in the Equation
View node:

infoTable = info.getInfoTable(type);

where type defines the type of table to return. It can have the value 'Weak' to return
the weak form equations, 'Constraint' to return the constraint types table, or
'Expression' to return the variable expressions table.

EXAMPLE: ACCESS AND MODIFY THE EQUATION WEAK FORM

This example continues from the Example: Implement and Solve a Heat Transfer
Problem and modifies the model equation.

To retrieve information about the physics interface create an info object:

ht = model.component('comp1').physics('ht');
info = ht.feature('solid1').featureInfo('info');

From the info object access the weak form equation by entering:

infoTable = info.getInfoTable('Weak')

This returns a string variable that contains both the name of the weak equation variable
and the equation of the physics implemented in the weak form as in below:

infoTable =

[] 'ht.streamline' 'root.comp1.ht…' '4' 'Spatial' 'Domain 1' '3'

[] '(ht.dfluxx*t…' 'root.comp1.ht…' '4' 'Spatial' 'Domain 1' '3'
[] '-ht.C_eff*(h…' 'root.comp1.ht…' '4' 'Spatial' 'Domain 1' '3'

MODELING PHYSICS | 119

 The output shows that the heat equation is defined at the second row and the weak
expression at the second column. Enter:

infoTable(2,2)

to get the weak equation as a string variable. The result of this command is:

ans =
(ht.dfluxx*test(Tx)+ht.dfluxy*test(Ty)+ht.dfluxz*test(Tz))*ht.d

To access the equation in the node root.comp1.ht.solid1.weak$2; for example, to

modify the equation and lock the expression, run the commands:

equExpr = '200[W/(m*K)]*(-Tx*test(Tx)-Ty*test(Ty)-Tz*test(Tz))';
info.set(infoTable(2,3), {equExpr});

These commands set the heat conductivity to a constant value directly within the heat
balance equation.

You can now compute the solution and display the results in a figure:

model.study('std1').run;;
mphplot(model,'pg1','rangenum',1);

Adding Global Equations

To add a global equation in the model use the command:

model.component(<ctag>).physics.create(<odestag>,
'GlobalEquations')

To define the name of the variable to be solved by the global equation, enter:

120 | CHAPTER 3: BUILDING MODELS

 ode.set('name', <idx>, <name>)

where ode is a link to a Global Equations node and <idx> is the index of the global
equation, and <name> a string with the name of the variable.

Set the expression <expr> of the global equation with:

ode.set('equation', <idx>, <expr>)

where <expr> is defined as a string variable.

Initial value and initial velocity can be set with the commands:

ode.set('initialValueU', <idx>, <init>)

ode.set('initialValueUt', <idx>, <init_t>)

where <init> and <init_t> are the initial value expression for the variable and its
time derivative respectively.

EXAMPLE: SOLVE AN ODE PROBLEM

This example illustrates how to solve the following ODE in a COMSOL model:

·
·· u
u + --- + 1 = 0
2

u0 = 0
·
u 0 = 20

MODELING PHYSICS | 121

 Code for use with MATLAB ®
model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
ge = comp1.physics.create('ge', 'GlobalEquations');
ge1 = ge.feature('ge1');
ge1.set('name', 1, 1, 'u');
ge1.set('equation', 1, 1, 'utt+0.5*ut+1');
ge1.set('initialValueU', 1, 1, 'u0');
ge1.set('initialValueUt', 1, 1, 'u0t');
model.param.set('u0', 0);
model.param.set('u0t', 20);
std1 = model.study.create('std1');
std1.feature.create('time', 'Transient');
std1.feature('time').set('tlist', 'range(0,0.1,20)');
std1.run;
model.result.create('pg1', 1);
model.result('pg1').set('data', 'dset1');
model.result('pg1').feature.create('glob1', 'Global');
model.result('pg1').feature('glob1').set('expr', {'comp1.u'});
mphplot(model,'pg1')

Defining Model Settings Using External Data File

To use tabulated data from files in a model, use the interpolation function available
under the Global Definitions node or the Definitions node of the model.

To add an interpolation function under the Global Definitions node, enter:

model.func.create(<functag>, 'Interpolation')

If you have several model nodes in your model and you want to attach it to the
specified component node <ctag>, enter:

model.component(<ctag>).func.create(<functag>, 'Interpolation')

where <ctag> is the tag of the model node to attach the interpolation function.

Then you can interpolate data specified by a table inside the function (default), or
specified in an external file.

When using an interpolation table, set the interpolation data for each row of the table
with the commands:

func.setIndex('table', <t_value>, <i>, 1)

func.setIndex('table', <ft_value>, <i>, 2)

where func is a link to a function node and <t_value> is the interpolation parameter
value and <ft_value> is the function value. <i> is the index (0-based) in the
interpolation table. If the number of rows is large then it takes a long time to create

122 | CHAPTER 3: BUILDING MODELS

the table element by element. Instead store all the data as a cell array of strings and
set the values all at once:

func.set('table', data)

To use an external file change the source for the interpolation and specify the file,
where filename is the name (including the path) of the data file:

func.set('source', 'file')
func.set('filename', <filename>)

Several interpolation methods are available. Choose the one to use with the command:

func.set('interp', method)

The string method can be set as one of the following alternatives:

• 'neighbor', for interpolation according to the nearest neighbor method,

• 'linear', for linear interpolation method,
• 'cubicspline', for cubic spline interpolation method, or
• 'piecewisecubic', piecewise cubic interpolation method.
You can also decide how to handle parameter values outside the range of the input data
by selecting an extrapolation method:

func.set('extrap', method)

The string method can be one of these values:

• 'const', to use a constant value outside the interpolation data,

• 'linear', for linear extrapolation method,
• 'nearestfunction', to use the nearest function as extrapolation method, or
• 'value', to use a specific value outside the interpolation data.

model.func() in the COMSOL Multiphysics Programming Reference

Manual or type at the MATLAB prompt: mphdoc(model.func).

Access the User-Defined Physics Interface

Using COMSOL with MATLAB, to run a model made with a user-defined physics
interface created with the COMSOL Physics Builder you need to save the compiled
archive (.jar) in your user home folder .comsol/<version>/archives, where you

MODELING PHYSICS | 123

 replace <version> with the current version of COMSOL. Any compressed archive
(with extension .jar) is loaded next time COMSOL with MATLAB starts.

124 | CHAPTER 3: BUILDING MODELS

C r e a t i ng S e le ct i on s
In this section:

• The Selection Node

• Coordinate-Based Selections
• Selection Using Adjacent Geometry
• Displaying Selections

Creating Named Selections

The Selection Node

Use a Selection node to define a collection of geometry entities in a central location in
the model. The selection can easily be accessed in physics or mesh features or during
results analysis. For example, you can refer collectively to a set of boundaries that have
the same boundary conditions, which also have the same mesh size settings.

A selection feature can be one of these types:

• explicit, to include entities explicitly defined by their definitions indices,

• ball, to include entities that fall with a set sphere,
• cylinder, to include entities that fall with a set cylinder, and
• box, to include entities that fall within a set box.

Selection can also be combined by Boolean operations, such as Union, Intersection,

and Difference.

SETTING AN EXPLICIT SELECTION

Create an explicit selection with the command:

model.component(<ctag>).selection.create(<seltag>, 'Explicit')

To specify the domain entity dimension to use in the selection node, enter:

sel.geom(sdim)

CREATING SELECTIONS | 125

 where sel is a link to an Explicit Selection node and sdim is the space dimension that
represents the different geometric entities:

• 3 for domains,
• 2 for boundaries/domains,
• 1 for edges/boundaries, and
• 0 for points.

Set the domain entity indices in the selection node with the command:

sel.set(<idx>)

where <idx> is an array of integers that list the geometric entity indices to add in the
selection.

Coordinate-Based Selections

DEFINING A BALL SELECTION NODE

The Ball selection node is defined by a centerpoint and a radius. The selection can
include geometric entities that are completely or partially inside the ball. The selection
can be set up by using either the COMSOL API directly or the mphselectcoords
function. There are different ways to define the ball selections: Ball Selection Using the
COMSOL API or Ball Selection Using mphselectcoords.

Ball Selection Using the COMSOL API

To add a ball selection to a model object enter:

model.component(<ctag>).selection.create(<seltag>, 'Ball')

To set the coordinates (<x0>, <y0>, <z0>) of the selection centerpoint, enter:

sel.set('posx', <x0>)
sel.set('posy', <y0>)
sel.set('posz', <z0>)

where sel is a link to a Ball Selection node and <x0>, <y0>, <z0> are double values.

Specify the ball radius <r0> with the command:

sel.set('r', <r0>)

where <r0> is a double floating-point value.

To specify the geometric entity level, enter:

sel.set('entitydim', edim)

126 | CHAPTER 3: BUILDING MODELS

where edim is an integer defining the space dimension value (3 for domains, 2 for
boundaries/domains, 1 for edges/boundaries, and 0 for points).

The selection also specifies the condition for geometric entities to be selected:

sel.set('condition', condition)

where condition can be:

• 'inside', to select all geometric entities completely inside the ball,

• 'intersects', to select all geometric entities that intersect the ball (default),
• 'somevertex', to select all geometric entities where at least some vertex is inside
the ball, or
• 'allvertices', to select all geometric entities where all vertices are inside the ball.

Ball Selection Using mphselectcoords

The function mphselectcoords retrieves geometric entities enclosed by a ball.

To get the geometric entities enclosed by a ball of radius r0, with its center positioned
at (x0,y0,z0) enter the command:

idx = mphselectcoords(model, <geomtag>, [<x0>,<y0>,<z0>], ...

entitytype,'radius',<r0>)

where <geomtag> is the tag of geometry where the selection, and entitytype, can be
one of 'point', 'edge', 'boundary', or 'domain'.

The above function returns the entity indices list. Use it to specify a feature selection
or to create an explicit selection as described in Setting an Explicit Selection.

You can also refine the search using several search ball. To do so set the coordinates as
a NxM array where N corresponds of the number of point to use and M the space
dimension of the geometry as in the command below:

idx = mphselectcoords(model, <geomtag>, ...

[<x0>,<y0>,<z0>; <x1>,<y1>,<z1>;...], entitytype)

This returns the geometric entity indices that have vertices near both the given
coordinates using the tolerance radius.

To include any geometric entities in the selection that have at least one vertex inside
the search ball, set the property include to 'any':

idx = mphselectcoords(model, <geomtag>, ...

[<x0>,<y0>,<z0>; <x1>,<y1>,<z1>], entitytype, 'include', 'any');

CREATING SELECTIONS | 127

 In case the model geometry is finalized as an assembly, you have distinct geometric
entities for each part of the assembly (pair). Specify the adjacent domain index to avoid
selection of any overlapping geometric entities. Set the adjnumber property with the
domain index:

idx = mphselectcoords(model, <geomtag>, [<x0>,<y0>,<z0>], ...

entitytype,'radius',<r0>,'adjnumber',<idx>)

where <idx> is the domain index adjacent to the desired geometric entities.

DEFINING A BOX SELECTION NODE

The Box selection node is defined by two diagonally opposite points of a box (in 3D)
or rectangle (in 2D). There are different ways to define the box selections: Box
Selection Using the COMSOL API or Box Selection Using mphselectbox

Box Selection Using the COMSOL API

This command adds a box selection to the model object:

model.component(<ctag>).selection.create(<seltag>, 'Box')

To specify the points (<x0>, <y0>, <z0>) and (<x1>, <y1>, <z1>), enter:

sel.set('xmin', <x0>)
sel.set('ymin', <y0>)
sel.set('zmin', <z0>)
sel.set('xmax', <x1>)
sel.set('ymax', <y1>)
sel.set('zmax', <z1>)

where sel is a link to a Box Selection node and <x0>, <y0>, <z0>, <x1>, <y1>, <z1>
are double values.

To specify the geometric entities levels use the command:

sel.set('entitydim', edim)

where edim is an integer defining the space dimension value (3 for domains, 2 for
boundaries/domains, 1 for edges/boundaries, and 0 for points).

The selection also specifies the condition for geometric entities to be selected:

sel.set('condition', condition)

where condition can be:

• 'inside', to select all geometric entities completely inside the ball,

• 'intersects', to select all geometric entities that intersect the ball (default),

128 | CHAPTER 3: BUILDING MODELS

• 'somevertex', to select all geometric entities where at least some vertex is inside
the ball, or
• 'allvertices', to select all geometric entities where all vertices are inside the ball.

Box Selection Using mphselectbox

The function mphselectbox retrieves geometric entities enclosed by a box (in 3D) or
rectangle (in 2D).

To get the geometric entities of type entitytype enclosed by the box defined by the
points (x0,y0,z0) and (x1,y1,z1), enter the command:

idx = mphselectbox(model,<geomtag>,...
[<x0> <x1>;<y0> <y1>;<z0> <z1>], entitytype)

where <geomtag> is the geometry tag where the selection is applied, and entitytype
can be one of 'point', 'edge', 'boundary', or 'domain'.

The above function returns the entity indices list. Use it to specify a feature selection
or to create an explicit selection as described in Setting an Explicit Selection.

By default the function searches for the geometric entity vertices near these
coordinates using the tolerance radius. It returns only the geometric entities that have
all vertices inside the box or rectangle. To include any geometric entities in the
selection that have at least one vertex inside the search ball, set the property include
to 'any':

idx = mphselectbox(model,<geomtag>,...
[<x0> <x1>;<y0> <y1>;<z0> <z1>], entitytype,'include','any')

In case the model geometry is finalized as an assembly (pair), you have distinct
geometric entities for each part of the assembly. Specify the adjacent domain index to
avoid selection of overlapping geometric entities. Set the adjnumber property with the
domain index:

idx = mphselectbox(model,<geomtag>,...
[<x0> <x1>;<y0> <y1>;<z0> <z1>], entitytype, 'adjnumber', <idx>)

where <idx> is the domain index adjacent to the desired geometric entities.

RETRIEVING POINT COORDINATES USING A SELECTION

Use mphgetcoords to retrieve coordinates of the points that belong to a given
geometry. Run the command below to get the coordinates of the points that belong
to the desired geometric entity:

c = mphgetcoords(model,<geomtag>,entitytype,<idx>)

CREATING SELECTIONS | 129

 where <geomtag> is the geometry tag where the selection is applied, entitytype can
be one of 'point', 'edge', 'boundary', or 'domain' and <idx> is a integer array
containing the geometric entity indices. c is a Nx2 double array containing the point
coordinates where N is the number of points.

Selection Using Adjacent Geometry

Another approach is to select geometric entities and define the adjacent object. For
example, select edges adjacent to a specific domain or boundaries adjacent to a specific
point. There are different ways to create an adjacent selection: Adjacent Selection
Using the COMSOL API or Adjacent Selection Using mphgetadj

Adjacent Selection Using the COMSOL API

This command creates a selection node using adjacent geometric entities:

model.component(<ctag>).selection.create(<seltag>, 'Adjacent')

The geometric entity level needs to be specified with the command:

sel.set(edim)

where sel is a link to an Adjacent Selection node and edim is an integer defining the
space dimension value (3 for domains, 2 for boundaries/domains, 1 for
edges/boundaries, and 0 for points).

The Adjacent selection node only supports the Selection node as an input:

sel.set( 'Adjacent')

and specify the ball radius <r0> with the command:

sel.set('input', <seltag>)

where <seltag> is the tag of an existing Selection node.

Select the level of geometric entities to add in the selection with the command:

sel.set('outputdim', edim)

where edim is an integer defining the space dimension value (3 for domains, 2 for
boundaries/domains, 1 for edges/boundaries, and 0 for points).

If there are multiple domains in the geometry to include in the interior and exterior
selected geometric entities, then enter:

sel.set('interior', 'on')
sel.set('exterior', 'on')

130 | CHAPTER 3: BUILDING MODELS

To exclude the interior/exterior, select geometric entities and set the respective
property to 'off'.

Adjacent Selection Using mphgetadj

An alternative to the COMSOL API is to use the function mphgetadj to select
geometric entities using an adjacent domain.

To get a list of entities of type entitytype adjacent to the entity with the index
<adjnumber> of type adjtype, enter:

idx = mphgetadj(model, <geomtag>, returntype, adjtype, <adjnumber>)

where <geomtag> is the tag of geometry where the selection applies, returntype is
the type of geometry entities whose index are returned and adjtype is the type of
input geometric entity. The string variables returntype and adjtype can be one of
'point', 'edge', 'boundary', or 'domain'.

If <adjnumber> is an array, you can get the list of adjacent entities that connect the
input entities the best, to do so enter:

[idx,idx_cnct] = mphgetadj(model, <geomtag>, ...

returntype, adjtype, <adjnumber>)

The list returned by the function can be used to specify the selection for a model
feature or to create an explicit selection as described in Setting an Explicit Selection.

Displaying Selections
Use the function mphviewselection to display the selected geometric entities in a
MATLAB figure. This section also includes sections to Specify What to Display with
the Selection and Change Display Color and Transparency.

You can either specify the geometric entity index and its entity type or specify the tag
of a selection node available in the model.

To display the entity of type entitytype with the index <idx> enter:

mphviewselection(model, <geomtag>, <idx>, entitytype)

where <geomtag> is the geometry node tag, and <idx> is a positive integer array that
contains the entity indices. The string entitytype can be one of 'point', 'edge',
'boundary', or 'domain'.

If the model contains a selection node with the tag <seltag>, this selection can be
displayed with the command:

CREATING SELECTIONS | 131

 mphviewselection(model, <seltag>)

To plot the selection in an existing axes, set the property 'parent' with the axes
handle. For instance, the command below displays the selection in the current axis:

mphviewselection(model, <seltag>, 'parent', gca)

SPECIFY WHAT TO DISPLAY WITH THE SELECTION

• If the selected selection node is a Ball or Box selection, the ball or box selector is
display by default, to not show the selector, set the property 'showselector' to
'off'.
mphviewselection(model, <seltag>, 'showselector', 'off')

• To deactivate the geometry representation, set the property 'geommode' to 'off'

as in this command:
mphviewselection(model, <seltag>, 'geommode', 'off')

• The property 'vertexmode', 'edgemode' and 'facemode' support the value

'on' or 'off' in order to render the vertex, the edge and the face respectively in
the figure, as in this example line:
mphviewselection(model, <seltag>, 'facemode', 'off')

• To include vertex, edge and face number, set the property 'vertexlabels',
'facelabels' and 'edgelabels' respectively to 'on'.

• Change the marker used to represent the vertex with the property 'facemode'. In
the example command below the vertex are represented in the figure with a '+'
marker instead of the default '.':
mphviewselection(model, <seltag>, 'marker', '+')

• Specify the size of the marker with the property 'edgelabels', you can specify an
integer value corresponding to the number of pixels.

CHANGE DISPLAY COLOR AND TRANSPARENCY

• To change the color of the edge and the face use the property 'edgecolor' and
'facecolor' respectively. Specify the color of the vertex with the property
'markercolor'. Set the property with a character or using a RGB array. In this
example the edges are displayed in blue while the faces are displayed in the color
defined by the RGB array (0.5,0.5,0.5):
mphviewselection(model, <seltag>, 'edgecolor', 'b',...
'facecolor', [0.5 0.5 0.5])

• Specify the color for the selected edge and face with the properties
'edgecolorselected' and 'facecolorselected' respectively. Specify the color

132 | CHAPTER 3: BUILDING MODELS

 of the selected vertex with the property 'markercolorselected'. Use a character
or specify the color by its RGB array. These commands show how to set the edge to
a blue color and the face with the color defined by the RGB array (0.5, 0.5,0.5):
mphviewselection(model, <seltag>, 'edgecolorselected', 'b',...
'facecolorselected', [0.5 0.5 0.5])

• Specify the color for the vertex, edge, and face labels with the properties
'vertexlabelscolor', 'edgelabelscolor' and 'facelabelscolor' respectively.
You can use a character or the RGB array to specify the color.
• Control the transparency of the geometry representation with the property
'facealpha'. Set the property with a double included between 0 and 1. Using this
command the geometry is displayed with a transparency of 50%:
mphviewselection(model, <seltag>, 'facealpha', 0.5)

• Control the transparency of the selector representation with the property

'selectoralpha'. Set the property with a double included between 0 and 1.
Using this command, the selector is displayed with plain color:
mphviewselection(model, <seltag>, 'selectoralpha', 1)

CREATING SELECTIONS | 133

 Computing the Solution
This section describes the commands to use to compute the solution at the MATLAB
prompt. How to set up and run a study node but also how to set manual solver
sequence. This includes the following paragraphs:

• The Study Node

• The Solver Sequence Syntax
• Run the Solver Sequence
• Adding a Parametric Sweep
• Adding a Job Sequence
• Plot While Solving

• Introduction to Solvers and Studies in the COMSOL Multiphysics

Reference Manual
• Solvers and Study Steps in the COMSOL Multiphysics Programming
Reference Manual

The Study Node

A study node holds the nodes that define how to solve a model. These nodes are
divided into these broad categories:

• Study steps, which determines overall settings suitable for a certain study type,
• Solver sequence, and
• Job configurations for distributed parametric jobs, batch jobs, and cluster
computing.

Introduction to Solvers and Studies in the COMSOL Reference Manual

Create a study node by using the syntax:

model.study.create(<studytag>)

where studytag is a string that is used to define the study node.

134 | CHAPTER 3: BUILDING MODELS

The minimal definition for the study node consists in a study step that define the type
of study to use to compute the solution. To add a study step to the study node, use the
syntax:

study.feature.create(<ftag>, operation)

where study is a link to the study node. The string <ftag> is a string that is defined
to refer to the study step. The string operation is one of the basic study types, such
as Stationary, Transient, or Eigenfrequency, and more.

To specify a property value pair for a study step, enter:

study.feature(<ftag>).set(property, <value>)

where <ftag> is the string identifying the study step.

To generate the default solver sequence associated with the physics solved in the model
and compute the solution, run the study node with the command:

study.run

model.study() in the COMSOL Multiphysics Programming Reference

Manual

The Solver Sequence Syntax

If you do not want to use the default solver sequence created by the study node, you
can manually create one. To create a solver sequence, enter:

model.sol.create(<soltag>)

where <soltag> is a string used to refer to the solver sequence associated to a solution
object.

A solver sequence has to be connected to a study node, which is done with the
command:

sol.study(<studytag>)

where <studytag> is the tag of the study you want to associate the solver sequence
sol.

A solver sequence also requires the definition of these nodes:

• Study Step, where the study and study step is specified for compiling the equations
and computing the current solver sequence;

COMPUTING THE SOLUTION | 135

 • Dependent Variables, this node handles settings for the computation of dependent
variables, such as initial value and scaling settings but also the dependent variables
not solved for; and
• Solver node, where the type of solver to use is specified to compute the solution.

Add the nodes to the solver sequence with the command:

sol.feature.create(<ftag>, operation)

where sol is a link to a solver sequence node. The string <ftag> is a string that is
defined to refer to the node, for example, a study step. operation can be
'StudyStep', 'Variables', or 'Stationary'.

To specify a property value pair for a solver feature, enter:

feat.set(property, <value>)

where feat is a link to the solver sequence feature node.

For a list of the operations available for the solver node, see Features
Producing and Manipulating Solutions and Solvers and Study Steps, in
the COMSOL Multiphysics Programming Reference Manual.

Run the Solver Sequence

There are different ways to run the solver sequence:

• run the entire sequence,

• run up to a specified feature, or
• run from a specified feature.

Use the methods run or runAll to run the entire solver configuration node:

model.sol(<soltag>).run
model.sol(<soltag>).runAll

You can also use the method run(<ftag>) to run the solver sequence up to the solver
feature with the tag <ftag>:

model.sol(<soltag>).run(<ftag>)

When you want to continue solving a sequence, use the method runFrom(<ftag>) to
run the solver configuration from the solver feature with the tag <ftag>:

model.sol(<soltag>).runFrom(<ftag>)

136 | CHAPTER 3: BUILDING MODELS

Adding a Parametric Sweep
In addition to the study step that defines a study type, you can add a parametric sweep
to the study node. This is a study step that does not generate equations and can only
be used in combination with other study steps. You can formulate the sequence of
problems that arise when some parameters are varied in the model.

To add a parametric sweep to the study node, enter:

study.feature.create(<ftag>, 'Parametric')

where study is a link to a valid study node where to include the parametric sweep
defined with the tag <ftag>.

To add one or several parameters to the sweep, enter the command:

study.feature(<ftag>).setIndex('pname', <pname>, <idx>)

where <pname> is the name of the parameter to use in the parametric sweep and <idx>
the index number of the parameter. Set the <idx> to 0 to define the first parameter, 1
to define the second parameter, and so on.

Set the list of the parameter values with the command:

study.feature(<ftag>).setIndex('plistarr', <pvalue>, <idx>)

where <pvalue> contains the list of parameter values defined with either a string or
with a double array, and <idx> is the index number of the parameter and uses the same
value as for the parameter name.

If there are several parameters listed in the parametric sweep node, select the type of
sweep by entering:

study.feature(<ftag>).set('sweeptype', type)

where type is a string defining the sweep type, it can take either the value 'filled'
or 'sparse', referring to all combinations or specified combinations of the parameter
values, respectively.

Adding a Job Sequence

In the study node you can define a job sequence such as distributed parametric jobs,
batch jobs, and cluster computing. To create a batch node enter:

model.batch.create(<batchtag>, type)

COMPUTING THE SOLUTION | 137

 where <batchtag> is the tag of the job sequence and type is the type of job to define.
It can be either Parametric, Batch, or Cluster.

For a solver sequence you need to attach the job sequence to an existing study node.
Enter the command:

model.batch(<batchtag>).atach(<studytag>)

where <studytag> is the tag of the study node.

Each job type, such as parametric, batch, or cluster job, can be defined with specific
properties. Use the set method to add a property to the batch job:

model.batch(<batchtag>).set(property, <value>)

You can get the list of the properties in model.batch() in the COMSOL
Multiphysics Programming Reference Manual or type at the MATLAB
prompt: mphdoc(model.batch).

To run the batch sequence use the run method:

model.batch(<batchtag>).run

Plot While Solving

With the Plot While Solving functionality you can monitor the development of the
computation by updating predefined plots during the computation. Since the plots are
displayed on a COMSOL Multiphysics graphics window, start COMSOL with
MATLAB using a graphics COMSOL Multiphysics Server.

See the section Starting COMSOL Multiphysics with MATLAB using

the Graphics Server in the COMSOL Multiphysics Installation Guide.

To activate Plot While Solving, enter the command:

study.feature(<studysteptag>).set('plot', 'on')

where study is a link to a valid study node and <studysteptag> is a string that refers
to the study step.

Specify the plot group to plot by setting the plot group tag:

study.feature(<studysteptag>).set('plotgroup', <ptag>)

138 | CHAPTER 3: BUILDING MODELS

Only one plot group can be plotted during a computation. Use the probe feature
instead if you need to monitor several variables.

To activate Plot While Solving for a probe plot, enter the command:

study.feature(<studysteptag>).set('probesel', seltype)

where seltype is the type of probe selection, that can be 'none', 'all', or
'manual'.

In case the probe selection is set to manual you have to specify the list of the probe
variable to display. Enter the command:

study.feature(<studysteptag>).set('probes', <list>)

where <list> is the a cell array containing the list of the probe to use.

COMPUTING THE SOLUTION | 139

 Analyzing the Results
In this section:

• The Plot Group Syntax

• Displaying The Results
• The Dataset Syntax
• The Numerical Node Syntax
• Exporting Data

• Results Analysis and Plots in the COMSOL Multiphysics Reference

Manual
• Results in the COMSOL Multiphysics Programming Reference
Manual

The Plot Group Syntax

Result plots always appear in plot groups, which are added to the model by the create
method:

model.result.create(<pgtag>, sdim)

Select the string <pgtag> to identify the plot group and the integer sdim to set the
space dimension (1, 2, or 3) of the group.
To add a plot to a plot group, use the syntax:

pg.feature.create(<ftag>, plottype)

where pg is a link to a plot group node and plottype is a string that defines the plot
type.

Plots can have different attributes that modify the display. For example, the
Deformation attribute deforms the plot according to a vector quantity, the Height
Expression attribute introduces 3D height on a 2D table surface plot, and the Filter
attribute filters the plot using a logical expression. The type of plot determines which
attributes are available. Add an attribute to a plot with the command:

pg.feature(<ftag>).feature.create(<attrtag>, attrtype)

140 | CHAPTER 3: BUILDING MODELS

where attrtype is a string that defines the attribute type.

For a list of available plot types and corresponding attribute types, see
Results and model.result() in the COMSOL Multiphysics
Programming Reference Manual or type at the MATLAB prompt:
mphdoc(model.result).

Displaying The Results

There are different commands available to Display Plot Groups, Extract Plot Data,
Plot External Data, and to Add Data Plot to Model. A practical example of this is
included in Example: Plot mpheval Data.

DISPLAY PLOT GROUPS

Use the command mphplot to display a plot group available from the model

mphplot(model)

ANALYZING THE RESULTS | 141

 The figure contains a toolbar that allows the user to control the use of views, lights,
and camera settings set in the model. The COMSOL menu list the plot group
available, as well as geometry and mesh plot.

You can specify the plotgroup to display at the command line. For example, to display
the plot group <pgtag> enter:

mphplot(model, <pgtag>)

mphplot does not support image features, Material Appearance

subfeatures, and clipping views.

This renders the graphics in a MATLAB figure window. In addition you can plot
results in a COMSOL Multiphysics Graphics window if you start COMSOL with
MATLAB using a graphics COMSOL Multiphysics Server. To do this for a plot group
<pgtag> enter:

mphplot(model, <pgtag>, 'server', 'on')

See the section Starting COMSOL Multiphysics with MATLAB using

the Graphics Server in the COMSOL Multiphysics Installation Guide.

Another way to plot in a COMSOL Graphics window is to use the run method:

model.result(<pgtag>).run

Mac OS does not support plotting in a COMSOL Graphics window.

The default settings for plotting in a MATLAB figure do not display the color legend.
To include the color legend in a figure, use the property rangenum:

mphplot(model, <pgtag>, 'rangenum', <idx>)

where the integer <idx> identifies the plot for which the color legend should be
displayed.

142 | CHAPTER 3: BUILDING MODELS

EXTRACT PLOT DATA
In some situation it can be useful to extract data from a plot, for example, if you need
to manually edit the plot as it is allowed in MATLAB. To get a cell array, pd, which
contains the data for each plot feature available in the plot group <pgtag> enter:

pd = mphplot(model, <pgtag>)

The data fields contained in pd returned by mphplot are subject to change. The most
important fields are:

• p, the coordinates for each point that are used for creating lines or triangles.
• n, the normals in each point for the surfaces. These are not always available.
• t, contains the indices to columns in p of a simplex mesh, each column in t
representing a simplex.
• d, the data values for each point.
• rgb, the color values (red, green and blue) entities at each point.
If you don’t want to generate a figure when extracting the plot data structure, set the
property createplot to off as in the command below:

pd = mphplot(model, <pgtag>, 'createplot','off')

This is useful for instance on machine without graphics display support.

EXAMPLE: EXAMINING THE PLOT DATA

Reuse the first on-line example available for mphplot:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
std.feature.create('stat','Stationary');
std.run;
model.result.dataset.create('mir', 'Mirror3D');
pg = model.result.create('pg', 'PlotGroup3D');
pg.set('data', 'mir');
surf1 = pg.feature.create('surf1', 'Surface');
surf1.set('colortable', 'Thermal');
mphplot(model,'pg')
surf2 = pg.feature.create('surf2', 'Surface');
surf2.set('data', 'dset1').set('expr', 'ht.tfluxMag');

Now plot the result and extract the associated plot data structure:

ANALYZING THE RESULTS | 143

 pd = mphplot(model,'pg');

pd is a cell array containing three plot data structure, the first one corresponds the
outline of the geometry, the title, the legend and the color bar information (if any) in
the figure. The second and the third plot data structures correspond to the plot
defined by the features added to the plot group pg: surf1 and surf2 respectively.

To inspect the outline data of the geometry enter:

pd1 = pd{1}{1}

pd1 =
p: [3×1310 single]
t: [2×1048 int32]
rgb: [3×1 double]
type: 'line'
plottype: 'PlotGroup3D'
tag: 'pg'
preserveaspect: 'on'
title: 'Surface: Temperature (K) Surface: Total heat ...'
titlecolor: [3×1 double]
legendcolor: [3×1 double]

To investigate the plot data information of the second surface plot feature (surf2)
enter:

pd3 = pd{3}{1}
pd3 =
p: [3x4917 single]

144 | CHAPTER 3: BUILDING MODELS

 n: [3x4917 single]
t: [3x8964 int32]
d: [4917x1 single]
colortable: 'Rainbow'
cminmax: [0.3193 6.7120e+05]
rgb: [4917x3 single]
type: 'surface'
plottype: 'Surface'
tag: 'surf2'

Code for use with MATLAB ®

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
std.feature.create('stat','Stationary');
std.run;
model.result.dataset.create('mir', 'Mirror3D');
pg = model.result.create('pg', 'PlotGroup3D');
pg.set('data', 'mir');
surf1 = pg.feature.create('surf1', 'Surface');
surf1.set('colortable', 'Thermal');
mphplot(model,'pg')
surf2 = pg.feature.create('surf2', 'Surface');
surf2.set('data', 'dset1').set('expr', 'ht.tfluxMag');
pd = mphplot(model,'pg');

PLOT EXTERNAL DATA

Using the function mphplot you can also plot data that is specified directly as an input
argument. The supported data format is according to the structure provided by the
functions mphplot, mpheval and mphmesh.This allows you to plot data that has first
been extracted from the model. To plot the structure <data>, run the command:

mphplot(<data>)

If the data structure contains the value of several expressions, set the one to display in
the plot with the index property:

mphplot(<data>, 'index', <idx>)

where <idx> is a positive integer that corresponds to the expression to plot.

mphplot supports only plotting of data structures that are of the type
point, line or surface evaluations from mpheval.

Using the colortable option to select from several available color tables when
visualizing data:

ANALYZING THE RESULTS | 145

 mphplot(<data>, 'colortable', colorname)

Obtain a list of alternatives for colorname from the on-line help by entering:

help colortable

To disable the mesh displayed together with the data results, set the property mesh to
off as in this command:

mphplot(<data>, 'mesh', 'off')

ADD DATA PLOT TO MODEL

Use the command mphaddplotdata to create a plot group and incorporate the data
to a COMSOL Model. To create a plot group in the model using the plotdata data
and using the plottype type, enter the command:

mphaddplotdata(model, 'type', type, 'data', data)

where type is the type of plot, that can be 'arrow', 'line', 'surface',
'annotation', 'tube' or 'point'.

You can create a plottype using the plotdata in an existing plotgroup using the property
plotgroup as in the command below:

mphaddplotdata(model, 'type', type, 'data', data,...

'plotgroup', <pgtag> )

If you want to remove all existing plottype in the plotgroup enter:

mphaddplotdata(model, 'type', type, 'data', data,...

'plotgroup', <pgtag> , 'clearplot', 'on')

EXAMPLE: PLOT MPHEVAL DATA

This example extracts COMSOL data at the MATLAB prompt, modifies it and plots
the data in a MATLAB figure.

First load the Busbar model from the COMSOL Multiphysics Applications Libraries.
Enter:

model = mphopen('busbar');

To extract the temperature and the electric potential field, use the command mpheval:

dat = mpheval(model,{'T','V'},'selection',1);

To display the temperature field, using the thermal color table:

mphplot(dat,'index',1,'colortable','thermal');

146 | CHAPTER 3: BUILDING MODELS

Do a simple scaling of the electric potential then plot it using the default color table:

dat.d2 = dat.d2*1e-3;

Plot the newly evaluated data without the mesh:

mphplot(dat, 'index', 2, 'rangenum', 2, 'mesh', 'off');

To emphasize the geometry use the function mphgeom to display line plot on the same
figure:

hold on;
mphgeom(model, 'geom1', 'facemode', 'off')

ANALYZING THE RESULTS | 147

 Code for use with MATLAB ®
model = mphopen('busbar');
dat = mpheval(model,{'T','V'},'selection',1);
mphplot(dat,'index',1,'colortable','thermal');
dat.d2 = dat.d2*1e-3;
mphplot(dat, 'index', 2, 'rangenum', 2, 'mesh', 'off');
hold on;
mphgeom(model, 'geom1', 'facemode', 'off')

The Dataset Syntax

Use datasets to make solutions and meshes available for visualization and data analysis.
You can create Solution datasets, Mesh datasets, or visualization datasets (such as, for
instance, Cut Plane or Edge datasets). While Solution and Mesh datasets are self
defined, visualization datasets always refer to an existing Solution dataset.

See Datasets in the section Commands Grouped by Function of the

COMSOL Multiphysics Programming Reference Manual to get a list of
the available datasets.

All plots refer to datasets; the solutions are always available as the default
dataset.

To create a dataset at the MATLAB prompt, use the command:

model.result.dataset.create(<dsettag>, dsettype);

148 | CHAPTER 3: BUILDING MODELS

where dsettype is one of the available dataset types.

• Datasets in the COMSOL Multiphysics Reference Manual

• Use of Datasets in the COMSOL Multiphysics Programming
Reference Manual

The Numerical Node Syntax

Use the numerical node to perform numerical evaluation from within the COMSOL
Multiphysics model. Numerical operations such as computing averages, integrations,
maxima, or minima of a given expression are available. You can also perform point and
global evaluations.

To create a numerical node, enter:

model.result.numerical.create(<numtag>, numtype)

where numtype is the type of operation to be performed by the node.

For a list of the syntax of the numerical results type available, see About
Results Commands in the COMSOL Multiphysics Programming
Reference Manual.

To store the data needed to create a table and associate the table to the numerical node:

model.result.table.create(<tabletag>,'Table')
model.result.numerical(<numtag>).set('table',<tabletag>)

where <tabletag> is the tag of the table where you want to store the data evaluated
with the numerical operations defined with the tag <numtag>.

To extract the data stored in MATLAB into a table, use the methods getRealRow and
getImagRow, such as:

realRow = model.result.table(<tabletag>).getRealRow(<idx>)
imagRow = model.result.table(<tabletag>).getImagRow(<idx>)

where <idx> is the column index of the table <tabletag>.

ANALYZING THE RESULTS | 149

 For data evaluation in MATLAB you can also use the functions mpheval,
mphevalpoint, mphglobal, mphint2, mphinterp, mphmax, mphmean and mphmin.

Extracting Results

Exporting Data
Use the export node to generate an animation or to export data to an external file
(ASCII format). This section includes information about Animation Export, Data
Export, and the Animation Player.

ANIMATION EXPORT
Animations can be defined as two different types: a movie or an image sequence. The
movie generates file formats such as GIF (.gif), AVI (.avi), or flash (.swf); the
image sequence generates a sequence of images. Make sure COMSOL with MATLAB
using a graphics COMSOL Multiphysics Server to enable plot on server.

To learn how to start COMSOL with MATLAB using a graphics

COMSOL Multiphysics Server, see the COMSOL Multiphysics
Installation Guide.

To generate an animation, add an Animation node to the export method:

anim = model.result.export.create(<animtag>, 'Animation')

To change the animation type use the 'type' property according to:

anim.set('type', type)

where anim is a link to an animation node and type is either 'imageseq' or 'movie'.

To set the filename and finally create the animation, enter:

anim.set(typefilename, <filenname>)
anim.run

In the above, typefilename depends on the type of animation export:

'imagefilename' for an image sequence, 'giffilename' for a gif animation,
'flashfilename' for a flash animation, and 'avifilename' for an avi animation.

For a movie type animation, it is possible to change the number of frames per second
with the command:

150 | CHAPTER 3: BUILDING MODELS

 anim.set('fps', <fps_number>)

where <fps_number> is a positive integer that corresponds to the number of frames

per second to use.

For all animation types you can modify the width and the height of the plot with the
set method:

anim.set('width', <width_px>)
anim.set('height', <height_px>)

where, the positive integers <width_px> and <height_px> are the width and height
size (in pixels), respectively, to use for the animation.

DATA EXPORT
In order to save data to an ASCII file, create a Data node to the export method:

model.result.export.create(<datatag>, 'Data')

Set the expression expr and the file name filenname, and run the export:

model.result.export(<datatag>).setIndex('expr', <expr>, 0)
model.result.export(<datatag>).set('filename', <filenname>)

Set the export data format with the struct property:

model.result.export(<datatag>).set('struct', datastruct)

where datastruct can be set to 'spreadsheet' or 'sectionwise'.

See Data Formats in the COMSOL Multiphysics Programming

Reference Manual for details about the data formats used in the exported
data files.

To export the data in the specified file, run the export node:

model.result.export.(<datatag>).run

ANIMATION PLAYER
For transient and parametric studies, an animation player can be generated to create
interactive animations.

ANALYZING THE RESULTS | 151

 The player displays the figure on a COMSOL Graphics window. Make sure COMSOL
with MATLAB is started using a graphics COMSOL Multiphysics Server.

To learn how to start COMSOL with MATLAB using a graphics

COMSOL Multiphysics Server, see the COMSOL Multiphysics
Installation Guide.

To create a player feature node to the model enter the command:

model.result.export.create(<playtag>, 'Player')

Then associate the player with an existing plot group by setting the plotgroup
property:

model.result.export(<playtag>).set('plotgroup', <pgtag>)

where <pgtag> refers to the plot group, which is animated in the player.

The default frame number used to generate the animation is 25, you can also specify
the number of frame with the command:

model.result.export(<playtag>).set('maxframe', <maxnum>)

where <maxnum> is a positive integer value that corresponds to the maximum number
of frames to generate with the player.

Use the run method to generate the player:

model.result.export(<playtag>).run

Generating Report
You can use the Report Generator tool to report and documentate the model created
in COMSOL Multiphysics. At the command prompt, the function mphreport helps
to automatically add a default report template to the current model and generate it in
the HRTML or Micorsoft Word® format.

To add a default intermediate report templatein the model, enter the command:

mphreport(model, 'action', 'add')

This create a new report node with a unique tag, which 'rpt1' if the model did not
contain any report. You can specify the tag to identify the report node using the
property 'tag' as in the command below:

mphreport(model, 'action', 'add', 'tag', <rpttag>)

152 | CHAPTER 3: BUILDING MODELS

where <rpttag> is a string.

When creating a default report template you can choose between different template
format, enter the command:

mphreport(model, 'action', 'add', 'type', <type>)

where <type> is either 'brief', 'intermediate', or 'complete'.

To generate and visualize the report with the tag <rpttag>, enter:

mphreport(model, 'action', 'run', 'tag', <rpttag>,...

'filename', <fname>)

where <fname> is a string defining the name of the report file.

By default the output format is HTML, to generate a report in the Micorsoft Word
format enter:

mphreport(model, 'action', 'run', 'tag', <rpttag>,...

'filename', <fname>, 'format', 'docx')

ANALYZING THE RESULTS | 153

154 | CHAPTER 3: BUILDING MODELS
 4

Working With Models

This section introduces you to the functionality available for LiveLink™ for
MATLAB® including the wrapper functions and the MATLAB tools that can be
used and combined with a COMSOL Multiphysics® model object.

In this chapter:

• Using Workspace Variables in Model Settings

• Extracting Results
• Running Models in a Loop
• Running Models in Batch Mode
• Working with Matrices
• Extracting Solution Information and Solution Vectors
• Retrieving Xmesh Information
• Navigating the Model
• Handling Errors and Warnings
• Improving Performance for Large Models
• Creating a Custom GUI

155
 Using Workspace Variables in Model
Settings
LiveLink™ for MATLAB® allows you to define the model properties with MATLAB
variables or a MATLAB M-function.

In this section:

• The Set and SetIndex Methods

• Using a MATLAB® Function to Define Model Properties

The Set and SetIndex Methods

You can use MATLAB® variables to set properties of a COMSOL Multiphysics model.
Use the set or setIndex methods to pass the variable value from MATLAB to the
COMSOL model.

THE SET METHODS

Use the set method to assign parameter and/or property values. All assignments
return the parameter object, which means that assignment methods can be appended
to each other.

The basic method for assignment is:

something.set(name, <value>)

The name argument is a string with the name of the parameter/property. The <value>
argument can for example be a MATLAB integer or double array variable. <value>
can also be a string, in this case the value or expression is defined within the model
object.

When using a MATLAB variable, make sure that the value corresponds to the model
unit system. COMSOL can also take care of the unit conversation automatically; in this
case convert the MATLAB integer/double variable to a string variable and use the set
method as:

something.set(property, [num2str(<value>)'[unit]'])

where is the unit you want to set the value property.

156 | CHAPTER 4: WORKING WITH MODELS

THE SETINDEX METHODS
Use the setIndex method to assign values to specific indices (0-based) in an array or
matrix property. All assignment methods return the parameter object, which means
that assignment methods can be appended to each other:

something.setIndex(name, <value>, <index>)

The name argument is a string with the name of the property, <value> is the value to
set the property, which can be a MATLAB variable value or a string, and <index> is
the index in the property table.

When using a MATLAB variable make sure that the value corresponds to the model
unit system. COMSOL can automatically take care of the unit conversation; in this case
converting the MATLAB integer/double variable to a string variable and using the
set method as:

something.setIndex(name, [num2str(<value>)'[unit]'], <index>)

where [unit] is the unit you want to set the value property.

Using a MATLAB® Function to Define Model Properties

Use MATLAB® Function to define the model property. The function can either be
declared within the model object or called at the MATLAB prompt.

CALLING MATLAB FUNCTIONS WITHIN THE COMSOL MODEL OBJECT

LiveLink™ for MATLAB® enables you to declare a MATLAB M-function directly from
within the COMSOL Multiphysics model object. This is typically the case if you want
to call a MATLAB M-function from the COMSOL Desktop. The function being
declared within the model object accepts any parameter, variable, or expression
arguments defined in the COMSOL model object. However, to use a variable defined
at the MATLAB prompt, the variable has to be transferred first in the COMSOL
model as a parameter, for example (see how to set a MATLAB variable in the
COMSOL model in The Set and SetIndex Methods).

The function is evaluated any time the model needs to be updated. The model object
cannot be called as an input argument of the M-function.

Calling External Functions

USING WORKSPACE VARIABLES IN MODEL SETTINGS | 157

 CALLING MATLAB FUNCTIONS AT THE MATLAB PROMPT
Use a MATLAB function to define a model property with the set method:

feature.set(property, myfun(<arg>))

where myfun() is an M-function defined in MATLAB.

The function is called only when the command is executed at the MATLAB prompt.
The argument of the function <arg> called can be MATLAB variables. To include an
expression value from the model object, first extract it at the MATLAB prompt, as
described in Extracting Results.

The function myfun() accepts the model object model as an input argument as any
MATLAB variable.

158 | CHAPTER 4: WORKING WITH MODELS

Extracting Results
Use LiveLink™ for MATLAB® to extract at the MATLAB prompt the data computed
in the COMSOL Multiphysics® model. A suite of wrapper functions is available to
perform evaluation operations at the MATLAB prompt.

In this section:

• Extracting Data at Arbitrary Points

• Evaluating a Minimum of Expression
• Evaluating a Maximum of Expression
• Evaluating an Integral
• Evaluating an Expression Average
• Extracting Data at Node Points
• Evaluating an Expression at Geometry Vertices
• Evaluating Expressions on Particle/Ray Trajectories
• Evaluating a Global Expression
• Evaluating a Matrix Expression at Points
• Evaluating a Global Matrix
• Extracting Data From Tables

Extracting Data at Arbitrary Points

At the MATLAB® prompt, the function mphinterp evaluates the result at arbitrary
points. To evaluate an expression at specific point coordinates, call the function
mphinterp as in the command:

[d1,...] = mphinterp(model,{'e1',...},'coord',<coord>)

where e1,... are the COMSOL Multiphysics expressions to evaluate, <coord> is a

NxM double array, with N the space dimension of the evaluation domain, and M is the
number of evaluation points. The output d1,... is a PxM double array, where P is the
length of the inner solution.

Alternatively, specify the evaluation coordinates using a selection dataset:

data = mphinterp(model, <expr>, 'dataset', <dsettag>)

EXTRACTING RESULTS | 159

 where <dsettag> is a selection dataset tag defined in the model, for example, Cut
point, Cut Plane, Revolve, and so forth. <dsettag> can also be a mesh dataset tag, in
this case the evaluation is performed on the geometric mesh vertices.

The rest of this section has additional information for the function mphinterp:

• Specify the Evaluation Data

• Output Format
• Small-Signal Analysis, Prestressed Analysis, and Harmonic Perturbation Settings
• Specify the Evaluation Quality
• Other Evaluation Properties

SPECIFY THE EVALUATION DATA

The function mphinterp supports the following properties to set the data of the
evaluation to perform:

• dataset, specify the solution dataset to use in the evaluation:

data = mphinterp(model,<expr>,'coord',<coord>,'dataset',<dsettag>)

<dsettag> is the tag of a solution dataset or a mesh dataset. The default value is the
current solution dataset of the model. When a mesh dataset is specified the
expression <expr> can only be geometry or mesh expression.
• selection, specify the domain selection for evaluation:
data =
mphinterp(model,<expr>,'coord',<coord>,'selection',<seltag>)

where <seltag> is the tag of a selection node to use for the data evaluation.
<seltag> can also be a positive integer array that corresponds to the domain index
list. The default selection is All domains where the expression is defined. If the
evaluation point does not belong to the specified domain selection the output value
is NaN.
• edim, specify the element dimension for evaluation:
data = mphinterp(model,<expr>,'coord',<coord>,'edim',edim)

where edim is one of the strings 'point', 'edge', 'boundary' or 'domain'. One
can also use numerical values instead, which in 3D are the values from 0 to 3. The
default settings correspond to the model geometry space dimension. When using a
lower space dimension value, make sure that the evaluation point coordinates
dimension has the same size.

160 | CHAPTER 4: WORKING WITH MODELS

• ext, specify extrapolation control value to define how much outside the mesh the
interpolation searches. This ensures you return data for points that are outside the
geometry:
data = mphinterp(model,<expr>,'coord',<coord>,'ext',<ext>)

where <ext> is a double value corresponding to the search distance as a scale in

terms of the local element size. The default value is 0.1.

• solnum, specify the inner solution number for data evaluation. Inner solutions are
generated for the following analysis types: time domain, frequency domain,
eigenvalue, or stationary with continuation parameters:
data = mphinterp(model,<expr>,'coord',<coord>,solnum',<solnum>)

where <solnum> is an integer array corresponding to the inner solution index.

<solnum> can also be a string:'end' or 'all' to evaluate the expression for the last
inner solution, or all inner solution respectively. By default the evaluation is
performed on all inner solution.
• outersolnum, specify the outer solution number for data evaluation. Outer
solutions are generated with parametric sweeps:
data = mphinterp(model,<expr>,'coord',<coord>,...
'outersolnum',<outersolnum>)

where <outersolnum> is a positive integer corresponding to the outer solution

index. <outersolnum> can also be a string, 'all' or 'end', to evaluate the
expression for all or the last outer solution respectively. The default settings use the
first outer solution for the data evaluation.
• To evaluate the expression data at a specific time use the property t:
data = mphinterp(model,<expr>,'coord',<coord>,'t',<time>)

where <time> is a double array. The default value corresponds to all the stored time
steps.
• phase, specify the phase in degrees:
data = mphinterp(model,<expr>,'coord',<coord>,'phase',<phase>)

where <phase> is a double value.

OUTPUT FORMAT
The function mphinterp returns in the MATLAB workspace a double array. It also
supports other output formats.

To evaluate several expressions at once, make sure that the same number of output
variables are defined as there are expressions specified:

EXTRACTING RESULTS | 161

 [d1,...] = mphinterp(model,{'e1',...},'coord',<coord>)

To extract the unit of the evaluated expression, define an extra output variable:

[data, unit] = mphinterp(model,<expr>,'coord',<coord>)

with unit is a 1xN cell array where N is the number of expressions to evaluate.

Returns only the real part in the data evaluation with the property complexout:

data = mphinterp(model,<expr>,'coord',<coord>,'complexout','off')

To return an error if all evaluation points are outside the geometry, set the property
coorderr to on:

data = mphinterp(model,<expr>,'coord',<coord>,'coorderr','on')

By default the function returns the value NaN for any points that are outside the
domain.

SMALL-SIGNAL ANALYSIS, PRESTRESSED ANALYSIS, AND HARMONIC

PERTURBATION SETTINGS
For solutions with a stored linearization point, such as harmonic perturbation,
small-signal analysis, or prestressed analysis you can specify the evaluation method. Use
the mphinterp function with the property evalmethod:

data = mphinterp(model,<expr>,'coord',<coord>,'evalmethod',method)

where method can be one of the following value:

• 'harmonic', for harmonic perturbation analysis.

• 'linpoint', the expression is evaluated by taking the values of any dependent
variables from the linearization point of the solution.
• 'lintotal', the expression is evaluated by adding the linearization point and the
harmonic perturbation and taking the real part of this sum.
• 'lintotalavg', this is the same as evaluating using the lintotal property and
then averaging over all phases of the harmonic perturbation.
• 'lintotalrms', this is the same as evaluating using the lintotal property and
then taking the RMS over all phases of the harmonic perturbation.
• 'lintotalpeak', this is the same as evaluating using the lintotal property
solution and then taking the maximum over all phases of the harmonic perturbation.

If the property evalmethod is set to harmonic, you specify whether the expression
should be linearized or not with the property differential as shown below:

162 | CHAPTER 4: WORKING WITH MODELS

 data = mphinterp(model,<expr>,'coord',<coord>,...
'evalmethod','harmonic','differential',diffvalue)

The default property value settings ('on') evaluates the differential of the expression
with respect to the perturbation at the linearization point. If diffvalue is off, it
evaluates the expression by taking the values of any dependent variables from the
harmonic perturbation part of the solution.

SPECIFY THE EVALUATION QUALITY

With the property recover, you can specify the accurate derivative recovery:

data = mphinterp(model,<expr>,'coord',<coord>,'recover',recover)

where recover is either 'ppr', 'pprint', or 'off' (the default). Set the property to
ppr to perform recovery inside domains or set to pprint to apply recovery to all
domain boundaries. Because the accurate derivative processing takes time, the
property is disabled by default.

OTHER EVALUATION PROPERTIES

Set the unit property to specify the unit of the evaluation:

data = mphinterp(model,<expr>,'coord',<coord>,'unit',<unit>)

where unit is a cell array with the same size as expr.

To not use complex-value functions with real inputs, use the property complexfun:

data = mphinterp(model,<expr>,'coord',<coord>,'complexfun','off')

The default value uses complex-value functions with real inputs.

Use the property matherr to return an error for undefined operations or expressions:

data = mphinterp(model,<expr>,'coord',<coord>,'matherr','on')

Evaluating a Minimum of Expression

Use the function mphmin to evaluate the minimum of a given expression over an inner
solution list.

To evaluate the minimum of the COMSOL expressions e1,... use the command
mphmin:

[d1,...] = mphmin(model,{'e1',...},edim)

where edim is a string to define the element entity dimension: 'volume', 'surface',
or 'line'. edim can also be a positive integer (3, 2, or 1 respectively). The output

EXTRACTING RESULTS | 163

 variables d1,... are an NxP array where N is the number of inner solutions and P the
number of outer solutions.

The rest of this section has additional information for the function mphmin:

• Specify the Evaluation Data

• Output Format

SPECIFY THE EVALUATION DATA

The function mphmin supports the following properties to set the data of the
evaluation to perform:

• dataset, specify the solution dataset to use in the evaluation:

data = mphmin(model,<expr>,edim,'dataset',<dsettag>)

<dsettag> is the tag of a solution dataset. The default value is the current solution
dataset of the model.
• selection, specify the domain selection for evaluation:
data = mphmin(model,<expr>,edim,'selection',<seltag>)

where <seltag> is the tag of a selection node to use for the data evaluation.
<seltag> can also be a positive integer array that corresponds to the domain index
list. The default selection is all domains where the expression is defined. If the
evaluation point does not belong to the specified domain selection the output value
is NaN.
• solnum, specify the inner solution number for data evaluation. Inner solutions are
generated for the following analysis types: time domain, frequency domain,
eigenvalue, or stationary with continuation parameters:
data = mphmin(model,<expr>,edim,'solnum',<solnum>)

where <solnum> is an integer array corresponding to the inner solution index.

<solnum> can also be a string:'end' or 'all' to evaluate the expression for the last
inner solution or all inner solutions, respectively. By default the evaluation is
performed on all inner solutions.
• outersolnum, specify the outer solution number for data evaluation. Outer
solutions are generated with parametric sweeps:
data = mphmin(model,<expr>,edim,'outersolnum',<outersolnum>)

where <outersolnum> is a positive integer array corresponding to the outer

solution index. <outersolnum> can also be a string, 'all' or 'end', to evaluate

164 | CHAPTER 4: WORKING WITH MODELS

 the expression for all or the last outer solution, respectively. The default setting uses
the first outer solution for the data evaluation.
• To evaluate the expression data at a specific time use the property t:
data = mphmin(model,<expr>,edim,'t',<time>)

where <time> is a double array. The default value corresponds to all the stored time
steps.
• In case of data series, such as from a parametric or a transient study, an operation
can be applied. To perform data series operation use the function mphmin as in the
command below:
data = mphmin(model,<expr>,edim,'dataseries',<dataoperation>)

where <dataoperation> can be one of the following value: 'none' (no operation
performed), 'average' (to compute average of the selected series), 'integral'
(to integrate over series), 'maximum' (to evaluate the maximum over series),
'minimum' (to evaluate the minimum), 'rms' (to compute the root mean square),
'stddev' (to compute the standard deviation), or 'variance' (to compute the
variance).

OUTPUT FORMAT
The function mphmin also supports other output formats.

To extract the unit of the evaluated expression, define an extra output variable:

[data,unit] = mphmin(model,<expr>,edim)

where unit is a 1xN cell array, and N is the number of expressions to evaluate.

By default mphmin returns the results as a squeezed singleton. To get the full singleton
set the squeeze property to off:

data = mphmin(model,<expr>,edim,'squeeze','off')

Set the property matrix to off to return the data as a cell array instead of a double
array:

data = mphmin(model,<expr>,edim,'matrix','off')

Evaluating a Maximum of Expression

Use the function mphmax to evaluate the maximum of a given expression over an inner
solution list.

EXTRACTING RESULTS | 165

 To evaluate the maximum of the COMSOL Multiphysics expressions e1,... use the
command:

[d1,...] = mphmax(model,{'e1',...},edim)

where edim is a string to define the element entity dimension: 'volume', 'surface',
or 'line'. edim can also be a positive integer (3, 2, or 1 respectively). The output
variables d1,... are an NxP array where N is the number of inner solutions and P the
number of outer solutions.

The rest of this section has additional information for the function mphmax:

• Specify the Evaluation Data

• Output Format

SPECIFY THE EVALUATION DATA

The function mphmax supports the following properties to set the data of the
evaluation to perform:

• dataset, specify the solution dataset to use in the evaluation:

data = mphmax(model,<expr>,edim,'dataset',<dsettag>)

<dsettag> is the tag of a solution dataset. The default value is the current solution
dataset of the model.
• selection, specify the domain selection for evaluation:
data = mphmax(model,<expr>,edim,'selection',<seltag>)

where <seltag> is the tag of a selection node to use for the data evaluation.
<seltag> can also be a positive integer array that corresponds to the domain index
list. The default selection is all domains where the expression is defined. If the
evaluation point does not belong to the specified domain selection the output value
is NaN.
• solnum, specify the inner solution number for data evaluation. Inner solutions are
generated for the following analysis types: time domain, frequency domain,
eigenvalue, or stationary with continuation parameters:
data = mphmax(model,<expr>,edim,'solnum',<solnum>)

where <solnum> is an integer array corresponding to the inner solution index.

<solnum> can also be a string:'end' or 'all' to evaluate the expression for the last
inner solution or all inner solutions, respectively. By default the evaluation is
performed on all inner solutions.

166 | CHAPTER 4: WORKING WITH MODELS

• outersolnum, specify the outer solution number for data evaluation. Outer
solutions are generated with parametric sweeps:
data = mphmax(model,<expr>,edim,'outersolnum',<outersolnum>)

where <outersolnum> is a positive integer array corresponding to the outer

solution index. <outersolnum> can also be a string, 'all' or 'end', to evaluate
the expression for all or the last outer solution, respectively. The default setting uses
the first outer solution for the data evaluation.
• To evaluate the expression data at a specific time use the property t:
data = mphmax(model,<expr>,edim,'t',<time>)

where <time> is a double array. The default value corresponds to all the stored time
steps.
• In case of data series, such as from a parametric or a transient study, an operation
can be applied. To perform data series operation use the function mphmax as in the
command below:
data = mphmax(model,<expr>,edim,'dataseries', <dataoperation>)

where <dataoperation> can be one of the following value: 'none' (no operation
performed), 'average' (to compute average of the selected series), 'integral'
(to integrate over series), 'maximum' (to evaluate the maximum over series),
'minimum' (to evaluate the minimum), 'rms' (to compute the root mean square),
'stddev' (to compute the standard deviation), or 'variance' (to compute the
variance).

OUTPUT FORMAT
The function mphmax also supports other output formats.

To extract the unit of the evaluated expression, define an extra output variable:

[data,unit] = mphmax(model,<expr>,edim)

where unit is a 1xN cell array and N is the number of expressions to evaluate.

By default mphmax returns the results as a squeezed singleton. To get the full singleton
set the squeeze property to off:

data = mphmax(model,<expr>,edim,'squeeze','off')

Set the property matrix to off to return the data as a cell array instead of a double
array:

data = mphmax(model,<expr>,edim,'matrix','off')

EXTRACTING RESULTS | 167

 Evaluating an Integral
Evaluate an integral of expression with the function mphint2.

To evaluate the integral of the expression over the domain with the highest space
domain dimension call the function mphint2 as in this command:

[d1,...] = mphint2(model,{'e1',...},edim)

where e1,... are the expressions to integrate. The values d1,... are returned as a
1xP double array, with P the length of inner parameters. edim is the integration
dimension, which can be 'line', 'surface', 'volume', or an integer value that
specifies the space dimension (1, 2, or 3).

The rest of this section has additional information for the function mphint2:

• Specify the Integration Data

• Output Format
• Specify the Integration Settings

SPECIFY THE INTEGRATION DATA

The function mphint2 supports the following properties to set the data of the
evaluation to perform:

• dataset, specify the solution dataset to use in the integration:

data = mphint2(model,<expr>,edim,'dataset',<dsettag>)

<dsettag> is the tag of a solution dataset. The default value is the current solution
dataset of the model.
• selection, specify the integration domain:
data = mphint2(model,<expr>,edim,'selection',<seltag>)

where <seltag> is the tag of a selection node to use for the data evaluation.
<seltag> can also be a positive integer array that corresponds to the domain index
list. The default selection is all domains where the expression is defined. If the
evaluation point does not belong to the specified domain selection the output value
is NaN.
• solnum, specify the inner solution number for data evaluation. Inner solutions are
generated for the following analysis types: time domain, frequency domain,
eigenvalue, or stationary with continuation parameters:

168 | CHAPTER 4: WORKING WITH MODELS

 data = mphint2(model,<expr>,edim,'solnum',<solnum>)

where <solnum> is an integer array corresponding to the inner solution index.

<solnum> can also be a string:'end' or 'all' to evaluate the expression for the last
inner solution, or all inner solutions, respectively. By default the evaluation is
performed on all inner solutions.
• outersolnum, specify the outer solution number for data evaluation. Outer
solutions are generated with parametric sweeps:
data = mphint2(model,<expr>,edim,'outersolnum',<outersolnum>)

where <outersolnum> is a positive integer corresponding to the outer solution

index. <outersolnum> can also be a string, 'all' or 'end', to evaluate the
expression for all or the last outer solution respectively. The default settings use the
first outer solution for the data evaluation.
• To evaluate the expression data at a specific time use the property t:
data = mphint2(model,<expr>,edim,'t',<time>)

where <time> is a double array. The default value corresponds to all the stored time
steps.
• In case of data series, such as from a parametric or a transient study, an operation
can be applied. To perform data series operation use the function mphint2 as in the
command below:
data = mphint2(model,<expr>,edim,'dataseries',<dataoperation>)

where <dataoperation> can be one of the following value: 'none' (no operation
performed), 'average' (to compute average of the selected series), 'integral'
(to integrate over series), 'maximum' (to evaluate the maximum over series),
'minimum' (to evaluate the minimum), 'rms' (to compute the root mean square),
'stddev' (to compute the standard deviation), or 'variance' (to compute the
variance).

OUTPUT FORMAT
The function mphint2 also supports other output formats.

To extract the unit of the evaluated expression, define an extra output variable:

[data,unit] = mphint2(model,<expr>,edim)

with unit is a 1xN cell array where N is the number of expressions to evaluate.

By default mphint2 returns the results as a squeezed singleton. To get the full
singleton, set the squeeze property to off:

EXTRACTING RESULTS | 169

 data = mphint2(model,<expr>,edim,'squeeze','off')

Set the property matrix to off to return the data as a cell array instead of a double
array:

data = mphint2(model,<expr>,edim,'matrix','off')

SPECIFY THE INTEGRATION SETTINGS

To specify integration settings such as the integration method, integration order, or
axisymmetry assumption using these properties:

• method, specify the integration method, which can be either integration or

summation:
data = mphint2(model,<expr>,edim,'method',method)

where method can be 'integration' or 'summation'. The default uses the

appropriate method for the given expression.
• intorder, specify the integration order:
data = mphint2(model,<expr>,edim,'intorder',<order>)

where order is a positive integer. The default value is 4.

For datasets other than Solution, Particle, Cut*, Time Integral, Time
Average, Surface, and Line, the integration order does correspond to an
element refinement.

• intsurface or intvolume, compute surface or volume integral for axisymmetric

models:
data = mphint2(model,<expr>,edim,'intsurface','on')
data = mphint2(model,<expr>,edim,'intvolume','on')

Evaluating an Expression Average

Use the function mphmean to evaluate the average of a given expression over inner
solution lists. To evaluate the mean of the expressions e1,... use the command
mphmean:

[d1,...] = mphmean(model,{'e1',...},edim)

where edim is a string to define the element entity dimension: 'volume', 'surface',
or 'line'. edim can also be a positive integer (3, 2, or 1 respectively). The output
variables d1,... are an NxP array where N is the number of inner solutions and P the
number of outer solutions.

170 | CHAPTER 4: WORKING WITH MODELS

The rest of this section has additional information for the function mphmean:

• Specify the Evaluation Data

• Output Format
• Specify the Integration Settings

SPECIFY THE EVALUATION DATA

The function mphmean supports the following properties to set the data of the
evaluation to perform:

• dataset, specify the solution dataset to use in the evaluation:

data = mphmean(model,<expr>,edim,'dataset',<dsettag>)

<dsettag> is the tag of a solution dataset. The default value is the current solution
dataset of the model.
• selection, specify the domain selection for evaluation:
data = mphmean(model,<expr>,edim,'selection',<seltag>)

where <seltag> is the tag of a selection node to use for the data evaluation.
<seltag> can also be a positive integer array that corresponds to the domain index
list. The default selection is all domains where the expression is defined. If the
evaluation point does not belong to the specified domain selection the output value
is NaN.
• solnum, specify the inner solution number for data evaluation. Inner solutions are
generated for the following analysis types: time domain, frequency domain,
eigenvalue, or stationary with continuation parameters:
data = mphmean(model,<expr>,edim,'solnum',<solnum>)

where <solnum> is an integer array corresponding to the inner solution index.

<solnum> can also be a string:'end' or 'all' to evaluate the expression for the last
inner solution or all inner solutions, respectively. By default the evaluation is
performed on all inner solutions.
• outersolnum, specify the outer solution number for data evaluation. Outer
solutions are generated with parametric sweeps:
data = mphmean(model,<expr>,edim,'outersolnum',<outersolnum>)

where <outersolnum> is a positive integer array corresponding to the outer

solution index. <outersolnum> can also be a string, 'all' or 'end', to evaluate
the expression for all or the last outer solution, respectively. The default setting uses
the first outer solution for the data evaluation.

EXTRACTING RESULTS | 171

 • To evaluate the expression data at a specific time use the property t:
data = mphmean(model,<expr>,edim,'t',<time>)

where <time> is a double array. The default value corresponds to all the stored time
steps.
• In case of data series, such as from a parametric or a transient study, an operation
can be applied. To perform data series operation use the function mphmean as in the
command below:
data = mphmean(model,<expr>,edim,'dataseries',<dataoperation>)

where <dataoperation> can be one of the following value: 'none' (no operation
performed), 'average' (to compute average of the selected series), 'integral'
(to integrate over series), 'maximum' (to evaluate the maximum over series),
'minimum' (to evaluate the minimum), 'rms' (to compute the root mean square),
'stddev' (to compute the standard deviation), or 'variance' (to compute the
variance).

OUTPUT FORMAT
The function mphmean also supports other output formats.

To extract the unit of the evaluated expression, define an extra output variable:

[data,unit] = mphmean(model,<expr>,edim)

where unit is a 1xN cell array and N is the number of expressions to evaluate.

By default mphmean returns the results as a squeezed singleton. To get the full
singleton set the squeeze property to off:

data = mphmean(model,<expr>,edim,'squeeze','off')

Set the property matrix to off to return the data as a cell array instead of a double
array:

data = mphmean(model,<expr>,edim,'matrix','off')

SPECIFY THE INTEGRATION SETTINGS

You can specify integration settings such as an integration method or integration order
to perform the mean operation. The available integration properties are:

• method, specify the integration method, which can be either integration or

summation:

172 | CHAPTER 4: WORKING WITH MODELS

 data = mphmean(model,<expr>,edim,'method',method)

where method can be 'integration' or 'summation'. The default uses the

appropriate method for the given expression.
• intorder, specify the integration order:
data = mphmean(model,<expr>,edim,'intorder',<order>)

where <order> is a positive integer. The default value is 4.

Extracting Data at Node Points

The function mpheval lets you evaluate expressions on node points.

Call the function mpheval as in this command:

pd = mpheval(model, <expr>)

where <expr> is a string cell array that lists the expression to evaluate. The expression
has to be defined in the COMSOL model object in order to be evaluated.

pd is a structure with the following fields:

• expr contains the list of names of the expressions evaluated with mpheval;
• d1 contains the value of the expression evaluated. The columns in the data value
fields correspond to node point coordinates in columns in the field p. In case of
several expressions are evaluated in mpheval, additional field d2, d3,... are available;
• p contains the node point coordinates information. The number of rows in p is the
number of space dimensions;
• t contains the indices to columns in pd.p of a simplex mesh; each column in pd.t
represents a simplex;
• ve contains the indices to mesh elements for each node points; and
• unit contains the list of the unit for each evaluated expressions.

The rest of this section has additional information for the function mpheval:

• Specify the Evaluation Data

• Output Format
• Specify the Evaluation Quality
• Display the Expression in Figures

EXTRACTING RESULTS | 173

 SPECIFY THE EVALUATION DATA
The function mpheval supports the following properties to set the data of the
evaluation to perform:

• dataset, specify the solution dataset to use in the evaluation:

pd = mpheval(model, <expr>, 'dataset', <dsettag>)

<dsettag> is the tag of a solution dataset. The default value is the current solution
dataset of the model. Selection datasets such as Cut Point, Cut Line, Edge, Surface,
and so forth are not supported.
• selection, specify the domain selection for evaluation:
pd = mpheval(model, <expr>, 'selection', <seltag>)

where <seltag> is the tag of a selection node to use for the data evaluation.
<seltag> can also be a positive integer array that corresponds to the domain index
list. The default selection is all domains where the expression is defined. If the
evaluation point does not belong to the specified domain selection, the output value
is NaN.
• edim, specify the element dimension for evaluation:
pd = mpheval(model, <expr>, 'edim', edim)

where edim is one of the strings 'point', 'edge', 'boundary', or 'domain'. It is

also possible to use the corresponding integer which in 3D is in the range from 0 to
3. The default settings correspond to the model geometry space dimension. When
using a lower space dimension value, make sure that the evaluation point
coordinates dimension has the same size.

Use the function mphevalpoint to evaluate expressions at geometric

points (see Evaluating an Expression at Geometry Vertices).

• solnum, specify the inner solution number for data evaluation. Inner solutions are
generated for the following analysis types: time domain, frequency domain,
eigenvalue, or stationary with continuation parameters:
pd = mpheval(model, <expr>, 'solnum', <solnum>)

where <solnum> is an integer array corresponding to the inner solution index.

<solnum> can also be a string:'end' or 'all' to evaluate the expression for the last
inner solution, or all inner solution respectively. By default the evaluation is
performed on all inner solution.

174 | CHAPTER 4: WORKING WITH MODELS

• outersolnum, specify the outer solution number for data evaluation. Outer
solutions are generated with parametric sweeps:
pd = mpheval(model, <expr>, 'outersolnum', <outersolnum>)

where <outersolnum> is a positive integer corresponding to the outer solution

index. <outersolnum> can also be a string, 'all' or 'end', to evaluate the
expression for all or the last outer solution respectively. The default setting uses the
first outer solution for the data evaluation.
• To evaluate the expression data at a specific time use the property t:
pd = mpheval(model, <expr>, 't', <time>)

where <time> is a double array. The default value corresponds to all the stored time
steps.
• phase, specify the phase in degrees:
pd = mpheval(model, <expr>, 'phase', <phase>)

where <phase> is a double value.

• pattern, use Gauss point evaluation:
pd = mpheval(model, <expr>, 'pattern','gauss')

The default evaluation is performed on the Lagrange points.

OUTPUT FORMAT
The function mpheval returns a structure in the MATLAB workspace. You can specify
other output data formats.

To only obtain the data evaluation as a double array, set the property dataonly to on.
This is speeds up the call to COMSOL since the coordinate and element information
is not retrieved.

pd = mpheval(model, <expr>, 'dataonly', 'on')

Returns only the real part in the data evaluation with the property complexout:

pd = mpheval(model, <expr>, 'complexout', 'off')

SPECIFY THE EVALUATION QUALITY

Define mpheval function settings to specify the evaluation quality using these
properties:

• refine, specify the element refinement for evaluation:

EXTRACTING RESULTS | 175

 pd = mpheval(model, <expr>, 'refine', <refine>)

where <refine> is a positive integer. The default value is 1 which set the simplex
mesh identical to the geometric mesh. Many model use second order elements for
which a refine value of 2 must be used to use all the data in the model.
• smooth, specify the smoothing method to enforce continuity on discontinuous data
evaluation:
pd = mpheval(model, <expr>, 'smooth', smooth)

where smooth is either 'none', 'everywhere', or 'internal' (default). Set the

property to none to evaluate the data on elements independently, set to
everywhere to apply the smoothing to the entire geometry, and set to internal
to smooth the quantity inside the geometry (but no smoothing takes place across
borders between domains with different settings). The output with the same data
and same coordinates are automatically merged, which means that the output size
can differ depending on the smoothing method.
• recover, specify the accurate derivative recovery:
pd = mpheval(model, <expr>, 'recover', recover)

where recover is either 'ppr', 'pprint', or 'off' (default). Set the property to
ppr to perform recovery inside domains or set to pprint to perform recovery inside
domains. Because the accurate derivative processing takes time, the property is
disabled by default.

OTHER EVALUATION PROPERTIES

To not use complex-value functions with real inputs, use the property complexfun:

pd = mpheval(model, <expr>, 'complexfun', 'off')

The default value uses complex-valued functions with real inputs.

Use the property matherr to return an error for undefined operations or expressions:

pd = mpheval(model, <expr>, 'matherr', 'on')

DISPLAY THE EXPRESSION IN FIGURES

You can display an expression evaluated with mpheval in an external figure with the
function mphplot (see Displaying The Results).

Evaluating an Expression at Geometry Vertices

The function mphevalpoint returns the result of a given expression evaluated at the
geometry points:

176 | CHAPTER 4: WORKING WITH MODELS

 [d1,...] = mphevalpoint(model,{'e1',...})

where e1, … are the COMSOL expressions to evaluate. The output d1, … is an
N-by-P double array, where N is the number of evaluation points and P the length of
the inner solution.

The rest of this section has additional information for the function mphevalpoint:

• Specify the Evaluation Data

• Output Format

SPECIFY THE EVALUATION DATA

The function mphevalpoint supports the following properties to set the data of the
evaluation to perform:

• dataset, specify the solution dataset to use in the evaluation:

data = mphevalpoint(model,<expr>,'dataset',<dsettag>)
<dsettag> is the tag of a solution dataset. The default value is the current solution
dataset of the model.
• selection, specify the domain selection for evaluation:
data = mphevalpoint(model,<expr>,'selection',<seltag>)

where <seltag> is the tag of a selection node to use for the data evaluation.
<seltag> can also be a positive integer array that corresponds to the domain index
list. The default selection is all domains where the expression is defined. If the
evaluation point does not belong to the specified domain selection, the output value
is NaN.
• solnum, specify the inner solution number for data evaluation. Inner solutions are
generated for the following analysis types: time domain, frequency domain,
eigenvalue, or stationary with continuation parameters:
data = mphevalpoint(model,<expr>,'solnum',<solnum>)

where <solnum> is an integer array corresponding to the inner solution index.

<solnum> can also be a string:'end' or 'all' to evaluate the expression for the last
inner solution, or all inner solution respectively. By default the evaluation is
performed on all inner solution.
• outersolnum, specify the outer solution number for data evaluation. Outer
solutions are generated with parametric sweeps:

EXTRACTING RESULTS | 177

 data = mphevalpoint(model,<expr>,'outersolnum',<outersolnum>)

where <outersolnum> is a positive integer corresponding to the outer solution

index. <outersolnum> can also be a string, 'all' or 'end', to evaluate the
expression for all or the last outer solution respectively. The default settings use the
first outer solution for the data evaluation.
• To evaluate the expression data at a specific time use the property t:
data = mphevalpoint(model,<expr>,'t',<time>)

where <time> is a double array. The default value corresponds to all the stored time
steps.

Perform a data series operation with the dataseries property:

data = mphevalpoint(model,<expr>,'dataseries',dataseries)

where dataseries is either 'mean', 'int', 'max', 'min', 'rms', 'std', or 'var'.
Depending on the property value, mphevalpoint performs the following operations
— mean, integral, maximum, minimum, root mean square, standard deviation, or
variance, respectively.

When performing a minimum or maximum operation on the data series, you can
specify to perform the operation using the real or the absolute value. Set the property
minmaxobj to 'real' or 'abs', respectively:

data = mphevalpoint(model,<expr>,'dataseries',dataseries,...
'minmaxobj', valuetype)

By default valuetype is 'real'.

OUTPUT FORMAT
The function mphevalpoint supports other output formats.

To extract the unit of the evaluated expression, define an extra output variable:

[data,unit] = mphevalpoint(model,<expr>)

with unit is a 1xN cell array where N is the number of expressions to evaluate.

By default, mphevalpoint returns the results as a squeezed singleton. To get the full
singleton set the squeeze property to off:

data = mphevalpoint(model,<expr>,'squeeze','off')

Set the property matrix to off to return the data as a cell array instead of a double
array:

data = mphevalpoint(model,<expr>,'matrix','off')

178 | CHAPTER 4: WORKING WITH MODELS

Evaluating Expressions on Particle/Ray Trajectories
Evaluate expressions on particle trajectories with the function mphparticle and on
and ray trajectories with the function mphray.

mphray supports only Ray Trajectories datasets. mphparticle supports

both Ray Trajectories and Particle Trajectories datasets.

In this section you can replace the command mphparticle and mphray
as they support the same properties.

Evaluate expressions on particle and ray trajectories with either the function
mphparticle or mphray.

To evaluate the particle position and the particle velocity run mphparticle as in this
command:

pd = mphparticle(model)

pd is a structure containing the information about particle position and particle

velocity at every time step. The information is stored in the following fields:

• p contains the coordinates of the particle position along the trajectories. The data
are stored in a NxMxL array where N is the number of time steps, M the number of
evaluation point along the particle trajectories, and L the evaluation space
dimension.
• v contains the value of the particle velocity along the trajectories. The data are
stored in a NxMxL array where N is the number of time steps, M the number of
evaluation points along the particle trajectories, and L the evaluation space
dimension.
• t contains the list of evaluation time.

You can also specify expressions to evaluate along the particle trajectories. Run the
function mphparticle as in this command:

pd = mphparticle(model,'expr','e1')

where 'e1' is the expression to evaluate along the particle trajectories. The output
structure pd contains the fields p, v, and t (described above) with the following ones:

• unit contains the unit of the evaluated expression;

EXTRACTING RESULTS | 179

 • d1 contains the value of the expression. The data are stored in a NxM array where
N is the number of time steps and M the number of evaluation points along the
particle trajectories; and
• expr contains the list of the evaluated expression.

Use a string cell array to evaluate several expressions at once. The result of the
evaluation is then stored in the field d1,... corresponding to each evaluated
expression.

SPECIFY THE EVALUATION DATA

The function mphparticle supports the following properties to set the data of the
evaluation to perform:

• dataset, specify the solution dataset to use in the evaluation:

pd = mphparticle(model,'expr',<expr>,'dataset',<dsettag>)

<dsettag> is the tag of a particle solution dataset. The default value is the current
particle solution dataset of the model.
• To evaluate the expression data at a specific time use the property t:
pd = mphparticle(model,'expr',<expr>,'t',<time>)

where <time> is a double array. The default value corresponds to all the stored time
steps.

OUTPUT FORMAT
The function mphparticle also supports other output formats.

Set the property dataonly to on to return only the data related to the specified
expression:

pd = mphparticle(model,'expr',<expr>,'dataonly','on')

The output structure pd only contains the field unit, d#, expr, and t (described
above).

Evaluating a Global Expression

Evaluate a global expression with the function mphglobal.

To evaluate a global expression at the MATLAB® prompt, call the function

mphglobal as in this command:

[d1,...] = mphglobal(model,{'e1',...})

180 | CHAPTER 4: WORKING WITH MODELS

where e1,... are the COMSOL Multiphysics global expressions to evaluate. The
output values d1,... are returned as a Px1 double array, with P the length of inner
parameters.

The rest of this section has additional information for the function mphglobal:

• Specify the Evaluation Data

• Output Format
• Other Evaluation Properties

SPECIFY THE EVALUATION DATA

The function mphglobal supports the following properties to set the data of the
evaluation to perform:

• dataset, specify the solution dataset to use in the evaluation:

data = mphglobal(model,<expr>,'dataset',<dsettag>)

<dsettag> is the tag of a solution dataset. The default value is the current solution
dataset of the model.
• solnum, specify the inner solution number for data evaluation. Inner solutions are
generated for the following analysis types: time domain, frequency domain,
eigenvalue, or stationary with continuation parameters:
data = mphglobal(model,<expr>,'solnum',<solnum>)

where <solnum> is an integer array corresponding to the inner solution index.

<solnum> can also be a string:'end' or 'all' to evaluate the expression for the last
inner solution or all inner solutions, respectively. By default the evaluation is
performed on all inner solutions.
• outersolnum, specify the outer solution number for data evaluation. Outer
solutions are generated with parametric sweeps:
data = mphglobal(model,<expr>,'outersolnum',<outersolnum>)

where <outersolnum> is a positive integer corresponding to the outer solution

index. <outersolnum> can also be a string, 'all' or 'end' to evaluate the
expression for all or the last outer solution, respectively. The default settings uses the
first outer solution for the data evaluation.
• To evaluate the expression data at a specific time use the property t:
data = mphglobal(model,<expr>,'t',<time>)

where <time> is a double array. The default value corresponds to all the stored time
steps.

EXTRACTING RESULTS | 181

 • phase, specify the phase in degrees:
data = mphglobal(model,<expr>,'phase',<phase>)

where <phase> is a double value.

OUTPUT FORMAT
The function mphglobal also supports other output formats.

To extract the unit of the evaluated expression, define an extra output variable:

[data,unit] = mphglobal(model,<expr>)

with unit is a 1xN cell array where N is the number of expressions to evaluate.

Returns only the real part in the data evaluation with the property complexout:

data = mphglobal(model,<expr>,'complexout','off')

OTHER EVALUATION PROPERTIES

Set the unit property to specify the unit of the evaluation:

data = mphglobal(model,<expr>,'unit',<unit>)

where <unit> is a cell array with the same length as <expr>.

Use the property matherr to return an error for undefined operations or expressions:

data = mphglobal(model,<expr>,'matherr','on')

Evaluating a Matrix Expression at Points

The function mphevalpointmatrix returns the result of a given matrix expression
evaluated at the points:

M = mphevalpointmatrix(model,<expr>,...)

where <expr> is the COMSOL expressions to evaluate. The output M is a matrix.

The rest of this section has additional information for the function
mphevalpointmatrix:

• Specify the Evaluation Data

• Output Format

182 | CHAPTER 4: WORKING WITH MODELS

SPECIFY THE EVALUATION DATA
The function mphevalpoint supports the following properties to set the data of the
evaluation to perform:

• dataset, specify the solution dataset to use in the evaluation:

M = mphevalpointmatrix(model,<expr>,'dataset',<dsettag>)
<dsettag> is the tag of a solution dataset. The default value is the current solution
dataset of the model.
• selection, specify the domain selection for evaluation:
M = mphevalpointmatrix(model,<expr>,'selection',<seltag>)

where <seltag> is the tag of a selection node to use for the data evaluation.
<seltag> can also be a positive integer array that corresponds to the domain index
list. The default selection is all domains where the expression is defined. If the
evaluation point does not belong to the specified domain selection, the output value
is NaN.
• solnum, specify the inner solution number for data evaluation. Inner solutions are
generated for the following analysis types: time domain, frequency domain,
eigenvalue, or stationary with continuation parameters:
M = mphevalpointmatrix(model,<expr>,'solnum',<solnum>)

where <solnum> is an integer array corresponding to the inner solution index.

<solnum> can also be a string:'end' or 'all' to evaluate the expression for the last
inner solution, or all inner solution respectively. By default the evaluation is
performed on all inner solution.
• outersolnum, specify the outer solution number for data evaluation. Outer
solutions are generated with parametric sweeps:
M = mphevalpoint(model,<expr>,'outersolnum',<outersolnum>)

where <outersolnum> is a positive integer corresponding to the outer solution

index. <outersolnum> can also be a string, 'all' or 'end', to evaluate the
expression for all or the last outer solution respectively. The default settings use the
first outer solution for the data evaluation.
• To evaluate the expression data at a specific time use the property t:
M = mphevalpointmatrix(model,<expr>,'t', <time>)

where <time> is a double array. The default value corresponds to all the stored time
steps.
• Perform a data series operation with the dataseries property:
M = mphevalpointmatrix(model,<expr>,'dataseries', dataseries)

EXTRACTING RESULTS | 183

 where dataseries is either 'none', 'average' or 'sum'.

Evaluating a Global Matrix

mphevalglobalmatrix evaluates the matrix variable such as S-parameters in a model
with several ports activated as a parametric sweep and a frequency-domain study.

Note: S-parameters evaluation requires the AC/DC Module or the RF Module.

To evaluate the global matrix associated to the expression <expr>, enter the
command:

M = mphevalglobalmatrix(model,<expr>)

The output data M is an NxN double array, where N is the number of port boundary
condition set in the model.

The rest of this section has additional information for the function
mphevalglobalmatrix:

• Specify the Evaluation Data

• Specify Matrix transformation

SPECIFY THE EVALUATION DATA

The function mphevalglobalmatrix supports the following properties to set the data
of the evaluation to perform:

• Set the solution dataset for evaluation with the property dataset:
M = mphevalglobalmatrix(model,<expr>,'dataset',<dsettag>)

where <dsettag> is the tag of a solution data.

• solnum, specify the inner solution number for data evaluation. Inner solutions are
generated for the following analysis types: time domain, frequency domain,
eigenvalue, or stationary with continuation parameters:
M = mphevalglobalmatrix(model,<expr>,'solnum',<solnum>)

where <solnum> is an integer array corresponding to the inner solution index.

<solnum> can also be a string:'end' or 'all' to evaluate the expression for the last
inner solution or all inner solutions, respectively. By default the evaluation is
performed on all inner solutions.

184 | CHAPTER 4: WORKING WITH MODELS

• outersolnum, specify the outer solution number for data evaluation. Outer
solutions are generated with parametric sweeps:
M = mphevalglobalmatrix(model,<expr>,'outersolnum',<outersolnum>)

where <outersolnum> is a positive integer corresponding to the outer solution

index. <outersolnum> can also be a string, 'all' or 'end' to evaluate the
expression for all or the last outer solution, respectively. The default settings uses the
first outer solution for the data evaluation.
• To evaluate the expression data at a specific time use the property t:
dM = mphevalglobalmatrix(model,<expr>,'t',<time>)

where <time> is a double array. The default value corresponds to all the stored time
steps.
• Perform a data series operation with the dataseries property:
M = mphevalglobalmatrix(model,<expr>,'dataseries', dataseries)

where dataseries is either 'none', 'average', or 'sum'.

• Perform a data series operation on outer data with the outerdataseries property:
M = mphevalglobalmatrix(model,<expr>,'outerdataseries',
outerdataseries)

where outerdataseries is either 'none', 'average', or 'sum'.

SPECIFY MATRIX TRANSFORMATION

To apply a transformation operation to compute the inverse of the matrix variable or
to convert between the impedance matrix, Z, the admittance matrix, Y, and the
S-parameter matrix S, use the trans property:

M = mphevalglobalmatrix(model,<expr>,'trans',<trans>)

where <trans> can be either: 'maxwellmutual' (from Maxwell to mutual

capacitance), 'mutualmaxwell' (from mutual to Maxwell capacitance), 'none' (no
transformation), 'inverse' (compute the inverse of the matrix), 'sy' (from S to Y
transformation), 'sz' (from S to Z transformation), 'ys' (from Y to S
transformation), 'yz' (from Y to Z transformation), 'zs' (from Z to S
transformation), or 'zy' (from Z to Y transformation).

For S to Y and Y to S transformation you need to specify the characteristic admittance,

to proceed use the command:

M = mphevalglobalmatrix(model,<expr>,'trans','sy','y0',<value>)
M = mphevalglobalmatrix(model,<expr>,'trans','ys','y0',<value>)

EXTRACTING RESULTS | 185

 where <value> is the characteristic admittance in siemens (S). The default value is 1 S.

For S to Z and Z to S transformation you need to specify the characteristic impedance,

to proceed use the command:

M = mphevalglobalmatrix(model,<expr>,'trans','sz','z0',<value>)

M = mphevalglobalmatrix(model,<expr>,'trans','zs','z0',<value>)

where <value> is the characteristic admittance in ohm (Ω).The default value is 1 Ω.

Global Matrix Evaluation section in the COMSOL Reference Manual.

Extracting Data From Tables

In the Table node you can store the data evaluated with the COMSOL Multiphysics
built-in evaluation method (see The Numerical Node Syntax).

Use mphtable to extract the data stored in the table with the tag <tbltag>. Enter:

tabl = mphtable(model,<tbltag>)

This creates a structure tabl made with the following fields:

• headers for the table,

• tag of the table,
• data of the extracted table, and
filename when the table is exported to file.

186 | CHAPTER 4: WORKING WITH MODELS

Running Models in a Loop
A common use of LiveLink™ for MATLAB® is to run models in a loop. MATLAB
provides several functionalities to run loops, including conditional statements and
error handling, and this section shows how to use that functionality together with the
COMSOL API syntax to run COMSOL Multiphysics® models in loops.

In this section:

• The Parametric Sweep Node

• Running Model in a Loop Using the MATLAB® Tools

The Parametric Sweep Node

Using the COMSOL API you can run models in loops. See Adding a Parametric
Sweep in the section Building Models.

By using the COMSOL built-in function to run models in loops, you can
ensure the model is saved automatically at each iteration. COMSOL also
offers tools to take advantage of clusters and distributed computer
architectures.

Running Model in a Loop Using the MATLAB® Tools

Use MATLAB® tools such as for or while statements to run your model in a loop.
The COMSOL API commands can be included in scripts using MATLAB commands.
To evaluate such a script you need to have MATLAB connected to a COMSOL server.

To run a model in a loop you do not need to run the entire M-file’s commands from
scratch. It is recommended to load a COMSOL model in MATLAB and run the loop
only over the desired operations. The COMSOL model is automatically updated when
running the study node.

RUNNING MODELS IN A LOOP | 187

 You can run an M-file for a model from scratch if required, for example, to generate
the geometry in loop.

The model run inside a MATLAB loop is not automatically saved. Make
sure to save the model at each iteration using the command mphsave to
save the model object.

If you are not interested in saving the entire model object at each
iteration, you can extract data and store it in the MATLAB workspace.
See Extracting Results to find the most suitable function to your model.

When running loops in MATLAB, the iteration progress is taken care of by MATLAB;
only the COMSOL commands are executed in the COMSOL server.

You can generate as many nested loops as needed and combine the loop with other
MATLAB conditional statements such as if and switch or error handling statements
such as try/catch. Or break the loop with break, or jump to the next loop iteration
with continue.

See the MATLAB help for more information about the MATLAB
commands for, while, if, switch, try/catch, break, and continue.

GEOMETRY PARAMETERIZATION
This example shows how to proceed to geometry parameterization using a MATLAB
for loop. The model consists of the busbar example available in the COMSOL
Multiphysics Applications Libraries; see the Introduction to COMSOL Multiphysics.

In this example the loop iterates over the busbar’s width, wbb. The solution for each
parameter value is displayed using the second plot group defined in the COMSOL
model. All the results are plotted in the same figure.

188 | CHAPTER 4: WORKING WITH MODELS

The results from the computation display in these plots:

Code for use with MATLAB ®

model = mphopen('busbar');
w = [5e-2 10e-2 15e-2 20e-2];
for i = 1:4
model.param.set('wbb',w(i));
model.study('std1').run;
subplot(2,2,i);
mphplot(model,'pg4','rangenum',1);
end

RUNNING MODELS IN A LOOP | 189

 Running Models in Batch Mode
Use LiveLink™ for MATLAB® to model in batch mode. At the MATLAB prompt you
can execute commands to set up the batch job using the COMSOL Multiphysics®
built-in method or run custom scripts directly from a command line. In this section:

• The Batch Node

• Running an M-File in Batch Mode
• Running an M-File in Batch Mode Without Display

The Batch Node

Using the COMSOL API you can run models in a loop. See Adding a Job Sequence.

Running an M-File in Batch Mode

Running COMSOL with MATLAB® in batch mode requires that you

have xterm installed on your machine. If this is not the case see Running
an M-File in Batch Mode Without Display.

To run in batch an M-script that runs COMSOL Model is required. Start COMSOL
with MATLAB at a terminal window with this command:

comsol mphserver matlab myscript

where myscript is the M-script, saved as myscript.m, that contains the operation to
run at the MATLAB prompt.

COMSOL Multiphysics does not automatically save the model. You need
to make sure that the model is saved before the end of the execution of
the script. See Loading and Saving a Model.

You can also run the script in batch without the MATLAB desktop and the MATLAB
splash screen. Enter this command:

comsol mphserver matlab myscript -nodesktop -mlnosplash

190 | CHAPTER 4: WORKING WITH MODELS

Running an M-File in Batch Mode Without Display
To connect COMSOL with a MATLAB® terminal requires that xterm is installed on
the machine. If this is not the case as it might be for a computation COMSOL server,
a workaround is to connect manually MATLAB to a COMSOL server with the
function mphstart.

These steps describe how to run an M-script that runs a COMSOL model:

1 In a system terminal prompt start a COMSOL Multiphysics Server with the

command:
comsol mphserver -silent &

2 In the same terminal window change the path to the COMSOL installation
directory:
cd COMSOL_path/mli

3 From that location, start MATLAB without display and run the mphstart function
in order to connect MATLAB to COMSOL:
matlab -nodesktop -nosplash -r "mphstart; myscript"

For more information about how to connect MATLAB to a COMSOL server see
Starting COMSOL® with MATLAB® on Windows ®/ Mac OSX / Linux®.

RUNNING MODELS IN BATCH MODE | 191

 Working with Matrices
In this section:

• Extracting System Matrices

• Set System Matrices in the Model
• Extracting State-Space Matrices
• Extracting Reduced Order State-Space Matrices

Extracting System Matrices

Extract the matrices of the COMSOL Multiphysics linearized system with the function
mphmatrix. To call the function mphmatrix, specify a solver node and the list of the
system matrices to extract:

str = mphmatrix(model, <soltag>, 'out', out)

where <soltag> is the solver node tag used to assemble the system matrices and out
is a cell array containing the list of the matrices to evaluate. The output data str
returned by mphmatrix is a MATLAB® structure, and the fields correspond to the
assembled system matrices.

See the Advanced section and the Assemble section in the COMSOL
Multiphysics Reference Manual, for more information about matrix
evaluation.

The system matrices that can be extracted with mphmatrix are listed in the table:

EXPRESSION DESCRIPTION

K Stiffness matrix
L Load vector
M Constraint vector
N Constraint Jacobian
D Damping matrix
E Mass matrix
NF Constraint force Jacobian

192 | CHAPTER 4: WORKING WITH MODELS

EXPRESSION DESCRIPTION

NP Optimization constraint Jacobian (*)

MP Optimization constraint vector (*)
MLB Lower bound constraint vector (*)
MUB Upper bound constraint vector (*)
Kc Eliminated stiffness matrix
Lc Eliminated load vector
Dc Eliminated damping matrix
Ec Eliminated mass matrix
Null Constraint null-space basis
Nullf Constraint force null-space matrix
ud Particular solution ud
uscale Scale vector
(*) Requires the Optimization Module.

SELECTING LINEARIZATION POINTS

The default selection of linearization points for the system matrix assembly is the
current solution of the solver node associated to the assembly.

If the linearization point is not specified when calling mphmatrix, the

COMSOL Multiphysics software automatically runs the entire solver
configuration before assembling and extracting the matrices.

Save time during the evaluation by manually setting the linearization point. Use the
initmethod property as in this command:

str = mphmatrix(model,<soltag>,'out',out,'initmethod',method)

where method corresponds to the type of linearization point — the initial value
expression ('init') or a solution ('sol').

To set the solution to use for the linearization point, use the property initsol:

str = mphmatrix(model,<soltag>,'out',out,'initsol',<initsoltag>)

where <initsoltag> is the solver tag to use for linearization points. You can also set
the initsol property to 'zero', which corresponds to using a null solution vector as
a linearization point. The default is the current solver node where the assemble node
is associated.

WORKING WITH MATRICES | 193

 For continuation, time-dependent, or eigenvalue analyses you can set the solution
number to use as a linearization point. Use the solnum property:

str = mphmatrix(model,<soltag>,'out',out,'solnum',<solnum>)

where <solnum> is an integer value corresponding to the solution number. The default
value is the last solution number available with the current solver configuration.

See Retrieving Xmesh Information to learn how to get relation between

the degrees of freedom information in the matrix system and coordinates
or element information.

SPECIFYING WHEN TO ASSEMBLE THE MATRICES IN THE SOLUTION

SEQUENCE
You can specify when in the solution sequence to assemble the system matrices, for
instance after computing the solution or if you have a solution sequence combining
different solver. By default the system matrices are assembled before running the first
solver, just after the first dependent variable node in the solution sequence. To specify
the node that precede the matrix extraction use the extractafter property:

str = mphmatrix(model,<soltag>,'out',out,'extractafter',<nodetag>)

where <nodetag> is the tag of a solution sequence node such as dependent variable or
solver nodes.

EIGENVALUE PROBLEMS
For eigenvalue problems, it is necessary to specify the eigenvalue name and the
eigenvalue linearization point. Used the property eigname to specify the name of the
eigenvalue and eigref to specify the value of eigenvalue linearization point:

str = mphmatrix(model,<soltag>,'out',out,'eigname',<eigname>)

str = mphmatrix(model,<soltag>,'out',out,'eigname',<eigname>,...
'eigref', <eigref>)

where <eigname> is a string and <eigref> a double.

ROW EQUILIBRATION, MATRIX SYMMETRY AND NULL-SPACE FUNCTION

The default assembly of the system matrices assumes row equilibration of the system
matrices. It is however possible to extract the unscaled matrices, to proceed set the
rowscale property to off:

str = mphmatrix(model,<soltag>,'out',out,'rowscale','off')

194 | CHAPTER 4: WORKING WITH MODELS

Set symmetry property to specify manually the symmetry type for the matrix
evaluation. The symmetry property support the following values:

str = mphmatrix(model,<soltag>,'out',out,'symmetry',sym)

where sym can be either of one of the following value:

• 'on', to assemble and extract the system matrices as symmetric.

• 'off', to assemble and extract the system matrices as non-symmetric.
• 'hermitian', to assemble and extract the system matrices as hermitian.
• 'auto', to let the solver assembly determine the type of the system matrices.

Use the nullfun property to specify the method for computation of matrices needed
for constraint handling:

• 'flnullorth', a method based on singular value decomposition;

• 'flspnull', to handle constraint matrices with nonlocal couplings using singular
sparse algorithm;
• 'explicitsp', to handle constraints by explicitly eliminating the DOFs on the
destination side of the explicit constraints. The remaining constraints are handled
using the Sparse method.
• 'explicitorth', to handle constraints by explicitly eliminating the DOFs on the
destination side of the explicit constraints. The remaining constraints are handled
using the Orthonormal method.
• 'auto', to let the software automatically determine the most appropriate method,
which uses an explicit handling of nodal constraints and one of the Orthonormal or
Sparse methods for the remaining constraints.

COMPLEX FUNCTION
If the system contains complex function, use the property complexfun to specify how
to handle such a function. Set this property to on to use complex-valued function with
real input:

str = mphmatrix(model,<soltag>,'out',out,'complexfun','on');

HANDLING UNDEFINED OPERATIONS

It is possible to disable the error for undefined operations during the assembly and
matrix evaluation, to proceed set the property matherr to off as in the command
below:

str = mphmatrix(model,<soltag>,'out',out,'matherr','off')

WORKING WITH MATRICES | 195

 EXTRACTING THE SYSTEM MATRICES
The following illustrates how to use the mphmatrix command to extract eliminated
system matrices of a stationary analysis and linear matrix system at the MATLAB
prompt.

The model consists of a linear heat transfer problem solved on a unit square with a 1e5
W/m^2 surface heat source and temperature constraint. Only one quarter of the
geometry is represented in the model. For simplification reasons, the mesh is made of
four quad elements and the discretization is set with linear element.

These commands set the COMSOL model object:

model = ModelUtil.create('Model2');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 2);
geom1.feature.create('sq1', 'Square');
geom1.run;

mat1 = comp1.material.create('mat1');
def = mat1.materialModel('def');
def.set('thermalconductivity',4e2);

ht = comp1.physics.create('ht', 'HeatTransfer', 'geom1');

ht.prop('ShapeProperty').set('boundaryFlux_temperature',false);
ht.prop('ShapeProperty').set('order_temperature',1);
hs1 = ht.feature.create('hs1','HeatSource',2);
hs1.selection.set(1);
hs1.set('Q',1,1e5);

temp1 = ht.feature.create('temp1','TemperatureBoundary',1);
temp1.selection.set([1 2]);

mesh1 = comp1.mesh.create('mesh1');
dis1 = mesh1.feature.create('dis1','Distribution');
dis1.selection.set([1 2]);
dis1.set('numelem',2);
mesh1.feature.create('map1','Map');

std1 = model.study.create('std1');
std1.feature.create('stat','Stationary');
std1.run;

To extract the solution vector of the computed solution, run the function mphgetu as
in this command:

U = mphgetu(model);

196 | CHAPTER 4: WORKING WITH MODELS

To assemble and extract the eliminated stiffness matrix and the eliminated load vector,
set the linearization point to the initial value expression by entering:

MA = mphmatrix(model ,'sol1', ...

'Out', {'Kc','Lc','Null','ud','uscale'},...
'initmethod','sol','initsol','zero');

Solve for the eliminated solution vector using the extracted eliminated system:

Uc = MA.Null*(MA.Kc\MA.Lc);

Combine the eliminated solution vector and the particular vector:

U0 = Uc+MA.ud;

Scale back the solution vector:

U1 = U0.*MA.uscale;

Now compare both solution vector U and U1 computed by COMSOL Multiphysics

and by the matrix operation, respectively.

Code for use with MATLAB ®

model = ModelUtil.create('Model');
comp1 = model.component.create('comp1', true);
geom1 = comp1.geom.create('geom1', 2);
geom1.feature.create('sq1', 'Square');
geom1.run;
mat1 = comp1.material.create('mat1');
def = mat1.materialModel('def');
def.set('thermalconductivity',4e2);
ht = comp1.physics.create('ht', 'HeatTransfer', 'geom1');
ht.prop('ShapeProperty').set('boundaryFlux_temperature',false);
ht.prop('ShapeProperty').set('order_temperature',1);
hs1 = ht.feature.create('hs1','HeatSource',2);
hs1.selection.set(1);
hs1.set('Q',1,1e5);
temp1 = ht.feature.create('temp1','TemperatureBoundary',1);
temp1.selection.set([1 2]);
mesh1 = comp1.mesh.create('mesh1');
dis1 = mesh1.feature.create('dis1','Distribution');
dis1.selection.set([1 2]);
dis1.set('numelem',2);
mesh1.feature.create('map1','Map');
std1 = model.study.create('std1');
std1.feature.create('stat','Stationary');
std1.run;
U = mphgetu(model);
MA = mphmatrix(model ,'sol1', ...
'Out', {'Kc','Lc','Null','ud','uscale'},...
'initmethod','sol','initsol','zero');

WORKING WITH MATRICES | 197

 Uc = MA.Null*(MA.Kc\MA.Lc);
U0 = Uc+MA.ud;
U1 = U0.*MA.uscale;

Set System Matrices in the Model

Use the function mphinputmatrix to set a linear matrix system to a model:

mphinputmatrix(model,<str>,<soltag>,<soltypetag>)

This command set the matrices of a linear system stored in the MATLAB® structure
<str> into the model. The linear system is associated to the solver sequence <soltag>
and is to be solved by the solver <soltypetag>.

mphinputmatrix only supports the solver types Stationary, Eigenvalue, and Time.

A valid structure <str> for a stationary solver includes the following fields:

FIELD DESCRIPTION

K Stiffness matrix
L Load vector
M Constraint vector
N Constraint Jacobian

A valid structure <str> for a time-dependent or an eigenvalue solver includes the

following fields:

EXPRESSION DESCRIPTION

K Stiffness matrix
L Load vector
M Constraint vector
N Constraint Jacobian
D Damping matrix
E Mass matrix

You can also include the Constraint force Jacobian vector, defined in the field NF.

Once the linear system is loaded in the model, you can directly run the solver.

The system matrices are not stored in the model when it is saved in the
MPH-format or loaded to the COMSOL Desktop.

198 | CHAPTER 4: WORKING WITH MODELS

SETTING A MODEL WITH A MODIFIED MATRIX SYSTEM
This example deals with heat transfer in solids physics. The geometry and physics
settings are already set in the model and saved in the MPH-format. The Model
MPH-file comes with the COMSOL installation.

At the MATLAB prompt you load the model and add an additional line heat source to
the model directly in the system matrix by manually changing the load vector. Then
compute the solution of the modified system in COMSOL.

Load the base Model MPH-file and display the geometry:

model = mphopen('model_tutorial_llmatlab.mph');
mphgeom(model)

This results in the following MATLAB figure:

Draw the line to be used as a line heat source in the model and plot the modified
geometry:

comp1 = model.component('comp1');
b1 = comp1.geom('geom1').feature.create('b1', 'BezierPolygon');
b1.set('p', {'1e-2' '5e-2'; '1e-2' '5e-2'; '1e-2' '1e-2'});
mphgeom(model,'geom1','edgelabels','on','facealpha',0.5);

WORKING WITH MATRICES | 199

 In the figure below you can see that the added line as the index 21:

Generate a mesh with finer mesh settings:

mesh1 = comp1.mesh('mesh1');
mesh1.feature.create('ftet1', 'FreeTet');
mesh1.feature('size').set('hauto', 3);
mesh1.run;
mphmesh(model)

Set the solver sequence associated to a stationary study node:

std1 = model.study.create('std1');
std1.feature.create('stat', 'Stationary');
sol1 = model.sol.create('sol1');
sol1.study('std1');
st1 = sol1.feature.create('st1', 'StudyStep');
st1.set('studystep', 'stat');

200 | CHAPTER 4: WORKING WITH MODELS

 v1 = sol1.feature.create('v1', 'Variables');
v1.set('control', 'stat');
sol1.feature.create('s1', 'Stationary');

Set the dependent variable discretization with linear shape function:

Shape = comp1.physics('ht').prop('ShapeProperty');
Shape.set('order_temperature', 1, 1);

The heat transfer interface automatically compute for internal DOFs in order to
evaluate fluxes accurately at the boundaries. Deactivate the internal DOFs with this
command:

Shape.set('boundaryFlux_temperature', false);

Now extract the matrices of the linear system associated to the solver sequence sol1:

ME = mphmatrix(model,'sol1','Out',{'K' 'L' 'M' 'N'},...

'initmethod','sol','initsol','zero');

To retrieve the degrees of freedom that belong to edge 21, you need to get the
geometric mesh data:

[stats,data] = mphmeshstats(model);

With the mesh data structure data, you can get the element indices that belong to
edge 2. Use the MATLAB find function to list all the indices:

elem_idx = find(data.elementity{1}==21)'

With the function mphxmeshinfo, retrieve the finite element mesh information
associated to solver sequence sol1:

info = mphxmeshinfo(model,'soltag','sol1','studysteptag','v1');

In the info structure you can get the DOFs indices that belong to the edge element
defined with the indices elem_idx:

dofs = info.elements.edg.dofs;
edgdofs_idx = [];
for i = 1:length(elem_idx)
edgdofs_idx = [edgdofs_idx; dofs(:,elem_idx(i))];
end

edgdofs_idx might contain duplicate DOFs indices. This is because the information
is from the element level; the duplicate indices correspond to the connecting node
between two adjacent elements.

First remove the duplicate entities:

unique_idx = unique(edgdofs_idx);

WORKING WITH MATRICES | 201

 Edit the load vector for the DOF that belong to edge 21, the total applied power is
50 W:

ME.L(unique_idx+1) = 50/length(unique_idx);

Now that the linear system has been modified, set it back in the model:

mphinputmatrix(model,ME,'sol1','s1')

Note: mphmatrix only assembles the matrix system for the dofs solved in the specified
solver configuration. mphinputmatrix insert the matrix system as defined by the user.
When inserting matrices in an existing model, the solution format may not be
compatible with the inserted system matrices.

In order to have a compatible xmesh solution format compatible with the size of the
inserted matrices, add a new equation form physics interface, solving only for one
variable.

gForm = comp1.physics.create('g', 'GeneralFormPDE', {'u'});

gForm.prop('ShapeProperty').set('order', 1);
gForm.prop('ShapeProperty').set('boundaryFlux', false);

Disable the Heat Transfer physics interface.

comp1.physics('ht').active(false);

Compute the solution of the added system:

model.sol('sol1').runAll;

Display the solution:

pg1 = model.result.create('pg1', 'PlotGroup3D');

pg1.feature.create('surf1', 'Surface');

202 | CHAPTER 4: WORKING WITH MODELS

 mphplot(model,'pg1','rangenum',1)

Code for use with MATLAB ®

model = mphopen('model_tutorial_llmatlab.mph');
mphgeom(model)
comp1 = model.component('comp1');
b1 = comp1.geom('geom1').feature.create('b1', 'BezierPolygon');
b1.set('p', {'1e-2' '5e-2'; '1e-2' '5e-2'; '1e-2' '1e-2'});
mphgeom(model,'geom1','edgelabels','on','facealpha',0.5);
mesh1 = comp1.mesh('mesh1');
mesh1.feature.create('ftet1', 'FreeTet');
mesh1.feature('size').set('hauto', 3);
mesh1.run;
mphmesh(model)
std1 = model.study.create('std1');
std1.feature.create('stat', 'Stationary');
sol1 = model.sol.create('sol1');
sol1.study('std1');
st1 = sol1.feature.create('st1', 'StudyStep');
st1.set('studystep', 'stat');
v1 = sol1.feature.create('v1', 'Variables');
v1.set('control', 'stat');
sol1.feature.create('s1', 'Stationary');
Shape = comp1.physics('ht').prop('ShapeProperty');
Shape.set('order_temperature', 1, 1);
Shape.set('boundaryFlux_temperature', false);
ME = mphmatrix(model,'sol1','Out',{'K' 'L' 'M' 'N'},...
'initmethod','sol','initsol','zero');
[stats,data] = mphmeshstats(model);
elem_idx = find(data.elementity{1}==21)'
info = mphxmeshinfo(model,'soltag','sol1','studysteptag','v1');
dofs = info.elements.edg.dofs;
edgdofs_idx = [];
for i = 1:length(elem_idx)

WORKING WITH MATRICES | 203

 edgdofs_idx = [edgdofs_idx; dofs(:,elem_idx(i))];
end
unique_idx = unique(edgdofs_idx);
ME.L(unique_idx+1) = 50/length(unique_idx);
mphinputmatrix(model,ME,'sol1','s1')
gForm = comp1.physics.create('g', 'GeneralFormPDE', {'u'});
gForm.prop('ShapeProperty').set('order', 1);
gForm.prop('ShapeProperty').set('boundaryFlux', false);
comp1.physics('ht').active(false);
model.sol('sol1').runAll;
pg1 = model.result.create('pg1', 'PlotGroup3D');
pg1.feature.create('surf1', 'Surface');
mphplot(model,'pg1','rangenum',1)

Extracting State-Space Matrices

Use state-space export to create a linearized state-space model corresponding to a
COMSOL Multiphysics model. You can export the matrices of the state-space form
directly to the MATLAB® workspace with the command mphstate.

This section includes information about The State-Space System, how to Extract
State-Space Matrices and Set Linearization Points and has an Extracting State-Space
Matrices.

THE STATE-SPACE SYSTEM

A state-space system is the mathematical representation of a physical model. The
system consistent in an ODE linking input, output, and state-space variable. A dynamic
system can be represented with the following system:

 dx
 ------- = Ax + Bu
 dt
 y = Cx + Du


An alternative representation of the dynamic system is:

·
M C x = M C Ax + M C Bu
y = Cx + Du

where x is the state variable vector.

If the components of the mass matrix MC are small, it is possible to approximate the
·
dynamic state-space model with a static model, where M C x = 0 :

204 | CHAPTER 4: WORKING WITH MODELS

 –1
y = ( D – C ( M C A ) M C B )u

Let Null be the PDE constraint null-space matrix and ud a particular solution fulfilling
the constraints. The solution vector U for the PDE problem can then be written

U = Null x + ud + u 0

where u0 is the linearization point, which is the solution stored in the sequence once
the state-space export feature is run.

CHOOSING THE INPUT

The input parameters should contain all parameters that are of interest as input to the
model. Moreover if you have any settings in model that are connected to the degrees
of freedom, like a constraint condition or a spring condition. These have to be declared
as input in your state-space system. When solving the state-space system in MATLAB,
subtract to these inputs the initial value of the corresponding DOF, as it is done in the
example Extracting State-Space Matrices.

EXTRACT STATE-SPACE MATRICES

The function mphstate requires that the input variables, output variables, and the list
of the matrices to extract in the MATLAB workspace are all defined:

str = mphstate(model, <soltag>, 'input', <input>, ...

'output', <output>, 'out', out);

where <soltag> is the solver node tag used to assemble the system matrices listed in
the cell array out, and <input> and <output> are cell arrays containing the list of the
input and output variables, respectively.

The output data str returned by mphstate is a MATLAB structure and the fields
correspond to the assembled system matrices.

The input variables need to be defined as parameters in the COMSOL model. The
output variables are defined as domain point probes or global probes in the COMSOL
model.

The system matrices that can be extracted with mphstate are listed in the table:

EXPRESSION DESCRIPTION

MA McA matrix
MB McB matrix
A A matrix

WORKING WITH MATRICES | 205

 EXPRESSION DESCRIPTION

B B matrix
C C matrix
D D matrix
Mc Mc matrix
Null Null matrix
ud ud vector
x0 x0 vector
To extract sparse matrices set the property sparse to on:

str = mphstate(model, <soltag>, 'input', <input>, ...

'output', <output>, 'out', out, 'sparse', 'on')

To keep the state-space feature node, set the property keepfeature to on:

str = mphstate(model, <soltag>, 'input', <input>, ...

'output', <output>, 'out', out, 'keepfeature', 'on')

SET LINEARIZATION POINTS

mphstate uses linearization points to assemble the state-space matrices. The default
linearization point is the current solution provided by the solver node, to which the
state-space feature node is associated. If there is no solver associated to the solver
configuration, a null solution vector is used as a linearization point.

The linearization point needs to be a steady-state solution.

You can manually select the linearization point to use. Use the initmethod property
to select a linearization point:

str = mphstate(model, <soltag>, 'input', <input>, ...

'output', <output>, 'out', out, 'initmethod', method)

where method corresponds to the type of linearization point — the initial value
expression ('init') or a solution ('sol').

To set the solution to use for the linearization point, use the property initsol:

str = mphstate(model, <soltag>, 'input', <input>, ...

'output', <output>, 'out', out, 'initsol', <initsoltag>)

where <initsoltag> is the solver tag to use for a linearization point. You can also set
the initsol property to 'zero', which corresponds to using a null solution vector as a

206 | CHAPTER 4: WORKING WITH MODELS

linearization point. The default is the current solver node where the assemble node is
associated.

For continuation, time-dependent, or eigenvalue analyses you can set which solution
number to use as a linearization point. Use the solnum property:

str = mphstate(model, <soltag>, 'input', <input>, ...

'output', <output>, 'out', out, 'solnum', <solnum>)

where <solnum> is an integer value corresponding to the solution number. The default
value is the last solution number available with the current solver configuration.

SIMULATION USING THE CONTROL SYSTEM TOOLBOX

The Control System Toolbox makes it possible to analyze state space models in
MATLAB for control design and to simulate such systems.

State space models can be defined using the ss or sparss function. The sparss
function can be used if the state space matrices are sparse. The system can be simulated
using the lsim function. In order to create a reduced order system using MATLAB we
will use the balred function. This function only accepts the use of full matrices.
Hence, the function ss is used for defining the state space system in MATLAB. Note
that calling the function ss with argument matrices that are sparse result in a set of
warnings. These warnings can be ignored.

The code below show how to use the Control System Toolbox to solve a dynamic
system from the A, B, C, and D matrices obtained using mphstate:

sys = ss(M.A, M.B, M.C, M.D);

u = repmat(<input>, length(<tspan>), 1);
[y,t] = lsim(sys, u, t);

where <input> is the input vector of the state space system, <tspan> the time step list.

One may obtain a reduced order model using the balred function.

redsys = balred(sys, <order>);

[y,t] = step(redsys, <tfinal>);

where <order> is the desired order reduction, and <tfinal> the simulation end time.

EXTRACTING STATE-SPACE MATRICES

In this section you will find an example that illustrate how to use the mphstate
function to extract the state-space matrices.

The problem studied here is a heat transfer model set with a heat source. In the
expected state space system the heat source is set as input. The temperature at specified

WORKING WITH MATRICES | 207

 location is used as output. The tutorial shows how to extract the state space matrices
and solve the system using the MATLAB functionalities.

This is the same problem solved as in Extracting Reduced Order State-Space Matrices
so you can compare the solution, and computational performance when solving the
problem with reduced order model state space system matrices.

First, load the model model_tutorial_llmatlab from the Application Library:

model = mphopen('model_tutorial_llmatlab');

This is model is used as base model for documentation tutorial. The geometry, mesh,
physics is set but for this specific problem you need to edit the model to use parameter
for the initial and external temperature as they will be used later in the state space
matrices export:

power = 30; Temp = 300; Text = 300; T0 = 293.15;

model.param.set('power', power);
model.param.set('Temp', Temp);
model.param.set('T0', T0);
model.param.set('Text', Text);
comp1 = model.component('comp1');
ht = comp1.physics('ht');
ht.feature('init1').set('Tinit', 'T0');
ht.feature('hf1').set('Text', 'Text');

You need now to add a time dependent study:

std = model.study.create('std');
time = std.feature.create('time','Transient');
time.set('tlist','range(0,1,50)');

Add a domain point probe plot, which will defines the output of the state space system:

pdom = comp1.probe.create('pdom', 'DomainPoint');

pdom.model('comp1');
pdom.set('coords3',[1e-2 1e-2 1e-2]);

Run the study and create a plot group to display the probe:

std.run;
pg1 = model.result.create('pg1', 'PlotGroup1D');
glob1 = pg1.create('glob1', 'Global');
glob1.set('expr', 'comp1.ppb1');

Extract the state-space system matrices Mc, MA, MB, C, and D, of the model with power,
Temp, and Text as input and the probe evaluation comp1.ppb1 as output:

M = mphstate(model,'sol1','out',{'Mc','MA','MB','C','D'},...
'input',{'power','Temp','Text'},...

208 | CHAPTER 4: WORKING WITH MODELS

 'output','comp1.ppb1');

Set the input power parameter and the reference temperature:

input = [power Temp-T0 Text-T0];

You can noticed that some of the input value are subtracted with the initial condition.
This is because these inputs are directly linked to the temperature degree of freedom
using a constraint condition.

func = @(t,x) M.MA*x + M.MB*input';

Set the solver option such as the mass and the Jacobian:

opt = odeset('mass',M.Mc,'jacobian',M.MA);

It is not required to define the Jacobian, but doing so speeds up the simulation.

The state space formulation that uses the mass matrix are prepared to use sparse and
hence can often be simulated much faster. In order to simulate systems that involve a
mass matrix on the left hand side one must switch to one of the simulation functions
that support a mass matrix. Such functions are ode15s, ode23s and ode23t. ode15s
and ode23t can solve problems with a singular mass matrix. Compute the state-space
system with the extracted matrices,

[t,x] = ode23s(func,0:1:50,zeros(size(M.MA,1),1),opt);
y = M.C*x';
y = y+T0;

Compare the solution computed with the system and the one computed with
COMSOL Multiphysics (see Figure 4-1):

plot(t,y,'r+');
hold on;
mphplot(model,'pg1');

WORKING WITH MATRICES | 209

 Figure 4-1: Temperature distribution computed with the state-space system (red marker)
and COMSOL Multiphysics (blue line).

Evaluate the steady-state temperature value:

G = M.D-M.C*(inv(M.MA))*M.MB;
y_inf = full(G*input');
y_inf = y_inf + T0

For some types of control system design the use of the matrices A, B, C, and D are
commonly used. Note that you need to keep the size of the matrices lower as you are
not using a mass matrix. Hence, you need to change the size of mesh in order to reduce
the number of degrees of freedom of the assembled system:

comp1.mesh('mesh1').autoMeshSize(7);

Extract the state-space system matrices A, B, C, and D, of the model with power, Temp,
and Text as input and the probe evaluation comp1.ppb1 as output:

M2 = mphstate(model,'sol1','out',{'A','B','C','D'},...
'input',{'power','Temp','Text'},...
'output','comp1.ppb1');

Compute the state-space system with the extracted matrices,

func = @(t,x) M2.A*x + M2.B*input';

[t,x] = ode45(func,0:1:50,zeros(size(M2.A,1),1));
y2 = M2.C*x';
y2 = y2+T0;

Compare the solution computed with the previous system (see Figure 4-2):

plot(t,y2,'k.')

210 | CHAPTER 4: WORKING WITH MODELS

Figure 4-2: Temperature distribution computed with the state-space system 1 (red '+'
marker), state-space system without mass matrix (black '.' marker) and COMSOL
Multiphysics (blue line).

Code for use with MATLAB ®

model = mphopen('model_tutorial_llmatlab');
power = 30; Temp = 300; Text = 300; T0 = 293.15;
model.param.set('power', power);
model.param.set('Temp', Temp);
model.param.set('T0', T0);
model.param.set('Text', Text);
comp1 = model.component('comp1');
ht = comp1.physics('ht');
ht.feature('init1').set('Tinit', 'T0');
ht.feature('hf1').set('Text', 'Text');
std = model.study.create('std');
time = std.feature.create('time','Transient');
time.set('tlist','range(0,1,50)');
pdom = comp1.probe.create('pdom', 'DomainPoint');
pdom.model('comp1');
pdom.set('coords3',[1e-2 1e-2 1e-2]);
std.run;
pg1 = model.result.create('pg1', 'PlotGroup1D');
glob1 = pg1.create('glob1', 'Global');
glob1.set('expr', 'comp1.ppb1');
M = mphstate(model,'sol1','out',{'Mc','MA','MB','C','D'},...
'input',{'power','Temp','Text'},'output','comp1.ppb1');
input = [power Temp-T0 Text-T0];
func = @(t,x) M.MA*x + M.MB*input';
opt = odeset('mass',M.Mc,'jacobian',M.MA);
[t,x] = ode23s(func,0:1:50,zeros(size(M.MA,1),1),opt);
y = M.C*x';

WORKING WITH MATRICES | 211

 y = y+T0;
plot(t,y,'r+');
hold on;
mphplot(model,'pg1');
G = M.D-M.C*(inv(M.MA))*M.MB;
y_inf = full(G*input');
y_inf = y_inf + T0
comp1.mesh('mesh1').autoMeshSize(7);
M2 = mphstate(model,'sol1','out',{'A','B','C','D'},...
'input',{'power','Temp','Text'},...
'output','comp1.ppb1');
func = @(t,x) M2.A*x + M2.B*input';
[t,x] = ode45(func,0:1:50,zeros(size(M2.A,1),1));
y2 = M2.C*x';
y2 = y2+T0;
plot(t,y2,'k.')

Extracting Reduced Order State-Space Matrices

COMSOL Multiphysics models often have a large number of degrees of freedom. This
leads to large state-space model when exported using mphstate. COMSOL
Multiphysics provides model reduction functionality, which can reduce the number of
states using an eigenvalue (or eigenfrequency) study leading to a low number of states
that can be used for simulation and analysis.

The function mphreduction returns state-space matrices or a MATLAB state-space

model using the ss function of a given reduced-order model set in the COMSOL
model.

See the Modal Reduced-Order Model and The Modal Solver Algorithm in the
COMSOL Multiphysics Reference Manual for more information.

EXTRACTING REDUCED ORDER STATE-SPACE MATRICES

In this section you will find an example that illustrate how to use the mphreduction
function to extract the reduced order model state-space matrices.

The problem studied here is a heat transfer model set with a heat source. In the
expected state space system the heat source is set as input. The temperature at specified
location is used as output. The tutorial shows how to extract the state space matrices
and solve the system using the MATLAB functionalities.

This is the same problem solved as in Extracting State-Space Matrices so you can
compare the solution, and computational performance when solving the problem with
full state space system matrices.

212 | CHAPTER 4: WORKING WITH MODELS

The following example sets up a model using commands from MATLAB. You need
the Heat Transfer Module or MEMS Module to run the model, as reduced order
modeling requires eigenvalue of the physical problem. If you don’t have access to one
of these modules, you can find the commands that sets up and solve the same problem
using the General Form PDE interface in the section Code for use with MATLAB® -
General Form PDE.

First, load the model model_tutorial_llmatlab from the Application Library:

model = mphopen('model_tutorial_llmatlab');

This is model is used as base model for documentation tutorial. The geometry, mesh,
physics is set but for this specific problem you need to edit the model to use parameter
for the initial and external temperature as they will be used later in the reduced order
model:

power = 30; Temp = 300; Text = 300; T0 = 293.15;

model.param.set('power', power);
model.param.set('Temp', Temp);
model.param.set('T0', T0);
model.param.set('Text', Text);
comp1 = model.component('comp1');
ht = comp1.physics('ht');
ht.feature('init1').set('Tinit', 'T0');
ht.feature('hf1').set('Text', 'Text');

This model is defined with a temperature constraint, you need to replace it with by a
heat flux condition as constraints are not supported as input for reduced order models:

ht.feature('temp1').active(false);
hf2 = ht.create('hf2', 'HeatFluxBoundary', 2);
hf2.selection().set(3);
hf2.set('HeatFluxType', 'ConvectiveHeatFlux');
hf2.set('h', '1e6');
hf2.set('Text', 'Temp');

You need now to add a time dependent study:

std1 = model.study.create('std1');
time = std1.feature.create('time','Transient');
time.set('tlist','range(0,1,50)');

Add a domain point probe plot, which will defines the output of the reduced order
model:

pdom = comp1.probe.create('pdom', 'DomainPoint');

pdom.model('comp1');
pdom.set('coords3',[1e-2 1e-2 1e-2]);

WORKING WITH MATRICES | 213

 Run the study and create a plot group to display the probe:

std1.run;
pg1 = model.result.create('pg1', 'PlotGroup1D');
glob1 = pg1.create('glob1', 'Global');
glob1.set('expr', 'comp1.ppb1');

A reduced order model requires eigenvalue solution of the problem. To create an

eigenfrequency study and specify the number of eigenvalue to 30 enter:

std2 = model.study().create('std2');
eig = std2.create('eig', 'Eigenfrequency');
eig.activate('ht', true);
eig.set('neigsactive', true);
eig.set('neigs', 30);
eig.set('shiftactive', true);

Before adding a reduced-order study to your model you need to set the input, here in
this example the power variable power, the bottom temperature Temp, and the exterior
temperature Text:

grmi1 =
model.common.create('grmi1','GlobalReducedModelInputs','');
grmi1.setIndex('name', 'power', 0);
grmi1.setIndex('name', 'Temp', 1);
grmi1.setIndex('name', 'Text', 2);

Now you can add a model reduction study.

std3 = model.study.create('std3');
mr = std3.create('mr','ModelReduction');
mr.set('trainingStudy','std2');
mr.set('trainingStep','eig');
mr.set('trainingRecompute','always');
mr.set('unreducedModelStudy','std1');
mr.set('unreducedModelStep','time');
mr.setIndex('qoiname','Tout',0);
mr.setIndex('qoiexpr','comp1.ppb1',0);
std3.run;

Here the model reduction study generates automatically a Time Dependent, Modal
Reduced-Order Model node you can identify with the tag rom1. To get the newly
generated reduced order model you can either type the command:

model.reduced;

or inspect the model using mphnavigator window by typing:

214 | CHAPTER 4: WORKING WITH MODELS

 mphnavigator

To visualize the reduced order node in the mphnavigator window, you

need to go the Settings menu and select Advanced.

A call to mphreduction creates the state-space matrices needed to simulate the

reduced-order system.

MR = mphreduction(model, 'rom1', ...

'out', {'MA' 'MB' 'C' 'D' 'Mc' 'x0'})

Now the reduced-order system can be simulated using the function ode23s.

input = [power Temp-T0 Text-T0];

func = @(t,x) MR.MA*x + MR.MB*input';
opt = odeset('mass',MR.Mc);
x0 = zeros(size(MR.MA,1),1);
[t,x2] = ode23s(func,0:1:50,x0,opt);
y2t = MR.C*x2';
y20 = MR.C*MR.x0;
y2 = y2t+y20;

Compare the solution computed with the system and the one computed with
COMSOL Multiphysics (see Figure 4-3):

plot(t,y2,'r+');
hold on;
mphplot(model,'pg1');

Figure 4-3: Temperature distribution computed with the reduced order model state-space
system (red marker) and COMSOL Multiphysics (blue line).

WORKING WITH MATRICES | 215

 Evaluate the steady-state temperature value:

G = -MR.C*(inv(MR.MA))*MR.MB;
y_inf = full(G*input');
y_inf = y_inf + T0

As an alternative to using ode23s, you can use functions in the Control System
Toolbox, which is an add-on to MATLAB. The matrices stored in MR can be used to
manually construct a state-space system using the function ss, or you can call
mphreduction using the return option to specify that mphreduction should do the
conversion.

sys = mphreduction(model, 'rom1', ...

'out', {'MA' 'MB' 'A' 'B' 'C' 'D' 'Mc' 'x0' 'Y0' 'Kr'}, ...
'return', 'ss')

The system can, for example, be simulated using the lsim function.

t = 0:1:50;
u = repmat(input,length(t),1);
figure(2)
[y,t] = lsim(sys, u,t);
grid on

Code for use with MATLAB ®

model = mphopen('model_tutorial_llmatlab');
power = 30; Temp = 300; Text = 300; T0 = 293.15;
model.param.set('power', power);
model.param.set('Temp', Temp);
model.param.set('T0', T0);
model.param.set('Text', Text);
comp1 = model.component('comp1');
ht = comp1.physics('ht');
ht.feature('init1').set('Tinit', 'T0');
ht.feature('hf1').set('Text', 'Text');
ht.feature('temp1').active(false);
hf2 = ht.create('hf2', 'HeatFluxBoundary', 2);
hf2.selection().set(3);
hf2.set('HeatFluxType', 'ConvectiveHeatFlux');
hf2.set('h', '1e6');
hf2.set('Text', 'Temp');
std1 = model.study.create('std1');
time = std1.feature.create('time','Transient');
time.set('tlist','range(0,1,50)');
pdom = comp1.probe.create('pdom', 'DomainPoint');
pdom.model('comp1');
pdom.set('coords3',[1e-2 1e-2 1e-2]);
std1.run;
pg1 = model.result.create('pg1', 'PlotGroup1D');
glob1 = pg1.create('glob1', 'Global');

216 | CHAPTER 4: WORKING WITH MODELS

 glob1.set('expr', 'comp1.ppb1');
std2 = model.study().create('std2');
eig = std2.create('eig', 'Eigenfrequency');
eig.activate('ht', true);
eig.set('neigsactive', true);
eig.set('neigsactive', true);
eig.set('neigs', 30);
eig.set('shiftactive', true);
grmi1 =
model.common.create('grmi1','GlobalReducedModelInputs','');
grmi1.setIndex('name', 'power', 0);
grmi1.setIndex('name', 'Temp', 1);
grmi1.setIndex('name', 'Text', 2);
std3 = model.study.create('std3');
mr = std3.create('mr','ModelReduction');
mr.set('trainingStudy','std2');
mr.set('trainingStep','eig');
mr.set('trainingRecompute','always');
mr.set('unreducedModelStudy','std1');
mr.set('unreducedModelStep','time');
mr.setIndex('qoiname','Tout',0);
mr.setIndex('qoiexpr','comp1.ppb1',0);
std3.run;
MR = mphreduction(model, 'rom1', ...
'out', {'MA' 'MB' 'C' 'D' 'Mc' 'x0'})
input = [power Temp-T0 Text-T0];
func = @(t,x) MR.MA*x + MR.MB*input';
opt = odeset('mass',MR.Mc);
x0 = zeros(size(MR.MA,1),1);
[t,x2] = ode23s(func,0:0.1:50,x0,opt);
y2t = MR.C*x2';
y20 = MR.C*MR.x0;
y2 = y2t+y20;
plot(t,y2,'r+');
hold on;
mphplot(model,'pg1');
G = -MR.C*(inv(MR.MA))*MR.MB;
y_inf = full(G*input');
y_inf = y_inf + T0

Code for use with MATLAB ® - General Form PDE

model = mphopen('model_tutorial_llmatlab');
power = 30; Temp = 300; Text = 300; T0 = 293.15;
model.param.set('power', power);
model.param.set('Temp', Temp);
model.param.set('T0', T0);
model.param.set('Text', Text);
comp1 = model.component('comp1');
comp1.physics('ht').active(false);
g = comp1.physics.create('g', 'GeneralFormPDE', {'u'});

WORKING WITH MATRICES | 217

 g.prop('Units').set('DependentVariableQuantity', 'temperature');
g.prop('Units').setIndex('CustomSourceTermUnit', 'W/m^3', 0, 0);
gfeq1 = g.feature('gfeq1');
gfeq1.setIndex('Ga', {'-material.k11*ux' '-material.k22*uy'
'-material.k33*uz'}, 0);
g.feature('init1').set('u', 'T0');
gfeq1.setIndex('f', 0, 0);
gfeq1.setIndex('da', 'material.rho*material.Cp', 0);
src1 = g.create('src1', 'SourceTerm', 3);
src1.selection.set([2]);
src1.setIndex('f', 'power/(1[cm]*1[cm]*1[mm])', 0);
flux1 = g.create('flux1', 'FluxBoundary', 2);
flux1.selection.set([7 8 10 11 12]);
flux1.setIndex('g', '10[W/(m^2*K)]*(Text-u)', 0);
flux2 = g.create('flux2', 'FluxBoundary', 2);
flux2.selection.set([3]);
flux2.setIndex('g', '1e6[W/(m^2*K)]*(Temp-u)', 0);
std1 = model.study.create('std1');
time = std1.feature.create('time','Transient');
time.set('tlist','range(0,1,50)');
pdom = comp1.probe.create('pdom', 'DomainPoint');
pdom.model('comp1');
pdom.set('coords3',[1e-2 1e-2 1e-2]);
std1.run;
pg1 = model.result.create('pg1', 'PlotGroup1D');
glob1 = pg1.create('glob1', 'Global');
glob1.set('expr', 'comp1.ppb1');
std2 = model.study().create('std2');
eig = std2.create('eig', 'Eigenfrequency');
eig.activate('ht', true);
eig.set('neigsactive', true);
eig.set('neigsactive', true);
eig.set('neigs', 30);
eig.set('shiftactive', true);
grmi1 =
model.common.create('grmi1','GlobalReducedModelInputs','');
grmi1.setIndex('name', 'power', 0);
grmi1.setIndex('name', 'Temp', 1);
grmi1.setIndex('name', 'Text', 2);
std3 = model.study.create('std3');
mr = std3.create('mr','ModelReduction');
mr.set('trainingStudy','std2');
mr.set('trainingStep','eig');
mr.set('trainingRecompute','always');
mr.set('unreducedModelStudy','std1');
mr.set('unreducedModelStep','time');
mr.setIndex('qoiname','Tout',0);
mr.setIndex('qoiexpr','comp1.ppb1',0);
std3.run;
model.reduced;

218 | CHAPTER 4: WORKING WITH MODELS

mphnavigator
MR = mphreduction(model, 'rom1', ...
'out', {'MA' 'MB' 'A' 'B' 'C' 'D' 'Mc' 'x0'})
input = [power Temp-T0 Text-T0];
func = @(t,x) MR.MA*x + MR.MB*input';
opt = odeset('mass',MR.Mc);
x0 = zeros(size(MR.MA,1),1);
[t,x2] = ode23s(func,0:1:50,x0,opt);
y2t = MR.C*x2';
y20 = MR.C*MR.x0;
y2 = y2t+y20;
plot(t,y2,'r+');
hold on;
mphplot(model,'pg1');
G = -MR.C*(inv(MR.MA))*MR.MB;
y_inf = full(G*input');
y_inf = y_inf + T0

WORKING WITH MATRICES | 219

 Extracting Solution Information and
Solution Vectors
In this section:

• Obtaining Solution Information

• Retrieving Solution Information and Solution Datasets Based on Parameter Values
• Extracting Solution Vector

Obtaining Solution Information

Get the solution object information with the function mphsolinfo. Specify only the
model object to obtain the information of the default solution object:

info = mphsolinfo(model)

This section includes information about Specifying the Solution Object and the
Output Format.

The function mphsolinfo replaces the function mphgetp. If you are

using the later you can now replace it as it will be removed in a future
version.

SPECIFYING THE SOLUTION OBJECT

To retrieve the information of a specific solution object, set the soltag property with
the solver tag soltag associated to the solution object:

info = mphsolinfo(model, 'soltag', <soltag>)

If there are several solution datasets attached to the solver, for example, solution
datasets with different selections, specify the dataset to use to get the solution object
information with the dataset property:

info = mphsolinfo(model, 'dataset', <dsettag>)

where dsettag the tag of the solution dataset to use.

220 | CHAPTER 4: WORKING WITH MODELS

OUTPUT FORMAT
The output info is a MATLAB® structure. The default fields available in the structure
are listed in the table:

FIELDS DESCRIPTION

soltag Tag of the solver associated to the solution object

study Tag of the study associated to the solution object
size Size of the solution vector
nummesh Number of mesh in the solution (for automatic remeshing)
sizes Size of solution vector and inner parameters for each mesh
soltype Solver type
solpar Parameter name
sizesolvals Length of parameter list
solvals Inner parameter value
paramsweepnames Outer parameter name
paramsweepvals Outer parameter value
label Solution node label
batch Batch information
dataset Tag of the solution dataset associated to the solution object

To get the information about the number of solutions, set the property nu to on:

info = mphsolinfo(model, 'nu', 'on')

The info structure is added with the following fields:

FIELDS DESCRIPTION

NUsol Number of solutions vectors stored

NUreacf Number of reaction forces vectors stored
NUadj Number of adjacency vectors stored
NUfsens Number of functional sensitivity vectors stored
NUsens Number of forward sensitivity vectors stored

The batch field is a structure including the following fields:

BATCH FIELDS DESCRIPTION

type The type of batch

psol Tag of the associated solver node

EXTRACTING SOLUTION INFORMATION AND SOLUTION VECTORS | 221

 BATCH FIELDS DESCRIPTION

sol Tag of the stored solution associated to psol

seq Tag of the solver sequence associated to psol

Retrieving Solution Information and Solution Datasets Based on

Parameter Values
A model can contain several solution vectors computed with different values of
parameters, such as time, eigenvalue, or model parameters. These solution vectors can
be available in different solution datasets. Use the function mphsolutioninfo to
retrieve the solution vector corresponding to a specified study parameter value.

The parameters used in a study can be group in two distinct solution number types:

• The inner solution, containing the solution computed with parameters such as
eigenvalues, time steps, or continuation parameter combinations.
• The outer solution, containing the solution computed with parameters defined in
parametric sweep.

To get information about all solution object and solution dataset combinations in the
model enter the command:

info = mphsolutioninfo(model)

The output info is a structure containing these fields:

FIELDS DESCRIPTION

solutions List of the solution object tags available in the model

sol# Substructure containing information related to the solution
number #

The substructure info.sol# has these fields:

FIELDS DESCRIPTION

dataset List of the tags of the dataset associated to the solution

study Tag of the study that computed the solution
sequencetype Type of solution node
cellmap Connections between parameters and inner/outer solution
numbers; the field is not available by default.
values Values of the parameters used in the solution
parameters Names of the parameters used in the solution

222 | CHAPTER 4: WORKING WITH MODELS

FIELDS DESCRIPTION

mapheaders Headers of the table stored in the map field

map Connections between the parameter values and the solution
number (inner and outer solutions)
You can also retrieve the solution objects and solution datasets related to a specific
parameter value with the command:

info = mphsolutioninfo(model,'parameters',{'e1','v1','tol1'})

where e1 is the expression name, v1 the value of the expression.

The property parameters can also be set as a 1xN cell array, where N corresponds to
the number of parameters to specify.

This section includes information about Specifying the Solution Object and the
Output Format. It also includes the section, Retrieving Solution Information.

SPECIFYING THE SOLUTION OBJECT

To retrieve the information of a specific solution object, set the soltag property with
the solver tag soltag associated to the solution object:

info = mphsolutioninfo(model, 'soltag', <soltag>)

If there are several solution datasets attached to the solver, for example, solution
datasets with different selections, specify the dataset to use to get the solution object
information with the dataset property:

info = mphsolutioninfo(model, 'dataset', <dsettag>)

where dsettag the tag of the solution dataset to use.

OUTPUT FORMAT
To include the cellmap field in the info.sol# substructure set the property cellmap
to on:

info = mphsolutioninfo(model, 'cellmap', 'on')

Improve the visibility of the map table by sorting the row using either the column
number or the name in the map header:

info = mphsolutioninfo(model, 'sort', <idx>)

where <idx> is a positive integer equal to the column number or a string

corresponding to the name of the column header.

EXTRACTING SOLUTION INFORMATION AND SOLUTION VECTORS | 223

 RETRIEVING SOLUTION INFORMATION
This example shows how to use the function mphsolutioninfo to retrieve solution
information in a mode combining a parametric sweep and transient analysis.

Start by loading the base model model_tutorial_llmatlab from the COMSOL

Multiphysics Application Libraries; this model contains base settings for a thermal
analysis:

model = mphopen('model_tutorial_llmatlab');

Now create a study combining a parametric sweep and a transient study step. The
parametric sweep consist by varying the parameters that set the heat source and the
bottom temperature. This is done with these commands:

std = model.study.create('std');
param = std.feature.create('param', 'Parametric');
param.setIndex('pname', 'power', 0);
param.setIndex('plistarr', '30 60 90',0);
param.setIndex('pname', 'Temp', 1);
param.setIndex('plistarr', '300 320', 1);
time = std.feature.create('time', 'Transient');
time.set('tlist', 'range(0,1,25)');

Set the sweep type to generate all possible combinations of the parameters power and
tf and compute the study:

param.set('sweeptype', 'filled');
std.run;

Once the solution is computed (it takes about 90 seconds), you can retrieve the
solution information in the model:

info = mphsolutioninfo(model)

The output info is a structure containing nine fields. By navigating in the info
structure you can retrieve how the solutions are stored in the model.

• info.sol1 contains the solution information related to the solver sequence sol1.
The associated dataset is dset1.
• info.sol2 contains the solution information for the parametric sequence. This
regroups the solution vectors computed for all outer parameters.

The other substructures contain the solution information for all possible outer solution
combinations.

Get the relation between the parameter values and the inner and outer solution
numbers:

224 | CHAPTER 4: WORKING WITH MODELS

 map = info.sol2.map

Retrieve the solution information related to the parameters power = 60 W:

info = mphsolutioninfo(model, 'parameters', {'power',60,0})

Retrieve the solution information related to the parameters power = 60 W,

Temp = 300 K and t = 10.4 seconds, for the time use a tolerance of 0.5 seconds to find
the appropriate inner solution number:

info = mphsolutioninfo(model, 'parameters', {{'power',60,0},...

{'Temp',300,0},{'t',10.4,0.5}})

To get the list of the solutions that contain the given parameters enter:

solnum = info.solutions

Code for use with MATLAB ®

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
param = std.feature.create('param', 'Parametric');
param.setIndex('pname', 'power', 0);
param.setIndex('plistarr', '30 60 90',0);
param.setIndex('pname', 'Temp', 1);
param.setIndex('plistarr', '300 320', 1);
time = std.feature.create('time', 'Transient');
time.set('tlist', 'range(0,1,25)');
param.set('sweeptype', 'filled');
std.run;
info = mphsolutioninfo(model)
map = info.sol2.map
info = mphsolutioninfo(model, 'parameters', {'power',60,0})
info = mphsolutioninfo(model, 'parameters', {{'power',60,0},...
{'Temp',300,0},{'t',10.4,0.5}})
solnum = info.solutions

Extracting Solution Vector

To extract the solution vector with the function mphgetu, enter:

U = mphgetu(model)

where U is an Nx1 double array, where N is the number of degrees of freedom of the
COMSOL Multiphysics model.

EXTRACTING SOLUTION INFORMATION AND SOLUTION VECTORS | 225

 This section includes information about Specifying the Solution and the Output
Format.

You can refer to the function mphxmeshinfo to receive the DOF name or
the node coordinates in the solution vector, see Retrieving Xmesh
Information.

SPECIFYING THE SOLUTION

Change the solver node to extract the solution vector with the property solname:

U = mphgetu(model, 'soltag', <soltag>)

where <soltag> is the tag of the solver node.

For solver settings that compute for several inner solutions, select the inner solution to
use with the solnum property:

U = mphgetu(model, 'solnum', <solnum>)

where <solnum> a positive integer vector that corresponds to the solution number to
use to extract the solution vector. For time-dependent and continuation analyses, the
default value for the solnum property is the last solution number. For an eigenvalue
analysis, it is the first solution number.

A model can contain different types of solution vectors — the solution of the problem,
the reaction forces vector, the adjoint solution vector, the functional sensitivity vector,
or the forward sensitivity. In mphgetu, you can specify the type of solution vector to
extract with the type property:

U = mphgetu(model, 'type', type)

where type is one of these strings 'sol', 'reacf', 'adj', or 'sens' used to extract the
solution vector, the reaction forces, the functional sensitivity, or the forward sensitivity,
respectively.

OUTPUT FORMAT
mphgetu returns the default the solution vector. Get the time derivative of the solution
vector Udot by adding a second output variable:

[U, Udot] = mphgetu(model)

In case the property solnum is set as a 1xM array and the solver node only uses one
mesh to create the solution, the default output is an NxM array, where N is the number
of degrees of freedom of the model. Otherwise, the output U is a cell array that contains

226 | CHAPTER 4: WORKING WITH MODELS

each solution vector. If you prefer to have the output in a cell array format, set the
property matrix to off:

U = mphgetu(model, 'solnum', <solnum>, 'matrix', 'off')

EXTRACTING SOLUTION INFORMATION AND SOLUTION VECTORS | 227

 Retrieving Xmesh Information
Use LiveLink™ for MATLAB® to retrieve low level information of the COMSOL
Multiphysics finite element model.

In this section:

• The Extended Mesh (Xmesh)

• Extracting Xmesh Information

The Extended Mesh (Xmesh)

The extended mesh (xmesh) is the finite element mesh used to compute the solution.
This contains the information about elements, nodes, and degrees of freedom such as
DOF names, position of the nodes in the assembled matrix system, or how elements
and nodes are connected.

Extracting Xmesh Information

The function mphxmeshinfo returns the extended mesh information. To get the
xmesh information of the current solver and mesh node, enter the command:

info = mphxmeshinfo(model)

where info is a MATLAB structure that contains the fields in the table:

FIELDS DESCRIPTION

soltag Tag of the solver node

ndofs Number of degrees of freedom
fieldnames List of field variables names
fieldndofs Number of degrees of freedom for each field variable
meshtypes List of the mesh type
geoms Tag of the geometry node used in the model
dofs Structure containing the dofs information
nodes Structure containing the nodes information
elements Structure containing the elements information

228 | CHAPTER 4: WORKING WITH MODELS

The dofs substructure contains the fields listed in the table:

FIELDS DESCRIPTION

geomnums 1-based geometry numbers for all DOFs.

coords Global coordinates for all DOFs in the model length unit. The kth
column of this matrix contains the coordinates of DOF number k.
nodes 0-based node numbers for all DOFs.
dofnames DOF names.
nameinds 0-based indices into dofNames() for all DOFs.
solvectorinds 0-based indices into solution vector for all DOFs.

The nodes substructure contains the fields listed in the table:

FIELDS DESCRIPTION

coords Global coordinates for all nodes. The nth column contains the
coordinates of node point number n.
dofnames DOF names in this geometry.
dofs 0-based DOF numbers for all nodes in this geometry.
dofs()[k][n] is the DOF number for DOF name
dofNames()[k] at node point n. A value of -1 means that there
is no DOF with this name at the node. Note: If there is a slit, only
one of the DOFs is given for each node point.

The elements substructure contains the fields listed in the table:

FIELDS DESCRIPTION

meshtypes List of the type of mesh available.

type Substructure containing the information of element of type type.

The type substructure lists the information for each element. The possible mesh types
are vtx, edg, quad, tri, quad, tet, hex, prism, and pyr. The substructure type
contains the fields listed in the table:

FIELDS DESCRIPTION

localcoords Local coordinates of nodes. The kth column contains the

coordinates of local node point number k.
localdofcoords The local coordinates for each local DOF (one column for
each local DOF).
localdofnames The name for each local DOF.

RETRIEVING XMESH INFORMATION | 229

 FIELDS DESCRIPTION

nodes 0-based node point indices for all mesh elements of type type.
A value -1 means that there is no node point at this location.
dofs 0-based DOF numbers for all mesh elements of type type. A
value -1 means that there is no DOF at this location.

SPECIFY THE INFORMATION TO RETRIEVE

To specify the solver node to retrieve the xmesh information, set the property soltag
as in this command:

info = mphxmeshinfo(model, 'soltag', <soltag>)

where <soltag> is the tag of the solver used to extract the xmesh information.

To retrieve the xmesh information for a specific study step node, specify it with the
property studysteptag:

info = mphxmeshinfo(model, 'studysteptag', <studysteptag>)

where <studysteptag> is the tag of either a compiled equation node or a variable

node.

In case several mesh cases have been used by a specific solver, for example, with an
automatic remeshing procedure, you can specify which mesh case to use to get the
discretization information:

info = mphxmeshinfo(model, 'meshcase', <meshcase>)

where <meshcase> is the mesh case number or the tag of the mesh case.

230 | CHAPTER 4: WORKING WITH MODELS

Navigating the Model
The model object contains all the finite element model settings. To retrieve the model
information you can navigate in the model object using a graphical user interface or
directly at the MATLAB® prompt. Learn how to get the list of predefined expressions
available for a given model and how to extract the value of these expressions and also
the properties of the method used in the model.

In this section:

• Navigating the Model Object Using a GUI

• Navigating The Model Object At The Command Line
• Finding Model Expressions
• Evaluating the Model Parameters
• Getting Feature Model Properties
• Getting Parameter and Variable Definitions
• Getting Selection Information

Navigating the Model Object Using a GUI

The usual approach to navigate through the model object in a graphical user interface
(GUI) is to load the model object at the COMSOL Desktop. Then transfer the model
object from the COMSOL Multiphysics Server to the COMSOL Desktop as in
Sharing the Model Between the COMSOL Desktop® and the MATLAB® Prompt.

An alternative approach is to call the function mphnavigator that displays the model
object information in a MATLAB® GUI. To run the function at the MATLAB prompt
enter the command:

mphnavigator

NAVIGATING THE MODEL | 231

 This command pops-up a MATLAB GUI as in this figure:

If you have installed the COMSOL apps in the MATLAB Apps ribbon, click the
COMSOL Model Navigator icon ( ).

Using the COMSOL Model Navigator apps ( ) only model objects

with the name model are supported.

If the COMSOL model objected is not stored in the MATLAB variable model enter
the command:

mphnavigator(<modelvar>)

where <modelvar> is the variable name in MATLAB that contains the model.

If a new model object is created with the same MATLAB object name,
restart mphnavigator in order to have the updated model information.

232 | CHAPTER 4: WORKING WITH MODELS

THE MENU BAR ITEMS

The mphnavigator GUI menu bar has the following options:

• The File menu, where the current model object can be saved in the MPH-format, a
new model object can be opened, and the mphnavigator window can be closed.
• The Tools menu lists the navigation tools available for the model object. Search is a
shortcut to the command mphsearch that starts a GUI to search expressions or tags
in the model object (see Finding Model Expressions). Solutions starts a GUI to
display the solution object available in the COMSOL Multiphysics model object.
Warnings and Errors lists the error or warning nodes available in the model object
(see Handling Errors and Warnings). Click Copy code: loop if you need to run a
specific node in loop, then you can paste the commands in a code editor, you will
get the following commands:
tag1 = <feattag1>;
tag2 = <feattag2>;
tags = cell(model.geom(tag1).feature(tag2).feature.tags);
for i=1:length(tags)
obj = model.geom(tag1).feature(tag2).feature(tags{i});
end

where <feattag1> and <feattag2> are the feature tags to reach the selected feature
in the model.

• The Settings menu only contains the Advanced options. Select or deselect the
advanced model object methods that are displayed in the Model Viewer tree.
• The Help menu.

THE SHORTCUT ICON

Just under the menu bar are several shortcut buttons.

NAVIGATING THE MODEL | 233

 These buttons are unavailable if no method has been selected in the Model Tree section.

• The Update properties and methods button ( ) update the Properties and Methods
section base on the change operated to the COMSOL model.
• The Open plot window button ( ) generate the plot in COMSOL of the selected
node and transfer the image in a MATLAB figure. This supports other plot than
geometry, mesh or plot group such as functions.
• The Plot button ( ) displays the geometry, the mesh, or a plot group in a
MATLAB figure.
• The Help button ( ) shows context-sensitive help for the currently selected node
in the model tree.
• The Help for COMSOL API button ( ) open the COMSOL API Documentation
corresponding to the selected node.
• The Report button ( ) generate report node in the model, you can choose
between brief, intermediate or complete report.
• The Write report button ( ), write the report in a document.

THE MODEL TREE SECTION

234 | CHAPTER 4: WORKING WITH MODELS

The Model Tree section has the list of the nodes of the model object. Use the scroll bar
to the right to scroll down the list and click the + icon to expand the model object
feature nodes.

When a feature node is selected, its associated command is listed just beneath the
model tree. Click Copy to copy syntax to the clipboard and then paste it in your script.

The Model Tree list is slightly different to the Model Builder list available in
the COMSOL Desktop. This is because mphnavigator displays all
feature nodes and does not use the same filter as in the COMSOL
Desktop to order the available feature nodes.

THE PROPERTIES SECTION

NAVIGATING THE MODEL | 235

 The Properties and the Value sections list the properties of a selected feature node and
the associated values respectively. The Allowed Value section list the allowed value for
the corresponding property.

Not all feature node returns a list of allowed value for the properties.

Click Copy set to copy to the clipboard the command that sets the selected property to
its current value.

Click Copy get to copy to the clipboard the command to get the value of the currently
selected property.

Click Copy Table to copy the entire properties table to the clipboard, then paste into a
text or spreadsheet editor.

Click Copy to copy a selected cell in the properties table.

THE METHODS SECTION

236 | CHAPTER 4: WORKING WITH MODELS

The Methods section lists all the methods associated to the feature node selected in the
Model Tree section.

Click Filter to filter the reduce the methods list to the one that returns simple
information that is easy and fast to evaluate.

Select a method in the list to get its associated syntax at the button of the Methods
section. Use the Copy button to copy the syntax to the clipboard. Click Copy call to
copy the method syntax associated to the selected feature node, the syntax is ready to
use at the MATLAB prompt.

Navigating The Model Object At The Command Line

Use the command mphmodel at the MATLAB® prompt to retrieve model object
information, such as tags for nodes and subnodes of a COMSOL Multiphysics model
object.

To get the list of the main feature nodes and the tags of the model object model, enter
the command:

mphmodel(model)

To list the subfeature of the node type model.feature enter the command:

mphmodel(model.feature)

To list the subfeature node of the feature node model.feature(<ftag>), enter:

mphmodel(model.feature(<ftag>))

Use the flag -struct to return the model object information to MATLAB structure:

str = mphmodel(model,'-struct')

str is a MATLAB structure and the fields consist of each feature node associated to the
root node.

Finding Model Expressions

Each model object contains predefined expressions that depend on the physics
interface used in the model.

The function mphsearch starts a MATLAB® GUI that displays the list of all the
expressions, constants, solution variables, or parameters available in the model object.
To run the function type at the MATLAB prompt:

mphsearch

NAVIGATING THE MODEL | 237

 This command pops-up a MATLAB GUI as in this figure:

If you have installed the COMSOL apps in the MATLAB Apps ribbon, click the
COMSOL Search icon ( ).

Using the COMSOL Model Search apps( ) only model object with
name model are supported.

If the COMSOL model objected is linked with a MATLAB object with a different
name than model, enter the command:

mphsearch(modelname)

where modelname is the name of the MATLAB object.

The table has the following for each entry, the:

• Name of the expression,

• Expression as it is set in the property value,
• Description if there is one set for the expression,
• Type of the expression, and the
• Path in the model object.

The Search section has a search tool to filter the list. Enter any string in the text field
and select where to search the string — in the name, the expression, or the description
of the table entry. You can also select the type you want to list. The expression type can

238 | CHAPTER 4: WORKING WITH MODELS

be Equation, Field, Function, Geom, Mesh, Tag, VarNames, or Weak. You can select
Starts with to search for any expression that start with the string enter in the text field.

• Click Go to display the result of the search. Click Clear to clear the search settings.
• Click Model info to get the model description in a separate window.
• Click Copy to copy any entry of the table to the clipboard.
• Click Close to close the mphsearch window.

Evaluating the Model Parameters

Use the command mphevaluate to evaluate the expression defined in the Parameters
node as in the command below:

value = mphevaluate(model, <expr>)

The evaluation does not require an existing solution dataset in the model, which means
you can evaluate the expression even if there is no solution computed in the model.

To evaluate multiple expressions at once, define <expr> as a cell array of expressions

as in the command below:

str = mphevaluate(model, {<expr1>, <expr2>, ...})

The output str is an array of structs with the same size as for the cell array of
expressions. For multiple parameters evaluation only a single output is permitted. The
struct contains the following fields: name, the parameter name; value, its value in the
unit specified in the field unit; def, the string for the definition of the parameter; and
descr the string description of the parameter.

To evaluate all the parameters defined in the table enter:

str = mphevaluate(model)

The output str is an array of structs with the same size as for the parameters table.

Get the full information of a model parameter expression with the command:

[value, unit, def, descr] = mphevaluate(model, <expr>)

where value, unit, def, and descr are the parameter value, the parameter unit, the
parameter definition and the parameter description respectively.

You can specify the unit of the expression to evaluate with the command:

value = mphevaluate(model, <expr>, <unit> )

where <unit> is the unit to evaluate the expression <expr> in.

NAVIGATING THE MODEL | 239

 To evaluate and return only the value of the expression use the command:

value = mphevaluate(model, <expr>, <unit>, 'value')

To evaluate and return only the unit of the expression use the command:

value = mphevaluate(model, <expr>, <unit>, 'unit')

To evaluate and return the definition string of the expression use the command:

value = mphevaluate(model, <expr>, <unit>, 'valueunit')

The above command can be used to set a new parameter using an existing one.

Getting Feature Model Properties

Use the command mphgetproperties to extract at the MATLAB® prompt the
properties of a specified node of the model object. Not all nodes contains properties.
Use the command:

str = mphgetproperties(model.<feature>)

where str is a MATLAB structure that lists all the properties and the value of the
feature node <feature>.

Some model node provides also a list of allowed value for their properties, to get such
a list enter the command:

[str, allowed] = mphgetproperties(model.<feature>)

this also returns the MATLAB structure allowed containing the list of the allowed
values for the properties of the feature node <feature>.

If you want to filter the properties in the output use the property propnames as in the
command below:

str = mphgetproperties(model.<feature>,'propnames',{<prop>,...})

where <prop> is the property name to show only in the output structure.

Set the property returnstrings to off to get the property data not as string as in the
command below:

str = mphgetproperties(model.<feature>,'returnstrings','off',...)

You can get the selection of the properties with the property showsel set to on as
described below:

str = mphgetproperties(model.<feature>,'showsel','on',...)

240 | CHAPTER 4: WORKING WITH MODELS

Getting Parameter and Variable Definitions
Use the command mphgetexpressions to get the expressions and the descriptions of
parameters and variables. Use the command:

expr = mphgetexpressions(<node>)

where <node> is the node to get the expressions from. Nodes that can be used are
model.param, model.variable(<tag>), and model.result.param.

expr is an Nx3 cell array where N is the number of expressions for this node.

Getting Selection Information

Use the function mphgetselection to retrieve the model selection information:

str = mphgetselection(model.selection(<seltag>))

where <seltag> is the tag a selection node defined in the model object. The output
®
str is a MATLAB structure with the following fields:

• dimension, the space dimension of the geometric entity selected,

• geom, the tag of the geometry node used in the selection,
• entities, the list of the entity indexes listed in the selection, and
• isGlobal, Boolean value to indicate if the selection is global or not.

NAVIGATING THE MODEL | 241

 Handling Errors and Warnings
In this section:

• Errors and Warnings

• Using MATLAB® Tools to Handle COMSOL® Exceptions
• Displaying Warnings and Errors in the Model

Errors and Warnings

COMSOL Multiphysics reports these types of problems:

• Errors, which prevents the program from completing a task, and

• Warnings, which are problems that do not prevent the completion of a task but that
might affect the accuracy or other aspects of the model.

For both errors and warnings a message is stored in a separate node located just below
the problematic model feature node.

In case of errors, a Java® Exception is thrown to MATLAB®, which also breaks the
execution of the script.

Using MATLAB® Tools to Handle COMSOL® Exceptions

When running a model that returns an error in MATLAB®, the execution of the script
is automatically stopped. You can use MATLAB tools to handle exceptions and prevent
the script from breaking. Use the try and catch MATLAB statements to offer
alternatives to a failed model.

In a loop, for example, use the try and catch statements to continue to the next
iteration. For automatic geometry or mesh generation you can use it to set the model
properties with alternative values that circumvent the problem.

Displaying Warnings and Errors in the Model

Use the command mphshowerrors to search in a given model object for warning or
error nodes. To display the error and warning messages and their location in the model
object enter the command:

mphshowerrors(model)

242 | CHAPTER 4: WORKING WITH MODELS

Alternatively mphshowerrors can also return the error and warning information in a
MATLAB® variable:

str = mphshowerrors(model)

where str is an Nx2 cell array, where N is the number of error and warning nodes that
contain the model object. str{i,1}, which contains the location in the model of the
i:th error or warning message, str{i,2} contains the message of the ith error or
warning message, and str{i,3} contains a cell arrays of the model tree nodes that
contain the error information. This last information makes it easier to extract error
information for automated processing of error and warning conditions.

HANDLING ERRORS AND WARNINGS | 243

 Improving Performance for Large
Models
Memory management is key to successful modeling. In COMSOL Multiphysics the
finite element model can store a large amount of data depending on the complexity of
the model. Exchanging such a large amount of data between MATLAB® and the
COMSOL server can be problematic in terms of memory management or execution
time. This section discusses the model settings if you are experiencing memory
problems or slowness of command execution.

• Allocating Memory
• Disabling Model Feature Update
• Disabling The Model History

Allocating Memory
COMSOL Multiphysics stores the data in Java®. If you are experiencing memory
problems during meshing, postprocessing operations, or when exchanging data
between the COMSOL server and MATLAB®, this can mean that the Java heap size
is set with too low a value.

Increasing the memory allocated for the Java process necessarily decreases
the memory available for the solver.

Either set The COMSOL Multiphysics Server Java Heap Size or Setting Manually the
Memory in MATLAB.

THE COMSOL MULTIPHYSICS SERVER JAVA HEAP SIZE

The Java heap size settings for the COMSOL Multiphysics Server process are stored
in the comsolmphserver.ini file. You can find this file in the
COMSOL55/Multiphysics/bin/<arch> directory. <arch> correspond to the
architecture of the machine where the COMSOL Multiphysics Server is running
(win64, maci64, or glnxa64).

Edit the file with a text editor, the Java heap settings are defined as in the following
lines:

244 | CHAPTER 4: WORKING WITH MODELS

 -Xss4m
-Xms40m
-Xmx1024m
-XX:MaxPermSize=256m

The values are given in Mb, modify these value to satisfy the model requirements.

SETTING MANUALLY THE MEMORY IN MATLAB

To modify the Java heap size you need to edit the java.opts file available under the
COMSOL with MATLAB start-up directory. The java.opts file is stored by default
with the following settings:

-Xss4m
-Xmx768m
-XX:MaxPermSize=256m

The values are given in Mb, modify these value to satisfy the model requirements.

To modify the MATLAB Java Heap size the java.opts file has to be stored at the
MATLAB start-up directory. This is the case when starting COMSOL with MATLAB.

If you are manually connecting MATLAB with a COMSOL Multiphysics

Server, make sure you have the java.opts at the MATLAB start-up
directory.

Disabling Model Feature Update

Every time a setting in changed in a model COMSOL Multiphysics automatically
checks the settings for that particular feature and updates any other feature that may
depend on the new setting. This ensures you that the features are built with updated
expressions and that any error messages appear as soon as possible after a COMSOL
command has been executed.

For models that contain a large amount of physics feature nodes this update operation
can take some time. For small models this is not an issue, but for larger models the
checks can be time consuming. It can help to deactivate the model feature update.

To disable the feature model update enter the command:

model.disableUpdates(true)

You have to enable the feature update again prior to computing the solution in order
to make sure that COMSOL works on an updated model definition. Enabling the

IMPROVING PERFORMANCE FOR LARGE MODELS | 245

 feature update is also necessary before building the geometry or the mesh in case these
are defined using expressions.

To enable the feature model update, enter the command:

model.disableUpdates(false)

Disabling The Model History

If you run a model in a loop you can experience a slowdown when the number of
iterations increases. This happens only with a large amount of iterations. The
increasing memory requirements for storing the model history explains this slowdown.
You can see all the operations performed on the model when saving it as an M-file. If
you run a model in a loop you do not need to store the model history because it
contains the same operations as many times as you have iterations in the loop. The
solution is to disable the history recording. To do this, enter the command:

model.hist.disable

When the model history is disabled you no longer see the commands used to set up
the model when saving it as an M-file.

The functions mphload and mphopen automatically disables the model

history when loading a model.

To activate the model history, enter the command:

model.hist.enable

246 | CHAPTER 4: WORKING WITH MODELS

Creating a Custom GUI
You can use the MATLAB® Guide or App Designer functionality to create a GUI and
connect the interface to a COMSOL Multiphysics model object. Each operation in the
GUI sets the value of a MATLAB variable or calls a MATLAB command. You can call
commands at the MATLAB prompt to set up a COMSOL model object or set
MATLAB variables in the COMSOL model object.

The figure below illustrates a GUI made in MATLAB and linked to a COMSOL model
object.

The simplified GUI only allows the user to compute a heat transfer problem on a given
geometry. The user can only change the radius and the position of the bottom circle
geometry. The heat source applied to the bottom circle is also defined by the user.

The button executes the building operation of the geometry and mesh. Another
button executes the computation of the solution.

CREATING A CUSTOM GUI | 247

248 | CHAPTER 4: WORKING WITH MODELS
 5

Calling External Functions

This section introduces you to the MATLAB® function callback from the
COMSOL Desktop® and COMSOL Multiphysics® model object.

In this chapter:

• Running External Function

• The MATLAB® Function Feature Node

249
 Running External Function
When running the model containing a MATLAB function feature node, COMSOL
Multiphysics automatically starts a MATLAB process that evaluates the function and
returns the value to the COMSOL model.

You do not need to start COMSOL with MATLAB to call a MATLAB

function from within the model; starting the COMSOL Desktop is
sufficient. The MATLAB process starts automatically to evaluate the
function.

On Linux operating systems, specify the MATLAB root directory path

MLROOT and load the gcc library when starting the COMSOL Desktop:
comsol -mlroot MLROOT -forcegcc.

Allowing External MATLAB Functions

To run MATLAB functions you need to allow external processes in the security
preferences.

In the COMSOL Desktop go to the Preferences and select Security, locate the General
section. To enable permanently external MATLAB function, in the list Allow external
MATLAB® functions select Yes. To enable external MATLAB function only once set the
list Allow external MATLAB® functions to Ask. Using the later option, you will be asked
to enable external functions when the function is run.

In the COMSOL Server™, you need to connect as administrator then go to

Administration>Preferences, in the Security section select Allow external MATLAB®
functions. Then click Save.

ALLOWING EXTERNAL PROCESSES IN THE COMMAND LINE

Add the flag -allowexternalmatlab on to the COMSOL start-up command to
enable external processes.

250 | CHAPTER 5: CALLING EXTERNAL FUNCTIONS

Disabling MATLAB® Splash Screen at Start-Up
To disable the MATLAB splash screen that pops-up when the MATLAB engine is
started you need to create the environment variable COMSOL_MATLAB_INIT before
starting COMSOL and set this variable with the value "matlab -nosplash".

Running a MATLAB® Function in Applications

To run an application from the COMSOL Server™ that uses an external MATLAB
function, it is recommended to embed the function M-file in the application. The
Application Builder offers the possibility to upload file that can then be accessible on
the server. Under the Libraries node, select Files, and in the Files node’s Settings
window, click the Add File to Library button ( ) to add the M-file to the library..

Applications that use external MATLAB function are not supported using
the COMSOL Client. It is only possible to use such applications using a
browser.

After embedding an M-file used by an existing MATLAB function, the

embedded M-file will only be used for new solutions. So it is
recommended to update any solutions that are saved in the application, if
they depend on the embedded M-file.

RUNNING EXTERNAL FUNCTION | 251

 The MATLAB® Function Feature Node
MATLAB® functions are global in scope and you can use them in a model to define
model settings such as

• Parameters
• Geometry settings
• Mesh settings
• Material properties
• Physics settings (domain conditions, boundary conditions, etc.)

Material properties and physics settings are evaluated while the model is solved
whereas the features can be used while the model is constructed.

• Defining a MATLAB® Function in the COMSOL® Model

• Setting the Function Directory Path in MATLAB®
• Adding a MATLAB® Function with the COMSOL® API Syntax
• Function Input/Output Considerations
• Updating Functions
• Defining Function Derivatives

Defining a MATLAB® Function in the COMSOL® Model

These topics are described for the MATLAB® function:

• Adding the MATLAB Function Node

• Defining the MATLAB Function
• Plotting the Function
• Example: Define the Hankel Function

ADDING THE MATLAB FUNCTION NODE

To evaluate a MATLAB function from in the COMSOL Multiphysics model you need
to add a MATLAB node in the model object where the function name, the list of the
arguments, and, if required, the function derivatives, are defined.

252 | CHAPTER 5: CALLING EXTERNAL FUNCTIONS

To add a MATLAB function node, on the Home toolbar, click Functions and select
Global>MATLAB( ).

The Settings window of the MATLAB node has these sections:

• Functions, where you declare the name of the MATLAB functions and their
arguments.
• Derivatives, where you define the derivative of the MATLAB functions with respect
to all function arguments.
• Plot Parameters, where you can define the limit of the arguments value in order to
display the function in the COMSOL Desktop Graphics window.

THE MATLAB® FUNCTION FEATURE NODE | 253

 DEFINING THE MATLAB FUNCTION
This figure illustrates the Settings window for MATLAB:

Under Functions, you define the function name and the list of the function arguments.

In the table columns and rows, enter the Function name and the associated function
Arguments. The table supports multiple function definitions. You can define several
functions in the same table or add several MATLAB nodes, as you prefer.

About Requirements for Functions and How to Test Them

Any function that you want to call using a MATLAB node must fulfill the following
requirements:

• It can take any number of inputs as vectors and must return a single output vector.
• The input vectors can be of arbitrary size, but in a single call the inputs vectors will
all have the same length. The returned vector must have exactly the same length as
the input vectors.

For example, functions such as + and - (plus and minus) work well on vector inputs,
but matrix multiplication (*, mtimes) and matrix power (^, mpower) do not. Instead,
use the elementwise array operators .* and .^. See also Function Input/Output
Considerations.

254 | CHAPTER 5: CALLING EXTERNAL FUNCTIONS

It is good practice to test your own functions and any MALTLAB functions that you
want to call from COMSOL Multiphysics by running them on the MATLAB
command line using vectors with suitable values as inputs. For example, using
besselj, a Bessel function of the first kind:

input = (1:10)'
size(input)
out = besselj(input, input)
size(out)

Here there are no errors, and the size of the input and output is the same.

As another example, test the corrcoef function for computing correlation

coefficients:

input = (1:10)'
size(input)
out = corrcoef(input, input)
size(out)

There are no errors when calling corrcoef using vector inputs, but the result does not
have the same size as the input and hence a call to corrcoef in this way will not work.

PLOTTING THE FUNCTION

Click the Plot button ( ) to display a plot of the function.

Click the Create Plot button ( ) to create a plot group under the Results node.

To plot the function you first need to define limits for the arguments. Expand the Plot
Parameters section and enter the desired value in the Lower limit and Upper limit
columns. In the Plot Parameters table the number of rows correspond to the number
of input arguments of the function. The first input argument corresponds to the top
row.

In case there are several functions declared in the Functions table, only the function that
has the same number of input arguments as the number of filled in rows in the Plot
Parameters table is plotted.

If several functions have the same number of input arguments, the first function in the
table (from top to bottom) is plotted. Use the Move up ( ) and Move down ( )
buttons to change the order of functions in the table.

THE MATLAB® FUNCTION FEATURE NODE | 255

 EXAMPLE: DEFINE THE HANKEL FUNCTION
Assume that you want to use MATLAB’s Bessel function of the third kind (Hankel
function) in a COMSOL model. Add a MATLAB function node, then define the
following settings:

FUNCTION NAME ARGUMENTS

besselh nu, x

To plot the function you need first to define the lower and upper limits for both nu
and x. In the Plot Parameters table set the first row (which corresponds to the first
argument nu) of the Lower limit column to 0 and the Upper limit column to 5 and set
the second row (corresponding of x) of the Lower limit column to 0 and the Upper limit
column to 10:

256 | CHAPTER 5: CALLING EXTERNAL FUNCTIONS

Click the Plot button ( ) to get this plot:

Setting the Function Directory Path in MATLAB®

To be able to run a model that use an external MATLAB® function, the path directory
of the function has to be set in MATLAB before it is called by COMSOL Multiphysics
to evaluate the function.

To proceed you have these options to set the directory path in MATLAB:

• the model MPH-file in the same directory as for the M-functions;

• Set the system environment variable COMSOL_MATLAB_PATH with the
M-functions directory path; or
• Use the Set Path window to specify the MATLAB search path. To open the window
type pathtool at the MATLAB prompt or in the MATLAB desktop go the Home
toolbar, Environment group.

Adding a MATLAB® Function with the COMSOL® API Syntax

To add a MATLAB® feature node to the COMSOL Multiphysics model using the
COMSOL API, enter the command:

THE MATLAB® FUNCTION FEATURE NODE | 257

 model.func.create(<ftag>, 'MATLAB')

Define the function name and function arguments with the command:

model.func(<ftag>).setIndex('funcs', <function_name>, 0, 0)
model.func(<ftag>).setIndex('funcs', <arglist>, 0, 1)

where <function_name> is a string set with the function name and <arglist> is a
string that defines the list of the input arguments.

Function Input/Output Considerations

The functions called from COMSOL Multiphysics must support vector arguments of
any length. COMSOL calls a MATLAB® function using vector arguments to reduce
the number of expensive calls from COMSOL to MATLAB. All common MATLAB
functions such as sin, abs, and other mathematical functions support vector
arguments.

When you write your own functions, remember that the input arguments
are vectors. The output must have the same size as the input. All
arguments and results must be double-precision vectors real or complex
valued.

Consider the following example function where the coefficient c depends on the x
coordinate:

function c = func1(x)
if x > 0.6
c = x/1.6;
else
c = x^2+0.3;
end

This function looks good at first but it does not work in COMSOL Multiphysics
because the input x is a vector:

• Element-by-element multiplication, division, and power must be used—that is, the

operators .*, ./, and .^. Replace expressions such as x/1.6 and x^2+0.3 with
x./1.6 and x.^2+0.3, respectively.
• The comparison x > 0.6 returns a matrix with ones (true) for the entries where the
expression holds true and zeros (false) where it is false. The function evaluates the
conditional statement if, and only if, all the entries are true (1).

258 | CHAPTER 5: CALLING EXTERNAL FUNCTIONS

You can replace the if statement with a single assignment to the indices retrieved from
the x > 0.6 operation and another assignment to the indices where x ≤ 0.6. The
function could then look like this:

function c = func2(x)
c = (x./1.6).*(x>0.6) + (x.^2+0.3).*(x<=0.6);

Updating Functions
If the function M-file is modified using a text editor, click Clear Functions to ensure that
the functions’ modifications are updated in the COMSOL Multiphysics model.

An alternative is to select the Clear functions automatically before solving check box.

Defining Function Derivatives

Automatic differentiation is not supported with MATLAB® functions. In case the
MATLAB function has Jacobian contributions, its derivatives with respect to the
function input arguments need to be defined. By default COMSOL Multiphysics
assumes the derivatives to be null.

Expand the Derivatives section to define the derivatives of the function with respect to
the function arguments. In the table define the derivative for each function argument.
In the Function column enter the function name, in the Argument column enter the
argument. Finally in the Function derivative column enter the expression for the
corresponding derivative.

The function derivatives can also be defined by additional MATLAB

functions.

The section, Example: Define the Hankel Function, defined the function derivative by
entering the following settings in the table:

FUNCTION ARGUMENT FUNCTION DERIVATIVE

besselh nu (besselh(nu-1,x)-besselh(nu+1,x))/2
besselh x (besselh(0,x)-besselh(2,x))/2

THE MATLAB® FUNCTION FEATURE NODE | 259

260 | CHAPTER 5: CALLING EXTERNAL FUNCTIONS
 6

Command Reference

The main reference for the syntax of the commands available with LiveLink™ for
MATLAB® is the COMSOL Multiphysics Programming Reference Manual. This
section documents additional interface functions that come with the product.

In this chapter:

• Summary of Commands
• Commands Grouped by Function

261
 Summary of Commands

colortable mphmeshstats
mphaddplotdata mphmin
mphapplicationlibraries mphmodel
mphcd mphnavigator
mphdoc mphopen
mpheval mphparticle
mphevalglobalmatrix mphplot
mphevalpoint mphquad2tri
mphevalpointmatrix mphray
mphevaluate mphreadstl
mphgeom mphreduction
mphgetadj mphreport
mphgetcoords mphsave
mphgetexpressions mphsearch
mphgetproperties mphselectbox
mphgetselection mphselectcoords
mphgetu mphshowerrors
mphglobal mphsolinfo
mphimage2geom mphsolutioninfo
mphinputmatrix mphstart
mphint2 mphstate
mphinterp mphsurf
mphinterpolationfile mphtable
mphlaunch mphtags
mphload mphthumbnail
mphmatrix mphversion
mphmax mphviewselection
mphmean mphwritestl
mphmeasure mphxmeshinfo
mphmesh

262 | CHAPTER 6: COMMAND REFERENCE

C o m m a nds Grou p ed b y Fu n c t i on
INTERFACE FUNCTIONS

FUNCTION PURPOSE

mphcd Change the directory to the directory of the model.

mphdoc Open help window for a certain topic
mphlaunch Launch a COMSOL Multiphysics Client, connect it to the
server and load a model.
mphload Load a COMSOL model MPH-file.
mphopen GUI for opening recent model files.
mphsave Save a COMSOL model.
mphstart Connect MATLAB to a COMSOL server.
mphthumbnail Set or get model thumbnail.
mphversion Return the version number of COMSOL Multiphysics

GEOMETRY FUNCTIONS

FUNCTION PURPOSE

mphgeom Plot a geometry in a MATLAB figure.

mphimage2geom Convert image data to geometry.
mphmeasure Measure entities in a geometry.
mphviewselection Display a geometric entity selection in a MATLAB figure.

MESH FUNCTIONS

FUNCTION PURPOSE

mphmesh Plot a mesh in a MATLAB figure.

mphmeshstats Return mesh statistics and mesh data information.

UTILITY FUNCTIONS

FUNCTION PURPOSE

mphevaluate Evaluate Parameters expressions in model.

mphgetadj Return geometric entity indices adjacent to each other.
mphgetcoords Return point coordinates of geometry entities.

COMMANDS GROUPED BY FUNCTION | 263

 FUNCTION PURPOSE

mphgetu Return solution vectors.

mphinputmatrix Add matrix system for a linear solver.
mphinterpolationfile Save data in file readable by the Interpolation feature.
mphmatrix Get model matrices.
mphquad2tri Convert plot data quad mesh into simplex mesh.
mphreadstl Read an STL file and returns the data as a struct.
mphreduction Return reduced-order state-space matrices for a model.
mphselectbox Select a geometric entity using a rubber band/box.
mphselectcoords Select a geometric entity using point coordinates.
mphsolinfo Get information about a solution object.
mphsolutioninfo Get information about solution objects and datasets
containing given parameters.
mphstate Get state-space matrices for dynamic systems.
mphsurf Create plot data structure from surf data.
mphwritestl Export plot data as an STL file.
mphxmeshinfo Extract information about the extended mesh.

264 | CHAPTER 6: COMMAND REFERENCE

POSTPROCESSING FUNCTIONS

FUNCTION PURPOSE

mphaddplotdata Add a data plot to a model.

mpheval Evaluate expressions on node points.
mphevalglobalmatrix Evaluate global matrix variables.
mphevalpoint Evaluate expressions at geometry vertices.
mphevalpointmatrix Evaluate matrix quantities at points in the geometry
mphglobal Evaluate global quantities.
mphint2 Perform integration of expressions.
mphinterp Evaluate expressions in arbitrary points or datasets.
mphmax Perform maximum of expressions.
mphmean Perform mean of expressions.
mphmin Perform minimum of expressions.
mphparticle Evaluate expressions on particle trajectories.
mphplot Render a plot group in a figure window.
mphray Evaluate expressions on particle and ray trajectories.
mphreport Generate report to model or write report.
mphtable Get table data.
mphray Evaluate expressions on ray trajectories.

MODEL INFORMATION AND NAVIGATION

FUNCTION PURPOSE

mphapplicationlibraries GUI for viewing the product Application Libraries.

mphgetexpressions Get the model variables and parameters.
mphgetproperties Get properties from a model node.
mphgetselection Get information about a selection node.
mphmodel Return tags for the nodes and subnodes in the
COMSOL model object.
mphnavigator GUI for viewing the COMSOL model object.
mphsearch GUI for searching expressions in the COMSOL model
object.
mphshowerrors Show messages in error and warning nodes in the
COMSOL model object.

COMMANDS GROUPED BY FUNCTION | 265

 FUNCTION PURPOSE

mphtags Get tags and names for nodes in a COMSOL model.

mphthumbnail Get and set model thumbnail images

colortable
Return a MATLAB® colormap for a COMSOL Multiphysics color table.

SYNTAX
map = colortable(name)

DESCRIPTION
map = colortable(name) returns the color table (of 1024 colors) for name, where
name can be one of the following strings:

AuroraAustralis — A color table that spans from white through green and indigo
to blue as for the colors in the aurora australis (southern light).

AuroraAustralisDark — A color table that spans from white through green and
indigo to blue as for the colors in the aurora australis (southern light) but does not
start with the full white color.

AuroraBorealis — This color table resembles the colors in the aurora borealis
(northern light). The color table spans from white through green and indigo to blue
but with a larger indigo portion.

Cividis — Uses yellow and blue colors in a color table that is suited for normal vision,
a deuteranomaly, or red-green colorblindness.

Cyclic — A color table that varies the hue component of the hue-saturation-value
color model, keeping the saturation and value constant (equal to 1). The colors begin
with red, pass through yellow, green, cyan, blue, magenta, and return to red. This table
is useful to display periodic functions and has a sharp color gradient.

Disco — This color table spans from red through magenta and cyan to blue.

Discolight — Similar to Disco but uses lighter colors.

Grayscale — A color table that uses no color, only the gray scale varying linearly from
black to white.

Grayprint — Varies linearly from dark gray (0.95, 0.95, 0.95) to light gray (0.05,
0.05, 0.05). This color table overcomes two disadvantages that the GrayScale color

266 | CHAPTER 6: COMMAND REFERENCE

table has when used for printouts on paper—it gives the impression of being
dominated by dark colors and that white cannot be distinguished from the
background.

HeatCamera — The colors range from black through blue, magenta, red, and yellow
to white, corresponding to the colors in an image from a heat camera.

HeatCameraLight — Similar to HeatCamera but with slightly lighter colors.

JupiterAuroraBorealis — the color table spans from black through blue to white
as for the Jupiter’s aurora.

Rainbow — The color ordering in this table corresponds to the wavelengths of the
visible part of the electromagnetic spectrum: beginning at the small-wavelength end
with dark blue, the colors range through shades of blue, cyan, green, yellow, and red.

Rainbowlight — Similar to Rainbow, this color table uses lighter colors.

Thermal — Ranges from black through red and yellow to white, which corresponds
to the colors iron takes as it heats up.

ThermalEquidistant — Similar to Thermal but uses equal distances from black to

red, yellow, and white, which means that the black and red regions become larger.

ThermalNarrow — Similar to Thermal but does not include the full black and white
end colors.

Traffic — Spans from green through yellow to red.

Trafficlight — Similar to Traffic but uses lighter colors.

Twilight — The color table uses colors associated with twilight (the illumination of
the Earth’s lower atmosphere when the Sun is not directly visible), spanning colors
from pink through white to blue.

Wave — Ranges linearly from blue to light gray, and then linearly from white to red.
When the range of the visualized quantity is symmetric around zero, the color red or
blue indicates whether the value is positive or negative, and the saturation indicates the
magnitude.

Wavelight — Similar to Wave and ranges linearly from a lighter blue to white (instead
of light gray) and then linearly from white to a lighter red.

Calling colortable is equivalent to calling the corresponding colormap function

directly.

COMMANDS GROUPED BY FUNCTION | 267

 EXAMPLE
Create a rainbow color map

map = colortable('Rainbow');
map = rainbow;

mphaddplotdata

SYNTAX
mphaddplotdata(model,'type',type,'data',data,...)

DESCRIPTION
mphaddplotdata(model,'type',type,'data',data,...) create a new plot using
the data data as a plottype type.

The function mphaddplotdata accepts the following property/value pairs:

TABLE 6-1: PROPERTY/VALUE PAIRS FOR THE MPHADDPLOTDATA COMMAND.

PROPERTY PROPERTY VALUE DEFAULT DESCRIPTION

arrowtype arrow | arrow Arrow type

arrowhead |
cone
arrowbase head | tail | tail Arrow base
center
clearplot off | on off Clear plot group
color vector | Color specification
string | index
| matrix
colortable string | cell Color table name
array
data matrix Coordinates data
edges off| on on Plot dataset edges
elementdata matrix Element data (low level)
latex off | on off Use LaTeX markup
normaldata matrix Data for normals (low level)
plotgroup string Plot group tag
radius vector Radius for tubes

268 | CHAPTER 6: COMMAND REFERENCE

TABLE 6-1: PROPERTY/VALUE PAIRS FOR THE MPHADDPLOTDATA COMMAND.

PROPERTY PROPERTY VALUE DEFAULT DESCRIPTION

type arrow | line | Plot type

surface |
annotation |
tube | point
vector matrix Direction vector

SEE ALSO
mphplot

mphapplicationlibraries
Graphical user interface (GUI) for viewing the Application Libraries.

SYNTAX
mphapplicationlibraries

DESCRIPTION
mphapplicationlibraries starts a GUI to visualize and access the example model
available in the COMSOL Application Libraries. The model MPH-file can be loaded
in MATLAB® and the model documentation PDF-file is accessible directly.

Models that are specific to LiveLink™ for MATLAB® also contains the script M-file.

COMMANDS GROUPED BY FUNCTION | 269

 Models that are specific to LiveLink™ for Simulink® also contains the simulation
SLX-file and opens directly in Simulink.

mphcd
Change directory to the directory of the model.

SYNTAX
mphcd(model)

DESCRIPTION
®
mphcd(model) changes the current directory in MATLAB to the directory where the
model was last saved.

SEE ALSO
mphload, mphsave

270 | CHAPTER 6: COMMAND REFERENCE

mphdoc
Open help window for a certain topic.

SYNTAX
mphdoc
mphdoc(node)
mphdoc(node,fname)
mphdoc api

DESCRIPTION
mphdoc opens the COMSOL documentation Help Desk.

mphdoc(node) opens the help window for the entry on node.

mphdoc(node,fname) opens the help window for the entry on node with the feature
fname.

mphdoc api opens a window with the JavaDoc help for the COMSOL API.

EXAMPLE
Create a model a model object:

model = ModelUtil.create('Model');

Get the documentation for the mesh node;

mphdoc(model.mesh)

Get the documentation of the geometry feature Rectangle:

mphdoc(model.geom,'Rectangle')

SEE ALSO
mphapplicationlibraries

mpheval
Evaluate expressions on node points.

SYNTAX
pd = mpheval(model,{e1,...,en},...)

COMMANDS GROUPED BY FUNCTION | 271

 DESCRIPTION
pd = mpheval(model,{e1,...,en},...) returns the post data pd for the
expressions e1,...,en.

The output value pd is a structure with fields expr, p, t, ve, unit and fields for data
values.

• The field expr contains the expression name evaluated.

• For each expression e1,...,en a field with the name d1,... dn is added with the
numerical values. The columns in the data value fields correspond to node point
coordinates in columns in p. The data contains only the real part of complex-valued
expressions.
• The field p contains node point coordinate information. The number of rows in p
is the number of space dimensions.
• The field t contains the indices to columns in p of a simplex mesh, each column in
t representing a simplex.
• The field ve contains indices to mesh elements for each node point.
• The field unit contains the list of the unit for each expression.
The function mpheval accepts the following property/value pairs:
TABLE 6-2: PROPERTY/VALUE PAIRS FOR THE MPHEVAL COMMAND.

PROPERTY PROPERTY DEFAULT DESCRIPTION

VALUE

complexfun off | on on Use complex-valued functions with

real input
complexout off | on on Return complex values
dataonly off | on off Only return expressions value
dataset String Dataset tag
edim point | Geometry Evaluate on elements with this
edge | space space dimension
boundary | dimension
domain | 0
| 1 | 2 | 3
matherr off | on off Error for undefined operations or
expressions
outersolnum Positive 1 Solution number for parametric
integer | sweep
all | end
pattern lagrange | lagrange Specifies if evaluation takes place in
gauss Lagrange points or in Gauss points

272 | CHAPTER 6: COMMAND REFERENCE

TABLE 6-2: PROPERTY/VALUE PAIRS FOR THE MPHEVAL COMMAND.

PROPERTY PROPERTY DEFAULT DESCRIPTION

VALUE

phase Scalar 0 Phase angle in degrees

recover off | ppr | off Accurate derivative recovery
pprint
refine Integer 1 Refinement of elements for
evaluation points
selection Integer vector all Set selection tag or entity number
| string | all
smooth internal | internal Smoothing setting
none |
everywhere
solnum Integer vector all Solutions for evaluation
| all | end
t Double array Times for evaluation
unit String | Cell Unit to use for the evaluation
array
The property Dataset controls which dataset is used for the evaluation. Datasets
contain or refer to the source of data for postprocessing purposes. Evaluation is
supported only on Solution datasets.

The property Edim decides which elements to evaluate on. Evaluation takes place only
on elements with space dimension Edim. If not specified, Edim equal to the space
dimension of the geometry is used. The setting is specified as one of the following
strings 'point', 'edge', 'boundary' or 'domain'. In previous versions it was only
possible to specify Edim as a number. For example, in a 3D model, if evaluation is done
on edges (1D elements), Edim is 1. Similarly, for boundary evaluation (2D elements),
Edim is 2, and for domain evaluation (3D elements), Edim is 3 (default in 3D).

Use Recover to recover fields using polynomial-preserving recovery. This techniques

recover fields with derivatives such as stresses or fluxes with a higher theoretical
convergence than smoothing. Recovery is expensive so it is turned off by default. The
value pprint means that recovery is performed inside domains. The value ppr means
that recovery is also applied on all domain boundaries.

The property Refine constructs evaluation points by making a regular refinements of

each element. Each mesh edge is divided into Refine equal parts.

COMMANDS GROUPED BY FUNCTION | 273

 The property Smooth controls if the post data is forced to be continuous on element
edges. When Smooth is set to internal, only elements not on interior boundaries are
made continuous.

The property Solnum is used to select the solution to plot when a parametric,
eigenvalue, or time-dependent solver has been used to solve the problem.

The property Outersolnum is used to select the solution to plot when a parametric
sweep has been used in the study.

When the property Phase is used, the solution vector is multiplied with
exp(i*phase) before evaluating the expression.

The expressions e1,...,en are evaluated for one or several solutions. Each solution
generates an additional row in the data fields of the post data output structure. The
properties Solnum and t control which solutions are used for the evaluations. The
Solnum property is available when the dataset has multiple solutions — for example, in
the case of parametric, eigenfrequency, or time-dependent solutions. The t property is
available only for time-dependent problems. If Solnum is provided, the solutions
indicated by the indices provided with the Solnum property are used. If t is provided,
solutions are interpolated. If neither Solnum nor t is provided, all solutions are
evaluated.

For time-dependent problems, the variable t can be used in the expressions ei. The
value of t is the interpolation time when the property t is provided, and the time for
the solution, when Solnum is used. Similarly, lambda and the parameter are available
as eigenvalues for eigenvalue problems and as parameter values for parametric
problems, respectively.

EXAMPLE
Evaluate the temperature at node points:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
std.feature.create('stat','Stationary');
std.run;
dat = mpheval(model,'T');

Evaluate both the total heat flux magnitude and the temperature:

data = mpheval(model,{'ht.tfluxMag', 'T'});

Evaluate the temperature and return the data only:

data = mpheval(model,'T','dataonly','on');

274 | CHAPTER 6: COMMAND REFERENCE

Evaluate the temperature at the node points in domain 2:

data = mpheval(model,'T','selection',2);

Evaluate the temperature at the node points on boundary 7:

data = mpheval(model,'T','selection',7,'edim','boundary');

Evaluate the temperature at second order Lagrange points:

data = mpheval(model,'T','refine',2);

Evaluate the temperature at the Gauss points:

data = mpheval(model,'T','pattern','gauss');

Evaluate the temperature at every time step computed with power set to 30:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
param = std.feature.create('param','Parametric');
time = std.feature.create('time','Transient');
time.set('tlist', 'range(0,1,25)');
param.setIndex('pname','power',0);
param.setIndex('plistarr','30 60 90',0);
std.run;

data = mpheval(model,'T','dataset','dset2');

Evaluate the temperature at the fifth time step:

data = mpheval(model,'T','dataset','dset2','solnum',5);

Evaluate the temperature at 10.5 sec and 15.2 sec:

data = mpheval(model,'T','dataset','dset2','t',[10.5,15.2]);

Evaluate the temperature at every time step computed with power set to 90:

data = mpheval(model,'T','dataset','dset2','outersolnum',3);

SEE ALSO
mphevalglobalmatrix, mphevalpoint, mphevalpointmatrix, mphglobal,
mphint2, mphinterp, mphparticle, mphray

mphevalglobalmatrix
Evaluate global matrix variables.

SYNTAX
M = mphevalglobalmatrix(model,expr,...)

COMMANDS GROUPED BY FUNCTION | 275

 DESCRIPTION
M = mphevalglobalmatrix(model,expr,...) evaluates the global matrix of the
variable expr and returns the full matrix M.

The function mphevalglobalmatrix accepts the following property/value pairs:

TABLE 6-3: PROPERTY/VALUE PAIRS FOR THE MPHEVALGLOBALMATRIX COMMAND.

PROPERTY PROPERTY VALUE DEFAULT DESCRIPTION

dataset String Dataset tag

dataseries none | average | none Data series operation
sum
outerdataseries none | average | none Outer data series
sum operation
outersolnum Positive integer | 1 Solution number for
end | all parametric sweep
solnum Integer vector | all Solution for evaluation
end | all
t Double array Time for evaluation
trans none | inverse | none The transformation to
maxwellmutual | apply to the matrix
mutualmaxwell | sy
| sz | ys | yz | zs |
zy
y0 Double array Taken from If trans is sy or ys: The
the physics characteristic admittance
interfaces
z0 Double array Taken from If trans is sz or zs: The
the physics characteristic impedance
interfaces

Note: S-parameters evaluation requires the RF module.

EXAMPLE
Load lossy_circulator_3d.mph from the RF Module’s Applications Libraries:

model = mphopen('lossy_circulator_3d.mph');

Evaluate the S-parameters matrix using the solution dataset dset4:

M = mphevalglobalmatrix(model,'emw.SdB','dataset','dset4');

276 | CHAPTER 6: COMMAND REFERENCE

SEE ALSO
mpheval, mphevalpoint, mphevalpointmatrix, mphglobal, mphint2, mphinterp,
mphparticle, mphray

mphevalpoint
Evaluate expressions at geometry vertices.

SYNTAX
[v1,...,vn] = mphevalpoint(model,{e1,...,en},...)
[v1,...,vn,unit] = mphevalpoint(model,{e1,...,en},...)

DESCRIPTION
[v1,...,vn] = mphevalpoint(model,{e1,...,en},...) returns the results from
evaluating the expressions e1,...,en at the geometry vertices. The values v1,...,vn
can either be a cell array or a matrix depending on the options.

[v1,...,vn,unit] = mphevalpoint(model,{e1,...,en},...) also returns the

unit of all expressions e1,...,en in the 1xN cell array unit.

The function mphevalpoint accepts the following property/value pairs:

TABLE 6-4: PROPERTY/VALUE PAIRS FOR THE MPHEVALPOINT COMMAND.

PROPERTY PROPERTY DEFAULT DESCRIPTION

VALUE

dataset String Dataset tag

dataseries none | mean | none Data series operation
int | max |
min | rms |
std | var
matrix off | on on Return a matrix if possible
minmaxobj real | abs real The value being treated if
dataseries is set to max or min
outersolnum Positive integer 1 Solution number for parametric
| all | end sweep
selection Integer vector All Set selection tag or entity number
| string | all domains
solnum Integer vector all Solutions for evaluation
| all | end
squeeze on | off on Squeeze singleton dimension
t Double array Times for evaluation

COMMANDS GROUPED BY FUNCTION | 277

 The property Dataset controls which dataset is used for the evaluation. Datasets
contain or refer to the source of data for postprocessing purposes. Evaluation is
supported only on Solution datasets.

The Dataseries property is used to control any filtering of the data series. The
supported operations are: average (mean), integral (int), maximum (max), minimum
(min), root mean square (rms), standard deviation (std) and variance (var).

Set the property Matrix to off to get the results in a cell array format.

In case the property Datseries is either min or max, you can specify how the values
are treated using the property Minmaxobj. Use either the real data or the absolute data.

The property Solnum is used to select the solution to plot when a parametric,
eigenvalue, or time-dependent solver has been used to solve the problem.

The expressions e1,...,en are evaluated for one or several solutions. Each solution
generates an additional row in the data fields of the post data output structure. The
properties Solnum and t control which solutions are used for the evaluations. The
Solnum property is available when the dataset has multiple solutions — for example in
the case of parametric, eigenfrequency, or time-dependent solutions. The t property is
available only for time-dependent problems. If Solnum is provided, the solutions
indicated by the indices provided with the Solnum property are used. If t is provided,
solutions are interpolated. If neither Solnum nor t is provided, all solutions are
evaluated.

For time-dependent problems, the variable t can be used in the expressions ei. The
value of t is the interpolation time when the property t is provided, and the time for
the solution, when Solnum is used. Similarly, lambda and the parameter are available
as eigenvalues for eigenvalue problems and as parameter values for parametric
problems, respectively.

EXAMPLE
Evaluate the temperature on all geometry points:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
std.feature.create('stat','Stationary'); std.run;

T = mphevalpoint(model,'T');

Evaluate the temperature on point 5:

T = mphevalpoint(model,'T','selection',5);

278 | CHAPTER 6: COMMAND REFERENCE

Evaluate the temperature and the magnitude of the total heat flux on point 5:

[T, heatflux, unit] = mphevalpoint(model,{'T','ht.tfluxMag'},...

'selection',5);

Evaluate the temperature at every time step computed with power set to 30:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
param = std.feature.create('param','Parametric');
time = std.feature.create('time','Transient');
time.set('tlist', 'range(0,1,25)');
param.setIndex('pname','power',0)
param.setIndex('plistarr','30 60 90',0);
std.run;

T = mphevalpoint(model,'T','selection',5,'dataset','dset2');

Evaluate the temperature at the seventh time step:

T = mphevalpoint(model,'T','selection',5,'dataset','dset2',...
'solnum',7);

Evaluate the temperature at 10.5 sec:

T = mphevalpoint(model,'T','selection',5,'dataset','dset2',...
't',10.5);

Evaluate the temperature on point 5 computed with power set to 90:

T = mphevalpoint(model,'T','selection',5,'dataset','dset2',...
'outersolnum',3)

Evaluate the temperature average over all time steps:

T_avg = mphevalpoint(model,'T','selection',5,...
'dataset','dset2','dataseries','average');

SEE ALSO
mpheval, mphevalglobalmatrix, mphevalpointmatrix, mphglobal, mphint2,
mphinterp, mphparticle, mphray

mphevalpointmatrix
Evaluate matrix quantities at points in the geometry.

SYNTAX
M = mphevalpointmatrix(model, expr, ...)

COMMANDS GROUPED BY FUNCTION | 279

 DESCRIPTION
M = mphevalpointmatrix(model, expr, ...) evaluates the point matrix of the
variable expr and returns the full matrix M.

The function mphevalpointmatrix accepts the following property/value pairs:

TABLE 6-5: PROPERTY/VALUE PAIRS FOR THE MPHEVAL COMMAND.

PROPERTY PROPERTY DEFAULT DESCRIPTION

VALUE

dataset String Dataset tag

dataseries none | none Dataseries operation
average | sum
outersolnum Positive integer 1 Solution number for parametric
| end | all sweep
selection String | positive Set selection tag or entity number
integer array
solnum Integer vector all Solution for evaluation
| end | all
t Double array Time for evaluation

SEE ALSO
mpheval, mphevalglobalmatrix, mphevalpoint, mphglobal, mphint2,
mphinterp, mphparticle, mphray

mphevaluate
Evaluate parameter expressions in models.

SYNTAX
mphevaluate(model,expr)
str = mphevaluate(model)
[value,unit,def,descr] = mphevaluate(model,expr,...)
[value,...] = mphevaluate(model,expr,unit)
[value,...] = mphevaluate(model,expr,asvalue)

DESCRIPTION
mphevaluate(model,expr) evaluates the expression expr defined in the Parameters
node.

str = mphevaluate(model) returns all the expressions defined in the Parameters

node as a structs arrays.

280 | CHAPTER 6: COMMAND REFERENCE

str = mphevaluate(model,{e1,...}) returns all the expressions defined in the cell
array {e1,...} as a structs arrays. When multiple expressions are evaluated only one
output is supported.

[value,unit,def,descr] = mphevaluate(model,expr,...) evaluates the

expression expr and return the unit (unit), the definition in the model (def) and the
description (descr).

[value,...] = mphevaluate(model,expr,unit) evaluates the expression expr in

the unit defined by unit.

value = mphevaluate(model,expr,unit,'value') returns only the value of the

expression expr.

unit = mphevaluate(model,expr,unit,'unit') returns only the unit of the

expression expr.

def = mphevaluate(model,expr,unit,'valueunit') returns as a string the value

and the unit of the expression expr. This is useful to set a new parameter based on an
existing one.

The evaluation does not require an existing solution dataset in the model.

EXAMPLE
Evaluate the parameter power defined in the model:

model = mphopen('model_tutorial_llmatlab');
power = mphevaluate(model,'power');

Evaluate Temp in degrees Celsius and its definition in the model:

[Temp,unit,def] = mphevaluate(model,'Temp','degC');

Evaluate an expression of parameters:

Temp = mphevaluate(model,'Temp+20[degC]','degF');

SEE ALSO
mpheval, mphglobal, mphinterp, mphparticle, mphray

mphgeom
Plot a geometry in a MATLAB® figure.

COMMANDS GROUPED BY FUNCTION | 281

 SYNTAX
mphgeom(model)
mphgeom(model,geomtag,...)
mphgeom(model,geomtag,'workplane',wptag,...)
h = mphgeom(model,geomtag,...)

DESCRIPTION
mphgeom(model) plots the model geometry in a MATLAB figure. If the model only
contains one geometry then the geomtag can be empty.

mphgeom(model,geomtag,...) plots the model geometry with the tag geomtag in a

MATLAB figure.

mphgeom(model,geomtag,'workplane',wptag...) plots the 2D geometry defined

in the workplane with the tag wptag in a MATLAB figure.

h = mphgeom(model,geomtag,...) also returns a handle of the plotted entities.

The function mphgeom accepts the following property/value pairs:

TABLE 6-6: PROPERTY/VALUE PAIRS FOR THE MPHGEOM COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

build on | off | on Build the geometry

current | before plotting
string
domainlabels on | off off Show domain labels
domainlabelscolor Char k Color for domain
labels
edgecolor Char k Edge color
edgelabels on | off off Show edge labels
edgelabelscolor Char k Color for edge labels
edgemode on | off on Show edges
entity point | edge Geometric entity to
| boundary | select
domain
facealpha Double 1 Set transparency value
facecolor Char | vector gray Face color
of RGB values
facelabels on | off off Show face labels
facelabelscolor Char k Color for face labels
facemode on | off on Show faces

282 | CHAPTER 6: COMMAND REFERENCE

TABLE 6-6: PROPERTY/VALUE PAIRS FOR THE MPHGEOM COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

parent Double Parent axes

selection Positive Selection
integer array
vertexlabels on | off off Show vertex labels
vertexlabelscolor Char k Color for vertex
labels
vertexmode on | off off Show vertices
view String | auto '' View settings
The Build property determines if mphgeom build the geometry prior to display it. If
the Build property is set with a geometry object tag, the geometry is built up to that
object. mphgeom only displays built geometry objects.

Use the Workplane property to show the 2D geometry that is define inside the
workplane with specified tag.

EXAMPLE
Plot the model geometry:

model = mphopen('model_tutorial_llmatlab.mph');
mphgeom(model)

Plot the model geometry with face labels:

mphgeom(model,'geom1','facelabels','on','facelabelscolor','r');

Plot boundaries 7, 8, 9 and 11:

mphgeom(model,'geom1','entity','boundary',...
'selection',[7:9,11]);

The geometry can be plotted with view settings applied. This results in a geometry plot
with grid, axes labels, lights, hiding etc. applied to the plot. Usually it is sufficient to
use the auto setting, but any valid view can be applied:

mphmesh(model, 'mesh1', 'view', 'auto')t

Plot the model geometry on an existing axis:

figure(2);
mphgeom(model, 'geom1','parent', gca);

SEE ALSO
mphmesh, mphviewselection

COMMANDS GROUPED BY FUNCTION | 283

 mphgetadj
Return geometric entity indices that are adjacent to each other.

SYNTAX
n = mphgetadj(model,geomtag,returntype,adjtype,adjnumber)
[n,m] = mphgetadj(model,geomtag,returntype,adjtype,adjnumber)

DESCRIPTION
n = mphgetadj(model,geomtag,returntype,adjtype,adjnumber) returns the
indices of the adjacent geometry entities.

[n,m] = mphgetadj(model,geomtag,returntype,adjtype,adjnumber) returns

the indices of the adjacent geometry entities in n. m contains the indices of entities that
connect the entities adjnumber the best.

returntype is the type of the geometry entities whose index are returned.

adjtype is the type of the input geometric entity.

The entity type can be one of 'point', 'edge', 'boundary', or 'domain' following
the entity space dimension defined below:

• 'domain': maximum geometry space dimension

• 'boundary': maximum geometry space dimension − 1

• 'edges': 1 (for 3D geometries only)

• 'point': 0

EXAMPLE
Return the indices of the boundaries adjacent to point 2:

model = mphopen('model_tutorial_llmatlab');

bnd_idx = mphgetadj(model, 'geom1', 'boundary', 'point', 2);

Return the indices of the points adjacent to domain 2:

pt_idx = mphgetadj(model, 'geom1', 'point', 'domain', 2);

Return the indices of the adjacent edges to boundaries 1, 2 and 9 and the indices of
the edges that connect the boundaries the best. Then show the results in a figure:

[idx1, idx2] = mphgetadj(model, 'geom1', 'edge',...

'boundary', [1, 2, 9]);
mphviewselection(model,'geom1',[1,2,9],'entity','boundary',...
'edgemode','off')

284 | CHAPTER 6: COMMAND REFERENCE

 hold on
mphviewselection(model,'geom1',idx1,'entity','edge',...
'edgemode','off','facemode','off','edgecolorselected','b')
mphviewselection(model,'geom1',idx2,'entity','edge',...
'edgemode','off','facemode','off','edgecolorselected','g')

SEE ALSO
mphgetcoords, mphselectbox, mphselectcoords

mphgetcoords
Return point coordinates of geometry entities.

SYNTAX
c = mphgetcoords(model,geomtag,entitytype,entitynumber)

DESCRIPTION
c = mphgetcoords(model,geomtag,entitytype,entitynumber) returns the
coordinates of the points that belong to the entity object with the type entitytype
and the index entitynumber.

The entitytype property can be one of 'point', 'edge', 'boundary' or 'domain'

following the entity space dimension defined below:

• 'domain': maximum geometry space dimension

• 'boundary': maximum geometry space dimension -1

• 'edge': 1 (only for 3D geometry)

'point': 0

EXAMPLE
Return the coordinates of points that belong to domain 2:

model = mphopen('model_tutorial_llmatlab');

c0 = mphgetcoords(model, 'geom1', 'domain', 2);

Return the coordinates of points that belong to boundary 5:

c1 = mphgetcoords(model, 'geom1', 'boundary', 5);

Return the coordinates of point number 10:

c2 = mphgetcoords(model, 'geom1', 'point', 10);

COMMANDS GROUPED BY FUNCTION | 285

 SEE ALSO
mphgetadj, mphselectbox, mphselectcoords

mphgetexpressions
Get the model variables and model parameters expressions.

SYNTAX
expr = mphgetexpressions(modelnode)

DESCRIPTION
expr = mphgetexpressions(modelnode) returns expressions from the node
modelnode as a cell array. expr contains the list of the variable names, the variable
expressions and the variable descriptions.

Note that not all nodes have expressions defined.

EXAMPLE
Get the expressions defined in the parameters node

model = mphopen('model_tutorial_llmatlab');
expr = mphgetexpressions(model.param)

SEE ALSO
mphgetproperties, mphgetselection, mphmodel, mphnavigator, mphsearch

mphgetproperties
Get the properties from a model node.

SYNTAX
str = mphgetproperties(modelnode)
[str,allowed] = mphgetproperties(modelnode)

DESCRIPTION
str = mphgetproperties(modelnode) returns the structure str containing the
properties that are defined for the node modelnode.

[str,allowed] = mphgetproperties(modelnode) also returns the structure

allowed containing the allowed values for the corresponding properties.

286 | CHAPTER 6: COMMAND REFERENCE

The function mphgetproperties accepts the following property/value pairs:
TABLE 6-7: PROPERTY/VALUE PAIRS FOR THE MPHGETPROPERTIES COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

propnames String cell List of properties to

array show only
returnstrings on | off on Return data as string
showsel on | off off Return the properties
selection

EXAMPLE
Build the mesh in the model model_tutorial_llmatlab.mph and get the mesh size
properties and their allowed values:

model = mphopen('model_tutorial_llmatlab');
mesh1 = model.component('comp1').mesh('mesh1');
mesh1.run;
msize = mesh1.feature('size');
[prop, allowed] = mphgetproperties(msize)

Get the min and max mesh size properties only:

prop = mphgetproperties(msize,'propnames',{'hmin','hmax'})

Return the property data not as string:

prop = mphgetproperties(msize,'propnames',{'hmin','hmax'},...
'returnstrings','off')

SEE ALSO
mphgetexpressions, mphgetselection, mphmodel, mphnavigator, mphsearch

mphgetselection
Get information about a selection node.

SYNTAX
info = mphgetselection(selnode)

DESCRIPTION
info = mphgetselection(selnode) returns the selection data of the selection node
selnode.

The output info is a MATLAB® structure defined with the following fields:

COMMANDS GROUPED BY FUNCTION | 287

 • dimension, the space dimension of the geometric entity selected.
• geom, the geometry tag.
• entities, the indices of the selected entities.
• isGlobal, a Boolean expression that indicates if the selection is global.

EXAMPLE
Add a selection node to the model busbar.mph and retrieve its information:

model = mphopen('model_tutorial_llmatlab.mph');
ball = model.selection.create('ball','Ball');
ball.set('entitydim',2);
ball.set('posz',11e-3');
ball.set('r',1e-5);
info = mphgetselection(model.selection('ball'))

SEE ALSO
mphgetexpressions, mphgetproperties, mphmodel, mphnavigator, mphsearch

mphgetu
Return a solution vector.

SYNTAX
U = mphgetu(model,...)
[U,Udot] = mphgetu(model,...)

DESCRIPTION
U = mphgetu(model) returns the solution vector U for the default solution dataset.

[U,Udot] = mphgetu(model,...) returns in addition Udot, which is the time

derivative of the solution vector. This syntax is available for a time-dependent solution
only.

For a time-dependent and parametric analysis type, the last solution is returned by
default. For an eigenvalue analysis type the first solution number is returned by default.

The function mphgetu accepts the following property/value pairs:

TABLE 6-8: PROPERTY/VALUE PAIRS FOR THE MPHGETU COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

soltag String Solver node tag

solnum Positive integer vector Solution for evaluation

288 | CHAPTER 6: COMMAND REFERENCE

TABLE 6-8: PROPERTY/VALUE PAIRS FOR THE MPHGETU COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

type String Sol Solution type

matrix off | on on Store as matrix if possible
The Solname property set the solution dataset to use associated with the defined solver
node.

Type is used to select the solution type. This is 'Sol' by default. The valid types are:
'Sol' (main solution), 'Reacf' (reaction force), 'Adj' (adjoint solution), 'Fsens'
(functional sensitivity) and 'Sens' (forward sensitivity).

If Solnum is a vector and the result has been obtained with the same mesh then the
solution is stored in a matrix if the Matrix option is set to 'on'.

EXAMPLE
Extract the solution vector:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
std.feature.create('stat','Stationary');
std.run;

U = mphgetu(model);

Extract the reaction force vector:

reacf = mphgetu(model,'type','reacf');

Extract the solution vectors for the first and the last time step:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
param = std.feature.create('param','Parametric');
time = std.feature.create('time','Transient');
time.set('tlist', 'range(0,1,25)');
param.setIndex('pname','power',0);
param.setIndex('plistarr','30 60 90',0);
std.run;

U = mphgetu(model,'solnum',[1,26]);

Extract the solution vector computed with power set to 30:

U = mphgetu(model,'soltag','sol3');

SEE ALSO
mphsolinfo

COMMANDS GROUPED BY FUNCTION | 289

 mphglobal
Evaluate global quantities.

SYNTAX
[d1,...,dn] = mphglobal(model,{e1,...,en},...)
[d1,...,dn,unit] = mphglobal(model,{e1,...,en},...)

DESCRIPTION
[d1,...,dn] = mphglobal(model,{e1,...,en},...) returns the results from
evaluating the global quantities specified in the string expression e1,..., en.

[d1,...,dn,unit] = mphglobal(model,{e1,...,en},...) also returns the unit

of the expressions e1,..., en. unit is a nx1 cell array.

The function mphglobal accepts the following property/value pairs:

TABLE 6-9: PROPERTY/VALUE PAIRS FOR THE MPHGLOBAL COMMAND.

PROPERTY PROPERTY VALUE DEFAULT DESCRIPTION

complexfun off | on on Use complex-valued functions

with real input
complexout off | on on Return complex values
dataset String Active solution Dataset tag
dataset
matherr off | on off Error for undefined operations
or expressions
outersolnum Positive integer | 1 Solution number for
all | end parametric sweep
phase Scalar 0 Phase angle in degrees
solnum Integer vector | all Solution for evaluation
all | end
t Double array Time for evaluation
unit String | cell array Unit to use for the evaluation

The property Dataset controls which dataset is used for the evaluation. Datasets
contain or refer to the source of data for postprocessing purposes. Evaluation is
supported only on solution datasets.

When the property Phase is used, the solution vector is multiplied with
exp(i*phase) before evaluating the expression.

290 | CHAPTER 6: COMMAND REFERENCE

The expressions ei are evaluated for one or several solutions. Each solution generates
an additional row in the output data array di. The properties solnum and t control
which solutions are used for the evaluations. The solnum property is available when
the dataset has multiple solutions — for example, in the case of parametric,
eigenfrequency, or time-dependent solutions. The t property is available only for
time-dependent problems. If solnum is provided, the solutions indicated by the indices
provided with the solnum property are used. If t is provided, solutions are
interpolated. If neither solnum nor t is provided, all solutions are evaluated.

For time-dependent problems, the variable t can be used in the expressions ei. The
value of t is the interpolation time when the property t is provided, and the time for
the solution, when solnum is used. Similarly, lambda and the parameter are available
as eigenvalues for eigenvalue problems and as parameter values for parametric
problems, respectively.

In case of multiple expression if the unit property is defined with a string, the same
unit is used for both expressions. To use different units, set the property with a cell
array. In case of inconsistent unit definition, the default unit is used instead.

Solnum is used to select the solution number when a parametric, eigenvalue, or

time-dependent solver has been used.

Outersolnum is used to select the outer solution number when a parametric sweep
has been used in the study step node.

EXAMPLE
Evaluate the maximum temperature in the model

model = mphopen('model_tutorial_llmatlab');
model.cpl.create('maxop','Maximum','geom1').selection.all;
std = model.study.create('std');
std.feature.create('stat','Stationary');
std.run;

maxT = mphglobal(model,'maxop(T)')

Evaluate the maximum temperature in the model in degrees Celsius

maxT = mphglobal(model,'maxop(T)','unit','degC')

Evaluate a global expression at every time step computed with power set to 30:

model = mphopen('model_tutorial_llmatlab');
model.cpl.create('maxop', 'Maximum', 'geom1').selection.all;
std = model.study.create('std');
param = std.feature.create('param','Parametric');

COMMANDS GROUPED BY FUNCTION | 291

 time = std.feature.create('time','Transient');
time.set('tlist', 'range(0,1,25)');
param.setIndex('pname','power',0);
param.setIndex('plistarr','30 60 90',0);
std.run;
maxT = mphglobal(model,'maxop(T)','dataset','dset2');

Evaluate maxop(T) for the first and fifth time step:

maxT = mphglobal(model,'maxop(T)',dataset','dset2',...
'solnum',[1,5]);

Evaluate maxop(T) at 20.512 sec:

maxT = mphglobal(model,'maxop(T)',dataset','dset2',...
't',20.512);

Evaluate maxop(T) at every time step computed with power set to 90:

maxT = mphglobal(model,'maxop(T)','dataset','dset2',...
'outersolnum',3);

SEE ALSO
mpheval, mphevalglobalmatrix, mphevalpoint, mphevalpointmatrix, mphint2,
mphinterp

mphimage2geom
Convert image data to a geometry.

SYNTAX
model = mphimage2geom(imagedata,level,...)

DESCRIPTION
model = mphimage2geom(imagedata,level,...) converts the image contained in
imagedata into a geometry which is returned in the model object model.

The contour of the image is defined by the value level. imagedata must be a 2D
matrix.

292 | CHAPTER 6: COMMAND REFERENCE

The function mphimage2geom accepts the following property/value pairs:
TABLE 6-10: PROPERTY/VALUE PAIRS FOR THE MPHIMAGE2GEOM COMMAND.

PROPERTY PROPERTY VALUE DEFAULT DESCRIPTION

compose on | off on Create compose nodes for

overlapping solids
curvetype auto | polygon auto Type of curve to create the
geometry object
geom Geometry node geom1 Geometry creation
minarea Value 1 Minimum area for interior
curves (in square pixels)
mindist Value 1 Minimum distance between
coordinates in curves (in
pixels)
modeltag String Model Model tag in a COMSOL
server
rectangle on | off off Insert rectangle in the
geometry
rtol Value 1e-3 Relative tolerance for
interpolation curves
scale Value 1 Scale factor from pixels to
geometry scale
type solid | closed | solid Type of geometry object
open

The default curve types creates a geometry with the best suited geometrical primitives.
For interior curves this is Interpolation Curves, and for curves that are touching the
perimeter of the image, Polygons is used.

To add the geometry created with mphimage2geom, specify the geometry node with
property geom.

EXAMPLE
Create a set of point coordinates:

p = (peaks+7)*5;

Display contour plot of the point data:

figure(1); [c,h] = contourf(p); clabel(c, h); colorbar

Create a geometry object following the contour made with point of value 50:

COMMANDS GROUPED BY FUNCTION | 293

 model = mphimage2geom(p, 50);
figure(2); mphgeom(model)

Create the same geometry object with a scale factor of 1e-3 and add it into an existing
3D model:

model = mphopen('model_tutorial_llmatlab');
wp1 = model.component('comp1').geom('geom1').feature.create('wp1',
'WorkPlane');
wp1.set('quickz', 1e-2);
mphimage2geom(p, 50,'scale',1e-3,wp1.geom);
mphgeom(model)

Create a geometry using MRI data. The geometry object is created following the
contour made with point of value 30 and disregard objects with an area (in pixel) lower
than 2:

mri = load('mri');
im = mri.D(:,:,1,1);
figure(1); image(im);
mphimage2geom(im, 30,'minarea',2);

mphinputmatrix
Add a matrix system for linear solvers.

SYNTAX
mphinputmatrix(model,str,soltag,soltypetag)

DESCRIPTION
mphinputmatrix(model,str,soltag,soltypetag) adds the system matrices and
vectors stored in the MATLAB® structure str to the model. The system matrices is
associated to the linear solver configuration defined with the tag soltag and solved
with the solver defined with the tag soltypetag.

soltypetag can only be one of the following solver type: Stationary, Eigenvalue,
Time.

A valid structure for a stationary solver includes the following fields:

TABLE 6-11: PROPERTY/VALUE PAIRS FOR THE MPHINT2 COMMAND.

FIELD NAME DESCRIPTION

K Stiffness matrix
L Load vector

294 | CHAPTER 6: COMMAND REFERENCE

TABLE 6-11: PROPERTY/VALUE PAIRS FOR THE MPHINT2 COMMAND.

FIELD NAME DESCRIPTION

M Constraint vector
N Constraint Jacobian
A valid structure for a time-dependent/ eigenvalue solver includes the following fields:
TABLE 6-12: PROPERTY/VALUE PAIRS FOR THE MPHINT2 COMMAND.

FIELD NAME DESCRIPTION

K Stiffness matrix
L Load vector
M Constraint vector
N Constraint Jacobian
D Damping matrix
E Mass matrix

There is also the possibility to include the constraint force Jacobian vector NF.

Once the matrix system is loaded in the model, the solver configuration is set ready to
run.

Note: The system matrices are not stored in the model when it is saved as an
MPH-file or loaded into the COMSOL Desktop.

EXAMPLE
Create a model with a square geometry

model = ModelUtil.create('Model');
comp = model.component.create('comp1', true);
geom = comp.geom.create('geom1', 2);
geom.create('sq1', 'Square');
geom.run;

Add an Equation General Form physics interface

g = comp.physics.create('g', 'GeneralFormPDE', 'geom1');

g.prop('ShapeProperty').set('order', 1)
g.prop('ShapeProperty').set('boundaryFlux', false);
cons = g.create('cons1', 'Constraint', 1).set('R', 'u');
cons.selection.set([1 2]);

Create a mapped mesh

COMMANDS GROUPED BY FUNCTION | 295

 map = comp.mesh.create('mesh1').create('map1', 'Map');
map.create('dis1', 'Distribution').set('numelem', 2);
map.feature('dis1').selection.set([1 2]);

Set-up the study and the solver configuration for a stationary problem:

std = model.study.create('std1');
std.create('stat', 'Stationary');
sol = model.sol.create('sol1');
sol.study('std1');
sol.feature.create('st1', 'StudyStep').set('studystep', 'stat');
sol.feature.create('v1', 'Variables');
sol.feature.create('s1', 'Stationary');

Extract the linear stationary matrix system in MATLAB:

str = mphmatrix(model,'sol1','out',{'K','L','M','N'},...
'initmethod','sol','initsol','zero');

Change the linear system by scaling the stiffness matrix:

str.K = str.K*0.5;

Insert the system matrix back to the model:

mphinputmatrix(model,str,'sol1','s1')

Run the solver configuration:

model.sol('sol1').runAll;

SEE ALSO
mphmatrix, mphxmeshinfo

mphint2
Perform integration of expressions.

SYNTAX
[v1,...,v2] = mphint2(model,{e1,...,en},edim,...)
[v1,...,v2,unit] = mphint2(model,{e1,...,en},edim,...)

DESCRIPTION
[v1,...,vn] = mphint2(model,{e1,...,en},...) evaluates the integrals of the
string expressions e1,...,en and returns the result in N matrices v1,...,vn with M
rows and P columns. M is the number of inner solution and P the number of outer
solution used for the evaluation. edim defines the element dimension, as a string: line,
surface, volume or as an integer value.

296 | CHAPTER 6: COMMAND REFERENCE

[v1,...,vn] = mphint2(model,{e1,...,en},...) also returns the units of the
integral in a 1xN cell array.

The function mphint2 accepts the following property/value pairs:

TABLE 6-13: PROPERTY/VALUE PAIRS FOR THE MPHINT2 COMMAND.

PROPERTY PROPERTY VALUE DEFAULT DESCRIPTION

dataseries none | average | none Data series operation

integral |
maximum | rms |
stddev | variance
dataset String Active solution Dataset tag
dataset
intorder Positive integer 4 Integration order
intsurface on | off off Compute surface integral
intvolume on | off off Compute volume integral
matrix on | off on Returns data as a matrix or
as a cell
method auto | auto Integration method
integration |
summation
outersolnum Positive integer | all 1 Solution number for
| end parametric sweep
selection Integer vector | all Selection list or named
string | all selection
solnum Integer vector | end all Solution for evaluation
| all
squeeze on | off on Squeeze singleton
dimensions
t Double array Time for evaluation

The property dataset controls which dataset is used for the evaluation. Datasets
contain or refer to the source of data for postprocessing purposes. Evaluation is
supported only on Solution datasets.

The expressions e1,...,en are integrated for one or several solutions. Each solution
generates an additional column in the returned matrix. The properties solnum and t
control which solutions are used for the integrations. The solnum property is available
when the dataset has multiple solutions — for example, in the case of parametric,
eigenfrequency, or time-dependent solutions. The t property is available only for
time-dependent problems. If solnum is provided, the solutions indicated by the indices

COMMANDS GROUPED BY FUNCTION | 297

 provided with the solnum property are used. If t is provided, solutions are
interpolated. If neither solnum nor t is provided, all solutions are evaluated.

For time-dependent problems, the variable t can be used in the expressions ei. The
value of t is the interpolation time when the property t is provided, and the time for
the solution, when solnum is used. Similarly, lambda and the parameter are available
as eigenvalues for eigenvalue problems and as parameter values for parametric
problems, respectively.

The unit property defines the unit of the integral, if a inconsistent unit is entered, the
default unit is used. In case of multiple expression, if the unit property is defined with
a string, the same unit is used for both expressions. To use different units, set the
property with a cell array. In case of inconsistent unit definition, the default unit is used
instead.

Solnum is used to select the solution number when a parametric, eigenvalue, or

time-dependent solver has been used.

Outersolnum is used to select the outer solution number when a parametric sweep
has been used in the study step node.

EXAMPLE
Integrate the normal heat flux across all boundaries:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
std.feature.create('stat','Stationary');
std.run;

[Q, unit] = mphint2(model,'ht.ntflux','surface');

Integrate the normal heat flux across all exterior boundaries

[Q, unit] = mphint2(model,'ht.ntflux','surface',...

'selection',[1:5,7:12]);

SEE ALSO
mpheval, mphevalglobalmatrix, mphevalpoint, mphevalpointmatrix, mphint2,
mphinterp, mphparticle, mphray

mphinterp
Evaluate expressions in arbitrary points or datasets.

298 | CHAPTER 6: COMMAND REFERENCE

SYNTAX
[v1,...,vn] = mphinterp(model,{e1,...,en},'coord',coord,...)
[v1,...,vn] = mphinterp(model,{e1,...,en},'dataset',dsettag,...)
[v1,...,vn,unit] = mphinterp(model,{e1,...,en},...)

DESCRIPTION
[v1,...,vn] = mphinterp(model,{e1,...,en},'coord',coord,...) evaluates
expressions e1,...en at the coordinates specified in the double matrix coord.
Evaluation is supported only on Solution datasets.

[v1,...,vn] = mphinterp(model,{e1,...,en},'dataset',dsettag,...)
evaluates expressions e1,...en on the specified dataset dsettag. In this case the
dataset needs to be of a type that defines an interpolation in itself, such as cut planes,
revolve, and so forth.

[v1,...,vn,unit] = mphinterp(model,{e1,...,en},...) returns in addition

the unit of the expressions.

The function mphinterp accepts the following property/value pairs:

TABLE 6-14: PROPERTY/VALUE PAIRS FOR THE MPHINTERP COMMAND.

PROPERTY PROPERTY VALUE DEFAULT DESCRIPTION

complexfun off | on on Use complex-valued functions

with real input
complexout off | on on Return complex values
coord Double array Coordinates for evaluation
coorderr off | on off Give an error message if all
coordinates are outside the
geometry
dataset String Auto Dataset tag
differential off | on on Whether the expression should
be linearized at the linearization
point. Applicable only if
evalmethod is harmonic
edim point | edge | Geometry Element dimension for
boundary | space evaluation
domain | 0 | 1 | 2 dimension
|3

COMMANDS GROUPED BY FUNCTION | 299

 TABLE 6-14: PROPERTY/VALUE PAIRS FOR THE MPHINTERP COMMAND.

PROPERTY PROPERTY VALUE DEFAULT DESCRIPTION

evalmethod linpoint | harmonic Applicable only for solutions

harmonic | with a stored linearization point.
lintotal | Controls if the linearization
lintotalavg | point, the perturbation, or a
lintotalrms |
combination should be used
lintotalpeak
when evaluating the expression.
ext Double between 0 0.1 Extrapolation distance: How
and 1 much outside the mesh that the
interpolation searches. The scale
is in terms of the local element
size.
matherr off | on off Error for undefined operations
or expressions
outersolnum Positive integer | 1 Solution number for parametric
all | end sweep
phase Scalar 0 Phase angle in degrees
recover off | ppr | off Accurate derivative recovery
pprint
selection Positive integer all Selection list
array | all
solnum Positive integer all Inner solutions for evaluation
array | all | end
t Double array Time for evaluation
unit String | Cell array Unit to use for the evaluation
The columns of the matrix coord are the coordinates for the evaluation points. If the
number of rows in coord equals the space dimension, then coord are global
coordinates, and the property edim determines the dimension in which the expressions
are evaluated. For instance, edim='boundary' means that the expressions are
evaluated on boundaries in a 3D model. If edim is less than the space dimension, then
the points in coord are projected onto the closest point on a domain of dimension
edim. If, in addition, the property selection is given, then the closest point on
domain number selection in dimension edim is used.

If the number of rows in coord is less than the space dimension, then these coordinates
are parameter values on a geometry face or edge. In that case, the domain number for
that face or edge must be specified with the property selection.

300 | CHAPTER 6: COMMAND REFERENCE

The expressions that are evaluated can be expressions involving variables, in particular
physics interface variables.

The matrices v1,...,vn are of the size k-by-size(coord,2), where k is the number of
solutions for which the evaluation is carried out, see below. The value of expression ei
for solution number j in evaluation point coord(:,m) is vi(j,m).

The vector pe contains the indices m for the evaluation points code(:,m) that are
outside the mesh, or, if a domain is specified, are outside that domain.

The property Data controls which dataset is used for the evaluation. Datasets contain
or refer to the source of data for postprocessing purposes. Evaluation is supported only
on Solution datasets. The active solution dataset is used by default.

The property edim decides which elements to evaluate on. Evaluation takes place only
on elements with space dimension edim. If not specified, edim equal to the space
dimension of the geometry is used. The setting is specified as one of the following
strings 'point', 'edge', 'boundary' or 'domain'. In previous versions it was only
possible to specify edim as a number. For example, in a 3D model, if evaluation is done
on edges (1D elements), edim is 1. Similarly, for boundary evaluation (2D elements),
edim is 2, and for domain evaluation (3D elements), edim is 3 (default in 3D).

The property evalmethod decides which solution to use in presence of linearization

point. Set the property value to harmonic to harmonic perturbation analysis,
linpoint evaluates the expression by taking the values of any dependent variables
from the linearization point of the solution, lintotal evaluates the expression by
adding the linearization point and the harmonic perturbation and taking the real part
of this sum. lintotalavg (lintotalrms and lintotalpeak) do the same as with
lintotal and then averaging (taking the RMS and taking the maximum respectively)
over all phases of the harmonic perturbation. When harmonic is selected, the property
differential specify to evaluate the differential of the expression with respect to the
perturbation at the linearization point (on) or to evaluates the expression by taking the
values of any dependent variables from the harmonic perturbation part of the solution
(off).

Use Recover to recover fields using polynomial-preserving recovery. This techniques

recover fields with derivatives such as stresses or fluxes with a higher theoretical
convergence than smoothing. Recovery is expensive so it is turned off by default. The
value pprint means that recovery is performed inside domains. The value ppr means
that recovery is also applied on all domain boundaries.

COMMANDS GROUPED BY FUNCTION | 301

 The property Refine constructs evaluation points by making a regular refinements of
each element. Each mesh edge is divided into Refine equal parts.

When the property phase is used, the solution vector is multiplied with
exp(i*phase) before evaluating the expression.

The expressions e1,...,en are evaluated for one or several solutions. Each solution
generates an additional row in the data fields of the post data output structure. The
properties solnum and t control which solutions are used for the evaluations. The
solnum property is available when the dataset has multiple solutions — for example, in
the case of parametric, eigenfrequency, or time-dependent solutions. The t property is
available only for time-dependent problems. If solnum is provided, the solutions
indicated by the indices provided with the solnum property are used. If t is provided,
solutions are interpolated. If neither solnum nor t is provided, all solutions are
evaluated.

For time-dependent problems, the variable t can be used in the expressions ei. The
value of t is the interpolation time when the property t is provided, and the time for
the solution, when solnum is used. Similarly, lambda and the parameter are available
as eigenvalues for eigenvalue problems and as parameter values for parametric
problems, respectively.

In case of multiple expression, if the unit property is defined with a string, the same
unit is used for both expressions. To use different units, set the property with a cell
array. In case of inconsistent unit definition, the default unit is used instead.

The property solnum is used to select the solution number when a parametric,
eigenvalue, or time-dependent solver has been used.

The property outersolnum is used to select the outer solution number when a
parametric sweep has been used in the study step node.

EXAMPLE
Evaluate the temperature at given coordinates:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
std.feature.create('stat','Stationary');
std.run;

coord = [0,0,1e-2;0,0,1e-2;0,1e-2,1e-2];
T = mphinterp(model,'T','coord',coord);

Evaluate both the temperature and the heat flux magnitude:

302 | CHAPTER 6: COMMAND REFERENCE

 [T,tfluxMag] = mphinterp(model,{'T','ht.tfluxMag'},...
'coord',coord);

Evaluate the temperature field on a structure grid:

x0 = [0,1e-2,2.5e-2,5e-2]; y0 = x0; z0 = [5e-3,1e-2,1.1e-2];

[x,y,z] = meshgrid(x0,y0,z0); xx = [x(:),y(:),z(:)]';
T = mphinterp(model,'T','coord',xx);

Evaluate the temperature on boundary 7 using global coordinates:

x0 = [0,5e-3,1e-2]; y0 = x0; z0 = [1.1e-2];

[x,y,z] = meshgrid(x0,y0,z0); xx = [x(:),y(:),z(:)]';
T = mphinterp(model,'T','coord',xx,'edim','boundary',...
'selection',7);

Evaluate the temperature and evaluation point global coordinates on boundary 7 using
local coordinates:

s10 = [0,0.25,0.5]; s20 = [0,0.25,0.5];

[s1,s2] = meshgrid(s10,s20); ss = [s1(:),s2(:)]';
[x,y,z,T] = mphinterp(model,{'x','y','z','T'},'coord',ss,...
'edim','boundary','selection',7);

Modify the extrapolation distance for point coordinates outside of the geometry:

coord = [5e-2;5e-2;1.1e-2];
T = mphinterp(model,'T','coord',coord)
T = mphinterp(model,'T','coord',coord,'ext',0.5);

Extract data using a cut line dataset. First create the cut line dataset, then evaluate the
temperature field along the line:

cln = model.result.dataset.create('cln', 'CutLine3D');

cln.setIndex('genpoints','1e-2',1,0);
cln.setIndex('genpoints','1e-2',0,2);
cln.setIndex('genpoints','5e-2',1,0);
T = mphinterp(model,'T','dataset','cln');

Evaluation including several solution

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
param = std.feature.create('param','Parametric');
time = std.feature.create('time','Transient');
time.set('tlist', 'range(0,1,25)');
param.setIndex('pname','power',0);
param.setIndex('plistarr','30 60 90',0);
std.run;

Evaluate the temperature at every time step computed with power set to 30:

coord = [0 0 1e-2;0 0 1e-2;0 1e-2 1e-2];

COMMANDS GROUPED BY FUNCTION | 303

 T = mphinterp(model,'T','coord',coord,'dataset','dset2');

Evaluate the temperature at the fifth time step:

T = mphinterp(model,'T','coord',coord,'dataset','dset2',...
'solnum',5);

Evaluate the temperature at 10.5 sec:

T = mphinterp(model,'T','coord',coord,'dataset','dset2',...
't',10.5);

Evaluate the temperature at every time step computed with power set to 90:

T = mphinterp(model,'T','coord',coord,'dataset','dset2',...
'outersolnum',3)

SEE ALSO
mpheval, mphevalglobalmatrix, mphevalpoint, mphevalpointmatrix, mphint2,
mphint2, mphparticle, mphray

mphinterpolationfile
Save data in files readable by the Interpolation feature.

SYNTAX
mphinterpolationfile(filename,type,data)
mphinterpolationfile(filename,type,data,xdata)
mphinterpolationfile(filename,type,data,xdata,ydata)

DESCRIPTION
mphinterpolationfile(filename,type,data) saves the NxM matrix data into
the text file filename with the format type. The interpolation coordinates are vectors
with values from 1 to N and 1 to M.

mphinterpolationfile(filename,type,data,xdata) saves the vector data and

the interpolation coordinate xdata into the text file filename with the format type.

mphinterpolationfile(filename,type,data,xdata,ydata) saves the matrix

data and the interpolation coordinate vectors xdata and ydata into the text file
filename with the format type.

type can be either 'grid', 'sectionwise' or 'spreadsheet'.

EXAMPLE
Create random 10x10 interpolation data to file using grid format:

304 | CHAPTER 6: COMMAND REFERENCE

 data = cumsum(0.1*randn(size(10)));
mphinterpolationfile('datagrid.txt','grid',data);

1D interpolation data to file using spreadsheet format:

t = 0:0.05:2*pi;
z = sin(cos(t)*4)+sin(51*t)*0.05;
mphinterpolationfile('dataspread.txt','spreadsheet',z,t);

2D interpolation data to file using sectionwise format:

z = magic(9); x = 1:9; y = 0:8;

mphinterpolationfile('datasection.txt','sectionwise',z,x,y);

SEE ALSO
mphquad2tri, mphreadstl, mphsurf, mphwritestl

mphlaunch
Launch COMSOL Desktop, connect it to the running server, and import an
application.

SYNTAX
mphlaunch
mphlaunch(model)
mphlaunch ModelTag
mphlaunch(..., timeout)

DESCRIPTION
mphlaunch launches a COMSOL Multiphysics Client and connect it to the same
server as MATLAB® is connected to. Then it imports the model on the server into the
COMSOL Multiphysics Client.

mphlaunch(model) does the same as above, but uses the model argument to select
which model is imported.

mphlaunch ModelTag uses the model with the tag 'ModelTag' in the server to be
imported. This can also be done using the syntax: mphlaunch('ModelTag')

mphlaunch(..., tms) uses the timeout tms (in milliseconds) to force MATLAB to
wait until the COMSOL server is free again. The default timeout value is 500. A
negative value results in no timeout.

EXAMPLE
Load the file model_tutorial_llmatlab.mph:

COMMANDS GROUPED BY FUNCTION | 305

 model = mphopen('model_tutorial_llmatlab');

Launch a COMSOL Multiphysics Client, connect it with the running server, import
the model defined as model:, and set a timeout of 1 s:

mphlaunch(model,1000);

mphload
Load a COMSOL Multiphysics model MPH-file.

SYNTAX
model = mphload(filename)
model = mphload(filename, mtag)
model = mphload(filename, mtag, '-history')
model = mphload(filename, mtag, pwd)
[model, filename] = mphload(filename, ...)

DESCRIPTION
model = mphload(filename) loads a COMSOL model object saved with the name
filename and assigns the default tag Model in the COMSOL server. If a model with
tag Model already exists and is also open in a COMSOL Multiphysics client, the loaded
model an index number is appended to the tag, for instance Model1. The model object
is accessible at the MATLAB prompt using the variable model.

model = mphload(filename, mtag) loads a COMSOL model object and assigns

the tag mtag in the COMSOL server.

model = mphload(filename, mtag, '-history') turns on model history

recording.

model = mphload(filename, mtag, pwd) loads the COMSOL model object saved
with the name filename protected with the password pwd.

model = mphload(mtag) link the model already loaded on the COMSOL server with
the tag mtag. The model object is accessible at the MATLAB prompt using the variable
model.

[model, filenameloaded] = mphload(filename, ...) also returns the full file

name filenameloaded of the file that was loaded.

The model tag mtag and the password pwd are defined as string.

306 | CHAPTER 6: COMMAND REFERENCE

If the model tag is the same as a model that is currently in the COMSOL server the
loaded model overwrites the existing one.

Note that MATLAB® searches for the model on the MATLAB path if an absolute path
is not supplied.

mphload turns off the model history recording by default, unless the property
'-history' is used.

The extension mph can be omitted.

mphload does not look for lock file when opening a model in the COMSOL server.

EXAMPLE
Load the file model_tutorial_llmatlab.mph:

model = mphload('model_tutorial_llmatlab');

Load the file model_tutorial_llmatlab.mph and set the model name in the
COMSOL server to Model2:

model = mphload('model_tutorial_llmatlab','Model2');

Load model_tutorial_llmatlab.mph and return the filename:

[model, filename] = mphload('model_tutorial_llmatlab');

SEE ALSO
mphopen, mphsave

mphmatrix
Get model matrices.

SYNTAX
str = mphmatrix(model,soltag,'Out',...)

DESCRIPTION
®
str = mphmatrix(model,soltag,'Out',{'A'},...) returns a MATLAB
structure str containing the matrix A assembled using the solver node soltag and
accessible as str.A, A being taken from the Out property list.

str = mphmatrix(model,soltag,fname,'Out',{'A','B',...}) returns a

MATLAB structure str containing the matrices A, B, … assembled using the solver

COMMANDS GROUPED BY FUNCTION | 307

 node solname and accessible as str.A and str.B, A and B being taken from the Out
property list.

The function mphmatrix accepts the following property/value pairs:

TABLE 6-15: PROPERTY/VALUE PAIRS FOR THE MPHMATRIX COMMAND

PROPERTY EXPRESSION DEFAULT DESCRIPTION

complexfun on | off off Use complex-valued

functions with real input
eigname String lambda Eigenvalue name
eigref Double 0 Value of eigenvalue
linearization point
extractafter Solution feature First dependent Specify at which point to
tag variable node extract the matrices
initmethod init | sol init Use linearization point
initsol String | zero Active solver Solution to use for
tag linearization
matherr on | off on Error for undefined
operations
nullfun flnullorth | auto Null-space function
flspnull |
flexplicit |
auto
out Cell array of List of matrices to assemble
strings
rowscale on | off on Row equilibration
solnum Positive integer | auto Solution number
auto
study Study tag {First Study to use with
study} initmethod
symmetry on | off | auto Symmetric matrices
hermitian |
auto

The following values are valid for the out property:

Property/Value Pairs for the property out.

PROPERTY EXPRESSION DESCRIPTION

out K Stiffness matrix

L Load vector
M Constraint vector

308 | CHAPTER 6: COMMAND REFERENCE

Property/Value Pairs for the property out.
PROPERTY EXPRESSION DESCRIPTION

N Constraint Jacobian
D Damping matrix
E Mass matrix
NF Constraint force Jacobian
NP Optimization constraint Jacobian (*)
MP Optimization constraint vector (*)
MLB Lower bound constraint vector (*)
MUB Upper bound constraint vector (*)
Kc Eliminated stiffness matrix
Lc Eliminated load vector
Dc Eliminated damping matrix
Ec Eliminated mass matrix
Null Constraint null-space basis
Nullf Constraint force null-space matrix
ud Particular solution ud
uscale Scale vector
(*) Requires the Optimization Module.

Note that the assembly of the eliminated matrices uses the current solution vector as
scaling method. To get the unscaled eliminated system matrices, it is required to set
the scaling method to 'none' in the variables step of the solver configuration node.

The load vector is assembled using the current solution available as linearization point
unless the initmethod property is provided. In case of the presence of a solver step
node in the solver sequence, the load vector correspond then to the residual of the
problem.

The function mphmatrix does not solve the problem as the assembly is performed
before the solver node in the solution sequence. You can specify the solution feature
node after which to assemble the system matrices with the property extractafter.
This is useful if you need to compute the solution before extracting the matrices or if
you have a solution sequence using different solver sequences and you want to extract
the matrices for a specific one.

The Advanced section in the COMSOL Multiphysics Reference Manual, describes the
functionality corresponding to the properties complexfun, nullfun, and rowscale.

COMMANDS GROUPED BY FUNCTION | 309

 Use the property symmetric to assemble the model matrix system as
symmetric/Hermitian, or you can use the automatic feature to find out (see Advanced
in the COMSOL Multiphysics Reference Manual).

EXAMPLE
Evaluate the system matrices of a stationary problem

model = mphopen('model_tutorial_llmatlab');
model.mesh('mesh1').autoMeshSize(8);
std = model.study.create('std1');
std.feature.create('stat', 'Stationary');
std.run;

Get the stationary matrix system, use the initial solution as linearization point:

str = mphmatrix(model,'sol1','out',{'K','L','M','N'},...
'initmethod','init');

Display the sparsity of the stiffness matrix and the constraint Jacobian and compute the
total load applied in the matrix system:

subplot(2,1,1); spy(str.K);subplot(2,1,2);spy(str.N)
Q = sum(str.L)

Get the eliminated matrix system, use the initial solution as linearization point:

str = mphmatrix(model,'sol1','out',{'Kc'},'initmethod','init');

Compare the sparsity between the eliminated and non-eliminated stiffness matrix:

subplot(2,1,1); hold on; spy(str.Kc,'r')

Evaluate the eliminated load vector using the current solution as linearization point:

str = mphmatrix(model,'sol1','out',{'Lc'},'initmethod','sol');

Evaluate the system matrices of a dynamic problem

model = mphopen('model_tutorial_llmatlab');
model.mesh('mesh1').autoMeshSize(8);
std = model.study.create('std1');
time = std.feature.create('time', 'Transient');
time.set('tlist', 'range(0,1,25)');
model.param.set('timestep', '1[s]');
std.run;

Get the dynamic matrix system:

str = mphmatrix(model,'sol1','out',{'E','D','K','L','M','N'});

Display the sparsity of the mass and stiffness matrices:

310 | CHAPTER 6: COMMAND REFERENCE

 subplot(1,2,1); spy(str.D); subplot(1,2,2); spy(str.K);

Get the eliminated dynamic matrix system:

str = mphmatrix(model,'sol1','out',{'Ec','Dc','Kc','Lc','M','N'});

Assemble the Jacobian using solution number 15 as linearization point. First run the
model to get available linearization point list:

std.run;
str = mphmatrix(model,'sol1','out',{'K'},...
'initmethod','sol','initsol','sol1','solnum',15);

Assemble the Jacobian using the zero vector as linearization point:

str = mphmatrix(model,'sol1','out',{'K'},...
'initmethod','sol','initsol','zero');

SEE ALSO
mphstate, mphxmeshinfo, mphinputmatrix

mphmax
Perform a maximum of expressions.

SYNTAX
[v1,...,vn] = mphmax(model,{e1,...,en},edim,...)
[v1,...,vn,unit] = mphmax(model,{e1,...,en},edim,...)

DESCRIPTION
[v1,...,vn] = mphmax(model,{e1,...,en},edim,...) evaluates the maximum
of the string expressions e1,...,en and returns the result in N matrices v1,...,vn
with M rows and P columns. M is the number of inner solution and P the number of
outer solution used for the evaluation. edim defines the element dimension: line,
surface, volume or as an integer value.

[v1,...,vn] = mphmax(model,{e1,...,en},edim,...) also returns the units of

the maximum in a 1xN cell array.

COMMANDS GROUPED BY FUNCTION | 311

 The function mphmax accepts the following property/value pairs:
TABLE 6-16: PROPERTY/VALUE PAIRS FOR THE MPHMAX COMMAND.

PROPERTY PROPERTY VALUE DEFAULT DESCRIPTION

dataseries none | average | none Data series operation

integral | maximum |
minimum | rms |
stddev | variance
dataset String Active Dataset tag
solution
dataset
matrix on | off on Returns data as a matrix or
as a cell
outersolnum Positive integer array | 1 Solution number for
all | end parametric sweep
selection Integer vector | string | all Selection list or named
all selection
solnum Integer vector | end | all Solution for evaluation
all
squeeze on | off on Squeeze singleton
dimensions
t Double array Time for evaluation

The property dataset controls which dataset is used for the evaluation. Datasets
contain or refer to the source of data for postprocessing purposes. Evaluation is
supported only on Solution datasets.

The maximum expressions e1,...,en is evaluated for one or several solutions. Each
solution generates an additional column in the returned matrix. The properties solnum
and t control which solutions are used for the evaluation. The solnum property is
available when the dataset has multiple solutions — for example, in the case of
parametric, eigenfrequency, or time-dependent solutions. The t property is available
only for time-dependent problems. If solnum is provided, the solutions indicated by
the indices provided with the solnum property are used. If t is provided, solutions are
interpolated. If neither solnum nor t is provided, all solutions are evaluated.

The property solnum is used to select the solution number when a parametric,
eigenvalue, or time-dependent solver has been used.

The property outersolnum is used to select the outer solution number when a
parametric sweep has been used in the study step node.

312 | CHAPTER 6: COMMAND REFERENCE

If the matrix property is set to off the output is cell arrays of length P containing cell
arrays of length M.

EXAMPLE
Evaluate the maximum temperature in the model domain:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
std.feature.create('stat','Stationary');
std.run;

maxT = mphmax(model,'T','volume');

Evaluate the maximum temperature on boundary 9:

maxT = mphmax(model,'T','surface','selection',9);

Evaluate maximum of expression using several solution:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
param = std.feature.create('param','Parametric');
time = std.feature.create('time','Transient');
time.set('tlist', 'range(0,1,25)');
param.setIndex('pname','power',0);
param.setIndex('plistarr','30 60 90',0);
std.run;

Evaluate the maximum of the temperature at every time step computed with power set
to 30:

maxT = mphmax(model,'T','volume','dataset','dset2');

Evaluate the maximum of the temperature at the fifth time step:

maxT = mphmax(model,'T','volume','dataset','dset2',...
'solnum',5);

Evaluate the maximum of the temperature at 10.5 sec and 15.2 sec:

maxT = mphmax(model,'T','volume','dataset','dset2',...
't',[10.5,15.2]);

Evaluate the maximum of the temperature at every time step computed with power set
to 90:

maxT = mphmax(model,'T','volume','dataset','dset2',....
'outersolnum',3);

SEE ALSO
mphmean, mphmin

COMMANDS GROUPED BY FUNCTION | 313

 mphmean
Perform a mean of expressions.

SYNTAX
[v1,...,vn] = mphmean(model,{e1,...,en},edim,...)
[v1,...,vn,unit] = mphmean(model,{e1,...,en},edim,...)

DESCRIPTION
[v1,...,vn] = mphmean(model,{e1,...,en},edim,...) evaluates the means of
the string expressions e1,...,en and returns the result in N matrices v1,...,vn with
M rows and P columns. M is the number of inner solution and P the number of outer
solution used for the evaluation. edim defines the element dimension: line, surface,
volume or as an integer value.

[v1,...,vn] = mphmean(model,{e1,...,en},edim,...) also returns the units of

the maximum in a 1xN cell array.

The function mphmean accepts the following property/value pairs:

TABLE 6-17: PROPERTY/VALUE PAIRS FOR THE MPHMEAN COMMAND.

PROPERTY PROPERTY VALUE DEFAULT DESCRIPTION

dataseries none | average | none Data series operation

integral | maximum |
minimum | rms |
stddev | variance
dataset String Active Dataset tag
solution
dataset
intorder Positive integer 4 Integration order
matrix off | on on Returns data as a matrix or
as a cell
method auto | integration | auto Integration method
summation
outersolnum Positive integer array | 1 Solution number for
all | end parametric sweep
selection Integer vector | string | all Selection list or named
all selection
solnum Integer vector | end | all Solution for evaluation
all

314 | CHAPTER 6: COMMAND REFERENCE

TABLE 6-17: PROPERTY/VALUE PAIRS FOR THE MPHMEAN COMMAND.

PROPERTY PROPERTY VALUE DEFAULT DESCRIPTION

squeeze on | off on Squeeze singleton

dimensions
t Double array Time for evaluation
The property dataset controls which dataset is used for the evaluation. Datasets
contain or refer to the source of data for postprocessing purposes. Evaluation is
supported only on Solution datasets.

The mean of expressions e1,...,en is evaluated for one or several solutions. Each
solution generates an additional column in the returned matrix. The properties solnum
and t control which solutions are used for the evaluation. The solnum property is
available when the dataset has multiple solutions — for example, in the case of
parametric, eigenfrequency, or time-dependent solutions. The t property is available
only for time-dependent problems. If solnum is provided, the solutions indicated by
the indices provided with the solnum property are used. If t is provided, solutions are
interpolated. If neither solnum nor t is provided, all solutions are evaluated.

The property solnum is used to select the solution number when a parametric,
eigenvalue, or time-dependent solver has been used.

The property outersolnum is used to select the outer solution number when a
parametric sweep has been used in the study step node.

If the matrix property is set to off the output is cell arrays of length P containing cell
arrays of length M.

EXAMPLE
Evaluate the mean temperature in the model domain:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
std.feature.create('stat','Stationary');
std.run;

maxT = mphmean(model,'T','volume');

Evaluate the mean temperature on boundary 9:

maxT = mphmean(model,'T','surface','selection',9);

Evaluate mean of expression using several solution:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');

COMMANDS GROUPED BY FUNCTION | 315

 param = std.feature.create('param','Parametric');
time = std.feature.create('time','Transient');
time.set('tlist', 'range(0,1,25)');
param.setIndex('pname','power',0);
param.setIndex('plistarr','30 60 90',0);
std.run;

Evaluate the mean of the temperature at every time step computed with power set to
30:

maxT = mphmean(model,'T','volume','dataset','dset2');

Evaluate the mean of the temperature at the fifth time step:

maxT = mphmean(model,'T','volume','dataset','dset2',...
'solnum',5);

Evaluate the mean of the temperature at 10.5 sec and 15.2 sec:

maxT = mphmean(model,'T','volume','dataset','dset2',...
't',[10.5,15.2]);

Evaluate the mean of the temperature at every time step computed with power set to
90:

maxT = mphmean(model,'T','volume','dataset','dset2',....
'outersolnum',3);

SEE ALSO
mphmax, mphmin

mphmeasure
Measure entities in geometry

SYNTAX
[measure, misc] = mphmeasure(model,geomtag,entity,...)

DESCRIPTION
[measure, misc] = mphmeasure(model,geomtag,entity,...) measure the
entity with the type entity in the geometry defined with the tag geomtag. entity
can be one of 'point', 'edge', 'boundary', or 'domain'.

316 | CHAPTER 6: COMMAND REFERENCE

The function mphmeasure accepts the following property/value pairs:
TABLE 6-18: PROPERTY/VALUE PAIRS FOR THE MPHMEASURE COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

Build on | off | string on Build the geometry before

plotting
Objects Cell array Selection matrix
Selection Positive integer all Selection (finalized
array geometry)
Usefinal on | off on Measure on the finalized
geometry

EXAMPLE
Get the volume of all domains in the geometry:

model = mphopen('busbar');
vm = mphmeasure(model,'geom1','domain')

Get the volume and the surface area of all domains in the geometry:

[vm, am]= mphmeasure(model,'geom1','domain')

Get midpoint and distance between points 2 and 4:

[m, dist] = mphmeasure(model,'geom1','point','selection',[2,4])

Length of edges bounded by vertex 1, 29, 41 and 42:

[~, a] = mphgetadj(model,'geom1','edge','point',[1,29,41,42])
mphviewselection(model,'geom1',[1,29,41,42],'point')
hold on
mphviewselection(model,'geom1',a,'edge','edgecolorselected','b')
m = mphmeasure(model,'geom1','edge','selection',a)

mphmesh
Plot a mesh in a MATLAB® figure window.

SYNTAX
mphmesh(model)
mphmesh(model,meshtag,...)
pd=mphmesh(model,meshtag,...)

DESCRIPTION
mphmesh(model) plots the mesh case in a MATLAB figure.

COMMANDS GROUPED BY FUNCTION | 317

 mphmesh(model,meshtag,...) plots the mesh case meshtag in a MATLAB figure.
If there is only one mesh in the model the meshtag can be left empty.

The function mphmesh accepts the following property/value pairs:

TABLE 6-19: PROPERTY/VALUE PAIRS FOR THE MPHMESH COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

Parent Double Parent axis

Edgecolor Char k Edge color
Edgelabels on | off off Show edge labels
Edgelabelscolor Char k Color for edge labels
Edgemode on | off on Show edges
Facealpha Double 1 Set transparency value
Facelabels on | off off Show face labels
Facelabelscolor Char k Color for face labels
Facemode on | off on Show faces
Meshcolor Char | vector gray Color for face element
Vertexcolor Char | vector k Color for vertices
Vertexlabels on | off off Show vertex labels
Vertexlabelscolor Char | vector k Color for vertex labels
Vertexmode on | off off Show vertices
View String | 'auto' | View settings
''

EXAMPLE
Plot the mesh case

model = mphopen('model_tutorial_llmatlab');
model.component('comp1').mesh.run;
mphmesh(model)

Create a second mesh case with an “extra fine” default mesh settings and plot it:

mesh = model.component('comp1').create('mesh2');
mesh.autoMeshSize(2);
mesh.run;
mphmesh(model,'mesh2','meshcolor','r');

The mesh can be plotted with view settings applied. This results in a mesh with grid,
axes labels, lights, hiding etc. applied to the plot. Usually it is sufficient to use the auto
setting, but any valid view can be applied:

318 | CHAPTER 6: COMMAND REFERENCE

 mphmesh(model, 'mesh1', 'view', 'auto')

Plot data can be returned from mphmesh. This can be used to create the plot later or
to extract information used to create the plot for further analysis

pd = mphmesh(model, 'mesh1');
mphplot(pd)

SEE ALSO
mphgeom, mphmeshstats, mphplot

mphmeshstats
Return mesh statistics and mesh data information.

SYNTAX
stats = mphmeshstats(model)
stats = mphmeshstats(model, meshtag, ...)
[stats,data] = mphmeshstats(model, meshtag, ...)

DESCRIPTION
stats = mphmeshstats(model) returns mesh statistics of the model mesh case in the
structure str.

stats = mphmeshstats(model, meshtag, ...) returns mesh statistics of a mesh

case meshtag in the structure str.

[stats,data] = mphmeshstats(model, meshtag, ...) returns in addition the

mesh data information such as vertex coordinates and definitions of elements in the
structure data.

The function mphmeshstats accepts the following property/value pairs:

TABLE 6-20: PROPERTY/VALUE PAIRS FOR THE MPHMESHSTATS COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

*
entity domain | Selected entity type
boundary |
edge | point
qualityhistogram Integer 20 Number of bins in the quality
distribution histogram

COMMANDS GROUPED BY FUNCTION | 319

 TABLE 6-20: PROPERTY/VALUE PAIRS FOR THE MPHMESHSTATS COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

qualitymeasure condition | volcircum Quality measure

growth |
maxangle |
skewness |
volcircum |
vollength
selection String | Selection tag or entity
Positive number
integer array
*
type ctx | edg | Restrict statistics to element
tri | quad | types. Can also be a cell
tet | pyr | array.
prism | hex
* Selection and Entity properties cannot be set if the data structure is returned.

The output structure stats contains the following fields:

TABLE 6-21: FIELDS IN THE STATS STRUCTURE

FIELD DESCRIPTION

meshtag Mesh case tag

geomtag Associated geometry tag
isactive Is the mesh node active
hasproblems Does the mesh have problems?
iscomplete Is the mesh built to completion?
sdim Space dimension
secondorderelements Does the mesh have second-order elements?
contributing Contributing physics and multiphysics
types Cell array with type names
numelem Vector with the number of elements for each type
qualitymeasure Quality measure
minquality Minimum quality
meanquality Mean quality
qualitydistr Quality distribution (vector)
minvolume Volume/area of the smallest element
maxvolume Volume/area of the largest element
volume Volume/area of the mesh

320 | CHAPTER 6: COMMAND REFERENCE

TABLE 6-21: FIELDS IN THE STATS STRUCTURE

FIELD DESCRIPTION

maxgrowthrate Maximum growth rate*

meangrowthrate Mean growth rate*
* Provides statistics for the entire selection regardless of the element type property.

The output structure data contains the following fields:

TABLE 6-22: FIELDS IN THE DATA STRUCTURE

FIELD DESCRIPTION

vertex Coordinates of mesh vertices

elem Cell array of definition of each element type
elementity Entity information for each element type

EXAMPLE
Get the mesh statistics:

model = mphopen('model_tutorial_llmatlab');
model.component('comp1').mesh.run;
stats = mphmeshstats(model)

Show the mesh quality distribution in a figure:

bar(linspace(0,1,20),stats.qualitydistr)

Get the mesh statistics and the mesh data:

[stats,data] = mphmeshstats(model);

Show the element vertices in a plot:

plot3(data.vertex(1,:), data.vertex(2,:), data.vertex(3,:), '.')

axis equal

Get the number of edge element:

numedgeelem = stats.numelem(strcmp(stats.types,'edg'))

SEE ALSO
mphmesh

mphmin
Perform a minimum of expressions.

COMMANDS GROUPED BY FUNCTION | 321

 SYNTAX
[v1,...,vn] = mphmin(model,{e1,...,en},edim,...)
[v1,...,vn,unit] = mphmin(model,{e1,...,en},edim,...)

DESCRIPTION
[v1,...,vn] = mphmin(model,{e1,...,en},edim,...) evaluates the minimum
of the string expressions e1,...,en and returns the result in N matrices v1,...,vn
with M rows and P columns. M is the number of inner solution and P the number of
outer solution used for the evaluation. edim defines the element dimension: line,
surface, volume or as an integer value.

[v1,...,vn] = mphmin(model,{e1,...,en},edim,...) also returns the units in

a 1xN cell array.

The function mphmin accepts the following property/value pairs:

TABLE 6-23: PROPERTY/VALUE PAIRS FOR THE MPHMIN COMMAND.

PROPERTY PROPERTY VALUE DEFAULT DESCRIPTION

dataseries none | average | none Data series operation

integral | maximum |
minimum | rms |
stddev | variance
dataset String Active Dataset tag
solution
dataset
matrix off | on on Returns data as a matrix or
as a cell
outersolnum Positive integer array | 1 Solution number for
all | end parametric sweep
selection Integer vector | string | all Selection list or named
all selection
solnum Integer vector | end | all Solution for evaluation
all
squeeze on | off on Squeeze singleton
dimensions
t Double array Time for evaluation

The property dataset controls which dataset is used for the evaluation. Datasets
contain or refer to the source of data for postprocessing purposes. Evaluation is
supported only on Solution datasets.

322 | CHAPTER 6: COMMAND REFERENCE

The mean of expressions e1,...,en is evaluated for one or several solutions. Each
solution generates an additional column in the returned matrix. The properties solnum
and t control which solutions are used for the evaluation. The solnum property is
available when the dataset has multiple solutions — for example, in the case of
parametric, eigenfrequency, or time-dependent solutions. The t property is available
only for time-dependent problems. If solnum is provided, the solutions indicated by
the indices provided with the solnum property are used. If t is provided, solutions are
interpolated. If neither solnum nor t is provided, all solutions are evaluated.

The property solnum is used to select the solution number when a parametric,
eigenvalue, or time-dependent solver has been used.

The property outersolnum is used to select the outer solution number when a
parametric sweep has been used in the study step node.

If the matrix property is set to off the output is cell arrays of length P containing cell
arrays of length M.

EXAMPLE
Evaluate the minimum temperature in the model domain:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
std.feature.create('stat','Stationary');
std.run;
maxT = mphmin(model,'T','volume');

Evaluate the minimum temperature on boundary 9:

maxT = mphmin(model,'T','surface','selection',9);

Evaluate minimum of expression using several solution:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
param = std.feature.create('param','Parametric');
time = std.feature.create('time','Transient');
time.set('tlist', 'range(0,1,25)');
param.setIndex('pname','power',0);
param.setIndex('plistarr','30 60 90',0);
std.run;

Evaluate the minimum of the temperature at every time step computed with power set
to 30:

maxT = mphmin(model,'T','volume','dataset','dset2');

COMMANDS GROUPED BY FUNCTION | 323

 Evaluate the minimum of the temperature at the fifth time step:

maxT = mphmin(model,'T','volume','dataset','dset2',...
'solnum',5);

Evaluate the minimum of the temperature at 10.5 sec and 15.2 sec:

maxT = mphmin(model,'T','volume','dataset','dset2',...
't',[10.5,15.2]);

Evaluate the minimum of the temperature at every time step computed with power set
to 90:

maxT = mphmin(model,'T','volume','dataset','dset2',....
'outersolnum',3);

SEE ALSO
mphmax, mphmean

mphmodel
Return tags for the nodes and subnodes in the COMSOL model object.

SYNTAX
mphmodel(model)
str = mphmodel(model,'-struct')

DESCRIPTION
mphmodel(model) returns the tags for the nodes and subnodes of the object model.

str = mphmodel(model,'-struct') returns the tags for the nodes and subnodes of
the object model as a MATLAB® structure str.

The function mphmodel can be used when navigating the model object and learning
about its structure. The mphmodel function is mainly designed for usage when working
on the command line and one needs to learn what nodes are placed under a particular
node.

EXAMPLE
Load the model busbar.mph and get the list of the nodes available under the root
node:

model = mphopen('busbar')
mphmodel(model)

Get the model information as a structure:

324 | CHAPTER 6: COMMAND REFERENCE

 res = mphmodel(model, '-struct')

SEE ALSO
mphgetexpressions, mphgetproperties, mphgetselection, mphnavigator,
mphsearch, mphshowerrors

mphnavigator
Graphical user interface (GUI) for viewing the COMSOL Multiphysics model object.

SYNTAX
mphnavigator
mphnavigator(modelv)

DESCRIPTION
mphnavigator opens the Model Object Navigator which is a graphical user interface
that can be used to navigate the model object and to view the properties and methods
of the nodes in the model tree.

The GUI requires that the COMSOL object is stored in a variable in the base
workspace (at the MATLAB® command prompt) with the name model.

COMMANDS GROUPED BY FUNCTION | 325

 mphnavigator(model) opens the model object defined with the name model in
Model Object Navigator.

EXAMPLE
Load busbar.mph from the Model Library:

mphopen busbar

Navigate the model object that is accessible with the variable model

mphnavigator

Load effective_diffusivity.mph from the Applications Libraries and set the

model object with the variable eff_diff:

eff_diff = mphopen('effective_diffusivity');

Navigate the model object that is accessible with the variable eff_diff

mphnavigator(eff_diff)

SEE ALSO
mphgetexpressions, mphgetproperties, mphgetselection, mphmodel,
mphsearch, mphshowerrors

326 | CHAPTER 6: COMMAND REFERENCE

mphopen
Graphical user interface (GUI) to open recent model files.

SYNTAX
mphopen
mphopen -dir dirpath
mphopen -clear
model = mphopen(filename)
model = mphopen(filename, mtag)
model = mphopen(filename, mtag, '-nostore')
model = mphopen(filename, mtag, '-history')
model = mphopen(filename, mtag, pwd)
[model, filenameloaded] = mphopen(filename,...)

DESCRIPTION
mphopen starts a GUI with the recent opened files list.

mphopen -dir dirpath starts a GUI with a list of the files in the specified directory
dirpath. If dirpath is not specified the working directory is taken by default.

mphopen -clear resets the recent opened files list.

model = mphopen(filename) loads a COMSOL model object saved with the name
filename and assigns the default tag Model in the COMSOL server. If a model with
tag Model already exists and is also open in a COMSOL Multiphysics client, the loaded
model an index number is appended to the tag, for instance Model1.

model = mphopen(filename, mtag) loads a COMSOL model object and assigns

the tag mtag in the COMSOL server.

model = mphopen(filename, mtag, '-nostore') does not update the recent

opened model list.

model = mphopen(filename, mtag, '-history') turns on history recording.

model = mphopen(filename, mtag, pwd) loads the COMSOL model in the file
protected with the password pwd.

[model, filenameloaded] = mphopen(filename, …) also returns the full file

name filenameloaded of the file that was loaded.

The model tag mtag and the password pwd are defined as string.

If the model tag is the same as a model that is currently in the COMSOL server the
loaded model overwrites the existing one.

COMMANDS GROUPED BY FUNCTION | 327

 Note that MATLAB® searches for the model on the MATLAB path if an absolute path
is not supplied.

mphopen turns off the model history recording by default, unless the property
'-history' is used.

The extension mph can be omitted.

mphopen does not look for a lock file when opening a model in the COMSOL server.

SEE ALSO
mphload, mphsave

mphparticle
Evaluate expressions on particle and ray trajectories.

SYNTAX
pd = mphparticle(model)
pd = mphparticle(model,'expr',{e1,...,en},...)

DESCRIPTION
mphparticle(model) returns particle position and particle velocity at all time steps
stored in the first particle dataset.

328 | CHAPTER 6: COMMAND REFERENCE

mphparticle(model,'expr',{e1,...,en},...) returns particle position, particle
velocity and expressions e1,..., en evaluated on particle trajectories.

The function mphparticle accepts the following property/value pairs:

TABLE 6-24: PROPERTY/VALUE PAIRS FOR THE MPHPARTICLE COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

expr String | Cell array Expressions to evaluate

dataonly on | off off Return only expression values
dataset String First Dataset tag
particle
dataset
t Double array Time for evaluation

The returned value pd is a structure with the following content

TABLE 6-25: FIELDS IN THE INFO STRUCTURE

FIELD CONTENT

v Velocity of the particles

p Position of the particles
d# Result of evaluation #
t Time for evaluation
expr Evaluated expressions
unit Unit of evaluations

Note: mphparticle only evaluates expressions using particle and ray datasets.

EXAMPLE
Load the model trapped_protons from the Applications Libraries:

model = mphopen('trapped_protons');

Extract the particle positions and particle velocities along the computed trajectories at
every time steps stored in the model:

pd = mphparticle(model)

Evaluate the mirror point latitude (Lm) and the particle equatorial pitch angle (Ea) at
t = 0.7 sec., extract only the data:

pd = mphparticle(model,'dataset','dset2',...

COMMANDS GROUPED BY FUNCTION | 329

 'expr',{'Lm','Ea*180/pi'},'t',0.7,'dataonly','on')

SEE ALSO
mpheval, mphevalpoint, mphint2, mphinterp, mphray

mphplot
Render a plot group in a figure window.

SYNTAX
mphplot(model)
mphplot(model,pgtag,...)
pd = mphplot(model,pgtag,...)
mphplot(pd,...)

DESCRIPTION
mphplot(model) opens a figure window and adds a menu where it is possible to
switch between all the different result plots in a model as well as any geometry and
mesh plots. A toolbar is added to the figure that allows the user to control the use of
views, lights, and camera settings.

mphplot(model,pgtag,...) renders the plot group tagged pgtag from the model
object model in a figure window in MATLAB®.

pd = mphplot(model,pgtag,...) also returns the plot data used in the MATLAB

figure in a cell array pd. pd contains ordinary MATLAB data and can later be used to
recreate the plot using mphplot. It is also possible to investigate and extract data from
the plot this way in order to create plots in other ways or to further analyze the data.
Set createplot to off if a plot should be not be created. Note that pd contains data
in single precision even though all calculations in COMSOL Multiphysics are carried
out in double precision.

mphplot(pd,...) makes a plot using the post data structure pd that is generated
using the function mpheval. Plots involving points, lines and surfaces are supported.

The function mphplot accepts the following property/value pairs:

TABLE 6-26: PROPERTY/VALUE PAIRS FOR THE MPHPLOT COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

colortable char rainbow Color table used for plotting post

data structure
createplot on | off on Create a plot

330 | CHAPTER 6: COMMAND REFERENCE

TABLE 6-26: PROPERTY/VALUE PAIRS FOR THE MPHPLOT COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

index Positive integer 1 Index of variable to use for

plotting post data structure
mesh on | off on Plot the mesh when using post
data structure
parent Double Set the parent axes
rangenum Positive integer none Color range bar (or legend) to
display
server on | off off Plot on server
view char | 'auto' | '' '' View settings tag

Note: The plot on server option requires that you start COMSOL with MATLAB in
graphics mode.

Only one color range bar and one legend bar is supported in a MATLAB figure. When
the option plot on server is active, all active color range bar are displayed.

The property createplot is useful when extracting plot data structure on machines
without a graphics display.

The data fields returned by mphplot are subject to change. The most important fields
are:

• p, the coordinates for each point that are used for creating lines or triangles.
• n, the normals in each point for the surfaces. These are not always available.
• t, contains the indices to columns in p of a simplex mesh, each column in t
representing a simplex.
• d, the data values for each point.
• rgb, the color values (red, green and blue) entities at each point.

EXAMPLE
Display the plot settings pg using a MATLAB figure

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
std.feature.create('stat','Stationary');
std.run;
model.result.dataset.create('mir', 'Mirror3D');

COMMANDS GROUPED BY FUNCTION | 331

 pg = model.result.create('pg', 'PlotGroup3D');
pg.set('data', 'mir');
surf1 = pg.feature.create('surf1', 'Surface');
surf1.set('colortable', 'Thermal');

mphplot(model,'pg')

Combine plot types on the same plot group:

surf2 = pg.feature.create('surf2', 'Surface');

surf2.set('data', 'dset1');
surf2.set('expr', 'ht.tfluxMag');

Display the plot group and the color range bar of the second plot type:

mphplot(model,'pg','rangenum',2)

Display the plot group on the server:

mphplot(model,'pg','server','on')

Display expression value evaluated using mpheval:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
std.feature.create('stat','Stationary');
std.run;

Extract temperature and total heat flux magnitude in domain 2:

pd = mpheval(model,{'T','ht.tfluxMag'},'selection',2);

Plot the temperature data using thermal color table:

mphplot(pd,'index',1,'colortable','Thermal','rangenum',1)

SEE ALSO
colortable, mpheval

mphquad2tri
Convert plot data quad mesh into simplex mesh.

SYNTAX
pdout = mphquad2tri(pdin)

DESCRIPTION
pdout = mphquad2tri(pdin) converts the plot data stored in the structure pdin into
the structure pdout using a simplex mesh.

332 | CHAPTER 6: COMMAND REFERENCE

The input and output structures, respectively pdin and pdout, are structures with
fields p, d, t, rgb and expr.

• The field p is a 2xN array containing the vertex coordinates.

• The field saved is a Nx1 array containing the data value at the vertices.
• The field t is an array containing the indices to columns in p of a quad mesh for
pdin and of a simplex mesh for pdout, each column in t representing respectively
a quad and a simplex.
• The field rgb is a Nx3 array containing the color model data as RGB values. This
field is optional.
• The field expr is a string containing the description of the data. This field is
optional.

EXAMPLE
Generate 3D surf data

[x,y] = meshgrid(-0.1:0.2:1.1,-0.4:0.2:0.4);
z = cumsum(0.1*randn(size(x)));

Create plotdata in a quad mesh:

pd = mphsurf(x,y,z);

Convert the plot data into triangle mesh:

pd = mphquad2tri(pd);
mphplot(pd)

SEE ALSO
mphsurf, mphreadstl, mphwritestl

mphray
Evaluate expressions on particle and ray trajectories.

SYNTAX
pd = mphray(model)
pd = mphray(model,'expr',{e1,...,en},...)

DESCRIPTION
pd = mphray(model) returns particle position and particle velocity at all time steps
stored in the first particle dataset.

COMMANDS GROUPED BY FUNCTION | 333

 pd = mphray(model,'expr',{e1,...,en},...) returns particle position, particle
velocity and expressions e1,...,en evaluated on particle or ray trajectories.

pd is a structure with fields p, v, d#, t, expr, and unit.

• The field p contains the position of the particles.

• The field v contains the velocity of the particles.
• The field d# contains the result of evaluation #.
• The field t contains the time for evaluation.
• The field expr contains the evaluated expressions.
• The field unit contains the unit of evaluations.

The function mphray accepts the following property/value pairs:

TABLE 6-27: PROPERTY/VALUE PAIRS FOR THE MPHRAY COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

dataonly on | off off Return only expression values

dataset String Dataset tag
expr String | Cell Expressions to evaluate
array
outersolnum Positive 1 Outer solution for evaluation
integer
solnum Positive all Inner solution for evaluation
integer array
| all | end
t Double array Time for evaluation
times on | off on Return times
velocities on | off on Return velocities

Note: mphray only evaluates expressions using particle and ray datasets.

SEE ALSO
mpheval, mphevalpoint, mphevalpoint, mphint2, mphinterp, mphparticle

mphreadstl
Read an STL file and return the data into a plot data structure.

334 | CHAPTER 6: COMMAND REFERENCE

SYNTAX
pd = mphreadstl(filename)

DESCRIPTION
pd = mphreadstl(filename) reads the STL file filename and returns the data into
the plot data structure pd.

pd is a structure with fields n, p, t, name, ref, rgb, expr, and d1.

• The field n contains the normal vector for the triangle.

• The field p contains node point coordinate information.
• The field t contains the indices to columns in p of a simplex mesh, each column in
t representing a simplex.

• The field name contains the name of the file the data come from.
• The field ref contains the header of the file.
• The field rgb contains the color model data at each vertices. The columns
correspond to node point coordinates in columns in p.
• The field expr contains the data description.
• The field d1 contains the data value at each vertices. The columns correspond to
node point coordinates in columns in p.

Both binary and ASCII STL files are supported.

SEE ALSO
mphquad2tri, mphsurf, mphwritestl

mphreduction
Return reduced-order state-space matrices for a model.

SYNTAX
data = mphreduction(model, romtag, ...)

DESCRIPTION
data = mphreduction(model, romtag, ...) extracts the reduced order state-space
matrices from the reduced order model romtag. The reduced order model can either
be created in the COMSOL GUI or via the API.

COMMANDS GROUPED BY FUNCTION | 335

 The function mphreduction accepts the following property/value pairs:
TABLE 6-28: PROPERTY/VALUE PAIRS FOR THE MPHREDUCTION COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

out Cell array of strings 'all' Names of output matrices

return struct | ss struct Return type

The following values are valid for the out property:

TABLE 6-29: PROPERTY/VALUE PAIRS FOR THE PROPERTY OUT.

PROPERTY EXPRESSION DESCRIPTION

out Kr Stiffness matrix

Dr Damping matrix
Dra Damping ratio matrix
Er Mass matrix
Br Input matrix
Cr Output matrix
F Input feedback matrix
B0r Initial value input matrix
B0rdot Initial value time derivative input matrix
Brdot Time derivative input matrix
Brdotdot Second time derivative input matrix
Mc Damping matrix
MA | A Stiffness matrix
MB | B Input matrix
D Input feedback matrix
C Output matrix
L Load vector
Y0 Output bias
U0 Initial value vector
Udot0 Initial derivative vector
Kud Stiffness matrix times ud

The return type ss requires that the Control System Toolbox is installed.

SEE ALSO
mphstate

336 | CHAPTER 6: COMMAND REFERENCE

mphreport
Generate report to model or write report.

SYNTAX
mphreport(model,...)
mphreport(model,'action','run','tag',rpttag,...)

DESCRIPTION
mphreport(model,...) generate report to the model model.

mphreport(model,'action','run','tag',rpttag,...) write report defined

with the tag rpttag.

The function mphreport accepts the following property/value pairs:

TABLE 6-30: PROPERTY/VALUE PAIRS FOR THE MPHREPORT COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

action add|run add Add or run report

filename String Report filename
format html | docx html Output format
open on | off on Open report when finished
tag String Report tag
template String User template
type brief | intermediate Report type
intermediate |
complete

EXAMPLE
Generate a brief report set with the tag 'myreport':

mphreport(model,'tag','myreport','type','brief')

Write report set with the tag 'myreport' to the Microsoft Word document
'modelreport.docx':

mphreport(model,'action','run','tag','myreport',...
'filename','modelreport','format','docx')

Do not open the report once written:

mphreport(model,'action','run','tag','myreport',...
'open','off')

COMMANDS GROUPED BY FUNCTION | 337

 mphsave
Save a COMSOL Multiphysics model.

SYNTAX
mphsave(model)
mphsave(model,filename,...)

DESCRIPTION
mphsave(model) saves the COMSOL model object model.

mphsave(model,filename,...) saves the COMSOL model object model to the file

named filename.

mphsave(filename) saves the unique COMSOL model that is loaded in the

COMSOL server to the file named filename.

mphsave(mtag) saves the COMSOL model loaded in the COMSOL server with the
tag mtag.

mphsave(mtag,filename,...) aves the COMSOL model loaded in the COMSOL

server with the tag mtag to the file named filename.

The function mphsave accepts the following property/value pairs:

TABLE 6-31: PROPERTY/VALUE PAIRS FOR THE MPHSAVE COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

component on | off off Save M-file using the component

syntax
optimize size | speed Optimize for speed or file size

If the file name is not provided, the model has to be saved previously on disk.

If the file name does not provide a path, the file is saved relatively to the current path
in MATLAB®.

The model can be saved as an MPH-file, Java file, or M-file. The file extension
determines which format that is saved.

Note: Model created with older version than COMSOL 5.3 cannot be saved using
the component syntax.

338 | CHAPTER 6: COMMAND REFERENCE

SEE ALSO
mphopen, mphload

mphsearch
Graphical user interface (GUI) for searching expressions in the COMSOL model
object.

SYNTAX
mphsearch(model)

DESCRIPTION
mphsearch(model) opens a graphical user interface that can be used to search
expressions in the model object model. Search using a text available in the name,
expression or description of the variable.

SEE ALSO
mphgetexpressions, mphgetproperties, mphgetselection, mphmodel,
mphnavigator

mphselectbox
Select geometric entity using a rubberband/box.

COMMANDS GROUPED BY FUNCTION | 339

 SYNTAX
n = mphselectbox(model,geomtag,boxcoord,entity,...)

DESCRIPTION
n = mphselectbox(model,geomtag,boxcoord,entity,...) returns the indices of
the geometry entities that are inside the rubberband domain (rectangle or box). This
method looks only on the vertex coordinates and does not observe all points on curves
and surfaces.

boxcoord set the coordinates of the selection domain, specified as a Nx2 array, where
N is the geometry space dimension.

entity can be one of point, edge, boundary, or domain following the entity space
dimension defined below:

• domain: maximum geometry space dimension

• boundary: maximum geometry space dimension − 1
• edges: 1 (for 3D geometries only)

The function mphpselectbox accepts the following property/value pairs:

TABLE 6-32: PROPERTY/VALUE PAIRS FOR THE MPHSELECTBOX COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

adjnumber Scalar None Adjacent entity number

When a model uses form an assembly, more than one vertex can have the same
coordinate if the coordinate is shared by separate geometry objects. In that case you
can use the adjnumber property to identify the domain that the vertices should be
adjacent to.

EXAMPLE
Find the domains using a box selection:

model = mphopen('model_tutorial_llmatlab');
coordBox = [-1e-3 11e-3;-1e-3 11e-3;9e-3 11e-3];
n = mphselectbox(model,'geom1',coordBox,'domain');

Find the boundaries inside the selection box:

n = mphselectbox(model,'geom1',coordBox,'boundary');

Find the boundaries inside the selection box that are adjacent to domain number 1:

n = mphselectbox(model,'geom1',coordBox,'boundary',...
'adjnumber',1);

340 | CHAPTER 6: COMMAND REFERENCE

Find geometric entity number in an assembly

model = mphopen('model_tutorial_llmatlab');
geom = model.component('comp1').geom('geom1');
geom.feature('fin').set('action','assembly');
geom.run('fin');

Find the boundaries within a box:

coordBox = [-1e-3,51e-3;-1e-3,51e-3;9e-3,11e-3];
n = mphselectbox(model,'geom1',coordBox,'boundary');

Find the boundary adjacent to domain 2:

n = mphselectbox(model,'geom1',coordBox,'boundary',...
'adjnumber',2);

SEE ALSO
mphgetadj, mphgetcoords, mphselectcoords, mphviewselection

mphselectcoords
Select a geometric entity using point coordinates.

SYNTAX
n = mphselectcoords(model,geomtag,coord,entity,...)

DESCRIPTION
n = mphselectcoords(model,geomtag,coord,entity,...) finds geometric
entity numbers based on their vertex coordinates.

One or more coordinates can be provided. The function searches for vertices near
these coordinates using a tolerance radius. The list of the entities that are adjacent to
such vertices is returned.

coord is a NxM array where N correspond of the number of point to use and M the
space dimension of the geometry.

entity can be one of point, edge, boundary or domain following the entity space
dimension defined below:

• domain: maximum geometry space dimension

• boundary: maximum geometry space dimension -1
• edges: 1(only for 3D geometry)

COMMANDS GROUPED BY FUNCTION | 341

 The function mphpselectcoords accepts the following property/value pairs:
TABLE 6-33: PROPERTY/VALUE PAIRS FOR THE MPHSELECTCOORDS COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

adjnumber Scalar Adjacent entity number

radius Scalar auto Search radius
include all | any all Include all or any vertices

When a model uses form an assembly, more than one vertex can have the same
coordinate if the coordinate is shared by separate geometry objects. In that case you
can use the adjnumber property to identify the domain that the vertices should be
adjacent to.

The radius property is used to specify the radius of the sphere or circle that the search
should be within. A small positive radius (based on the geometry size) is used by
default in order to compensate for rounding errors.

Use the property include when two point coordinates are used. Set it to all to select
objects within the search radius of all points. any returns objects within the search
radius of any points.

EXAMPLE
Find the geometric entity number:

model = mphopen('model_tutorial_llmatlab');
coord = [10e-3 0 10e-3;0 10e-3 10e-3];
n = mphselectcoords(model,'geom1',coord','point')

Return the indices of the point at coordinates within a search radius of 0.011:

n = mphselectcoords(model,'geom1',coord','point',...
'radius',0.011)

Return the indices of the boundaries that have a vertex within the search radius:

n = mphselectcoords(model,'geom1',coord','boundary',...
'radius',11e-3)

Return the indices of the edges that have a vertex within the search radius from all
points:

coord = [5e-3 0 10e-3;0 5e-3 10e-3];

n = mphselectcoords(model,'geom1',coord','edge',...
'radius',6e-3);

Return the indices of the edges that have a vertex within the search radius from at least
one point:

342 | CHAPTER 6: COMMAND REFERENCE

 n = mphselectcoords(model,'geom1',coord','edge',...
'radius',6e-3,'include','any');

Find geometric entity index in an assembly

model = mphopen('model_tutorial_llmatlab');
geom = model.component('comp1').geom('geom1');
geom.feature('fin').set('action', 'assembly');
geom.run('fin');

Return the indices of the boundaries that have any vertices within the search range of
a point:

coord = [0,0,10e-3];
n0 = mphselectcoords(model,'geom1',coord,'boundary')

Return the indices of the boundaries that also are adjacent to domain 1:

n1 = mphselectcoords(model,'geom1',coord,'boundary',...
'adjnumber',1);

Return the indices of the boundaries that also are adjacent to domain 2:

n1 = mphselectcoords(model,'geom1',coord,'boundary',...
'adjnumber',2);

SEE ALSO
mphgetadj, mphgetcoords, mphselectbox, mphviewselection

mphshowerrors
Show the messages in error nodes in the COMSOL Multiphysics model.

SYNTAX
mphshowerrors(model)
list = mphshowerrors(model)

DESCRIPTION
mphshowerrors(model) shows the error and warning messages stored in the model
and where they are located. The output is displayed in the command window.

list = mphshowerrors(model) returns the error and warning messages stored in

the model and where they are located in the N-by-3 cell array list, N corresponding to
the number of errors or warning found in the model object. The first column contains
the node of the error, the second column contain the error message and the third
column contains a cell arrays of the model tree nodes that contain the error

COMMANDS GROUPED BY FUNCTION | 343

 information, which can help for automated processing of error and warning
conditions.

mphsolinfo
Get information about a solution object.

SYNTAX
info = mphsolinfo(model,...)
info = mphsolinfo(model,'solname',soltag,...)

DESCRIPTION
info = mphsolinfo(model,...) returns information about the solution object.

The function mphsolinfo accepts the following property/value pairs:

TABLE 6-34: PROPERTY VALUE PAIRS FOR THE MPHSOLINFO COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

soltag String Active solution object Solution object tag

dataset String Active solution dataset Dataset tag
NU on | off off Get info about number of
solutions

The returned value info is a structure with the following content

TABLE 6-35: FIELDS IN THE INFO STRUCTURE

FIELD CONTENT

soltag Solution node tag

study Study node tag
size Size of the solution vector
nummesh Number of meshes in the solution (for automatic
remeshing)
sizes Size of the solution vector for each mesh and number of
time steps/parameters for each mesh
soltype Solver type (Stationary, Parametric, Time or Eigenvalue)
solpar Name of the parameter
sizesolvals Length of the parameter list
solvals Values of the parameters, eigenvalues, or time steps
paramsweepnames Parametric sweep parameter names
paramsweepvals Parametric sweep parameter values

344 | CHAPTER 6: COMMAND REFERENCE

TABLE 6-35: FIELDS IN THE INFO STRUCTURE

FIELD CONTENT

label Solution node label

batch Information about solutions for parametric sweeps
dataset List of datasets that directly use the solution
NUsol Number of solution vectors stored
NUreacf Number of reaction forces vectors stored
NUadj Number of adjacency vectors stored
NUfsens Number of functional sensitivity vectors stored
NUsens Number of forward sensitivity vectors stored
You can use the function mphgetu to obtain the actual values of the solution vector.
Note that these functions are low level functions and you most often would use
functions such as mphinterp and mpheval to extract numerical data from a model.

EXAMPLE
Get the information about the default solution object:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
std.feature.create('stat','Stationary');
std.run;
solinfo = mphsolinfo(model)

Get information of multiple solver solution:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
param = std.feature.create('param','Parametric');
time = std.feature.create('time','Transient');
time.set('tlist', 'range(0,1,25)');
param.setIndex('pname','power',0);
param.setIndex('plistarr','30 60 90',0);
std.run;

Get the information about the 1st outer solution (power = 30):

solinfo = mphsolinfo(model,'soltag','sol3');

Get the solution vector for 2nd outer solution (power = 60):

solinfo = mphsolinfo(model,'soltag','sol4');

SEE ALSO
mphgetu, mphxmeshinfo, mphsolutioninfo

COMMANDS GROUPED BY FUNCTION | 345

 mphsolutioninfo
Get information about solution objects and datasets containing given parameters.

SYNTAX
info = mphsolutioninfo(model)
info = mphsolutioninfo(model,'parameters',{{ei,vi,toli},...},…)

DESCRIPTION
info = mphsolutioninfo(model) returns information about all solution object and
solution dataset combinations in model.

info = mphsolutioninfo(model,'parameters',{{ei,vi,toli}, ...},…)

returns information about solution object and solution dataset containing the given
inner/outer solution parameters ei with the value equal to vi within the tolerance
toli.

The function mphsolutioninfo accepts the following property/value pairs:

TABLE 6-36: PROPERTY VALUE PAIRS FOR THE MPHSOLUTIONINFO COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

cellmap off | on off Set to return a cell version of the

map with headers
dataset String Active Dataset tag
solution
dataset
parameters Cell | Cell array Filter parameters, values, and
tolerances
soltag String | String cell Solution object tag
array
sort String | Scalar | auto Sort the map by column number or
auto header tag

The returned value info is a structure with the following content

TABLE 6-37: FIELDS IN THE INFO STRUCTURE

FIELD CONTENT

solutions List of matched solution tags

sol# Substructure containing information related to solution number
#

346 | CHAPTER 6: COMMAND REFERENCE

The substructure info.sol# has the following content
TABLE 6-38: FIELDS IN THE INFO.SOL# SUBSTRUCTURE

FIELD CONTENT

dataset Tag of the solution dataset

study Tag of the study associated to the solution
sequencetype Type of solution object
cellmap Cellmap describing the connections between parameters and
inner/outer solution numbers
values Parameters values used in the solution
parameters Parameters names used in the solution
mapheaders Headers for the map
map Map describing the connections between parameters and
inner/outer solution numbers

EXAMPLE
Load model_tutorial_llmatlab.mph:

model = mphopen('model_tutorial_llmatlab');

Create a study combining a parametric sweep and a transient study step:

std = model.study.create('std');
param = std.feature.create('param','Parametric');
time = std.feature.create('time','Transient');

Set the time stepping and the parametric sweep parameters:

time.set('tlist', 'range(0,1,25)');
param.setIndex('pname','power',0);
param.setIndex('plistarr','30 60 90', 0);

Run the study:

std.run;

Retrieve the solution information corresponding to power = 30 W:

info = mphsolutioninfo(model,'parameters',{'power',30,0})

Retrieve the solution information corresponding to power = 90 W and around

t = 10.4 sec. and its associated solution dataset:

info = mphsolutioninfo(model,'parameters',{{'power',90,0},...
{'t',10.4,0.5}})

Get the solution solution dataset associated:

COMMANDS GROUPED BY FUNCTION | 347

 dset = info.sol2.dataset

Get the inner and outer solution number:

solnum = info.sol2.map(end-1)
outersolnum = info.sol2.map(end)

SEE ALSO
mphgetu, mphxmeshinfo, mphsolinfo

mphstart
Connect MATLAB® to a COMSOL server.

SYNTAX
mphstart
mphstart(port)
mphstart(ipaddress, port)
mphstart(ipaddress, port, username, password)
mphstart(ipaddress, port, comsolpath)
mphstart(ipaddress, port, comsolpath, username, password)

DESCRIPTION
mphstart creates a connection with a COMSOL server using the default port number
(which is 2036).

mphstart(port) creates a connection with a COMSOL server using the specified

port number port.

mphstart(ipaddress, port) creates a connection with a COMSOL server using

the specified IP address ipaddress and the port number port. This command
assumes that the client and the server machine share the same login properties.

mphstart(ipaddress, port, username, password) creates a connection with a

COMSOL server using the specified IP address ipaddress and the port number
port, the username username and password password.

mphstart(ipaddress, port, comsolpath) creates a connection with a COMSOL

server using the specified IP address and port number using the comsolpath that is
specified. This is useful if mphstart cannot find the location of the COMSOL
Multiphysics installation.

mphstart(ipaddress, port, comsolpath, username, password) creates a

connection with a COMSOL server using the specified IP address, the port number,

348 | CHAPTER 6: COMMAND REFERENCE

the username and password using the comsolpath that is specified. This is useful if
mphstart cannot find the location of the COMSOL Multiphysics installation.

mphstart can be used to create a connection from within MATLAB when this is
started without using the COMSOL with MATLAB option. mphstart then sets up
the necessary environment and connect to COMSOL.

Prior to calling mphstart it is necessary to set the path of mphstart.m in the

MATLAB path or to change the current directory in MATLAB (for example, using the
cd command) to the location of the mphstart.m file.

A COMSOL server must be started prior to running mphstart.

mphstart connect to either a COMSOL Multiphysics Server (started with the

command: comsol mphserver) or the COMSOL Server™. To connect to the
COMSOL Server™ from a computer that has just MATLAB installed it is necessary to
run the COMSOL Server™ Client installer.

Once MATLAB is connected to the server, import the COMSOL class in order to use
the ModelUtil commands. To import the COMSOL class enter:

import com.comsol.model.*
import com.comsol.model.util.*

EXAMPLE
Connect manually MATLAB to a COMSOL Multiphysics Server and create a model:

mphstart
import com.comsol.model.*
import com.comsol.model.util.*
model = ModelUtil.create('Model');

Connect manually MATLAB to a COMSOL Multiphysics Server running on the

computer with the IP address 192.168.0.1 using port 2037:

mphstart('192.168.0.1',2037)

mphstate
Get state-space matrices for a dynamic system.

SYNTAX
str = mphstate(model,soltag,'Out',{'SP'})
str = mphstate(model,soltag,'Out',{'SP1','SP2',...})

COMMANDS GROUPED BY FUNCTION | 349

 DESCRIPTION
®
str = mphstate(model,soltag,'out',{'SP'}) returns a MATLAB structure
str containing the state-space matrix SP assembled using the solver node soltag and
accessible as str.SP, SP being taken from the Out property list.

str = mphstate(model,soltag,'out',{'SP1','SP2',...}) returns a MATLAB

structure str containing the state-space matrices SP1, SP2, ... assembled using the
solver node soltag and accessible as str.SP1and str.SP2. SP1 and SP2 being taken
from the out property list.

The function mphstate accepts the following property/value pairs:

TABLE 6-39: PROPERTY VALUE FOR THE MPHSTATE COMMAND

PROPERTY VALUE DEFAULT DESCRIPTION

out MA | MB | A | B | C | D | Output matrix

Mc | Null | ud | x0
keepfeature off | on off Keep the StateSpace feature in
the model
input String Input variables
output String Output variables
sparse off | on off Return sparse matrices
initmethod init | sol init Use linearization point
initsol soltag | zero soltag Solution to use for linearization
solnum Positive integer auto Solution number

The property Sparse controls whether the matrices A, B, C, D, M, MA, MB, and Null are
stored in the sparse format.

The equations correspond to the system below:

·
 M C x = M C Ax + M C Bu

 y = Cx + Du

where x are the state variables, u are the input variables, and y are the output variables.

A static linearized model of the system can be described by:

y = ( D – C ( M C A ) – 1 M C B )u

The full solution vector U can be then obtained from

U = Nullx + ud + u 0

350 | CHAPTER 6: COMMAND REFERENCE

where Null is the null-space matrix, ud the constraint contribution, and u0 is the
linearization point, which is the solution stored in the sequence once the state-space
export feature is run.

The matrices MC and MCA are produced by the same algorithms that do the
finite-element assembly and constraint elimination in COMSOL Multiphysics. MC and
MCA are the same as the matrices DC (eliminated mass matrix) and −KC (KC is the
eliminated stiffness matrix). The matrices are produced from an exact residual vector
Jacobian calculation (that is, differentiation of the residual vector with respect to the
degrees of freedoms x) plus an algebraic elimination of the constraints. The matrix C
is produced in a similar way (that is, the exact output vector Jacobian matrix plus
constraint elimination).

The matrices MCB and D are produced by a numerical differentiation of the residual
and output vectors, respectively, with respect to the input parameters (the algorithm
systematically perturbs the input parameters by multiplying them by a factor 1+10−8).

The input cannot be a variable constraint in the model.

EXAMPLE
Load model_tutorial_llmatlab.mph:

model = mphopen('model_tutorial_llmatlab');
comp1 = model.component('comp1');
comp1.mesh('mesh1').autoMeshSize(9);
std = model.study.create('std');
time = std.feature.create('time','Transient');
time.set('tlist','range(0,1,50)');

Edit the model to use parameter for the initial and external temperature:

model.param.set('T0','293.15[K]');
model.param.set('Text','300[K]');
ht = model.component('comp1').physics('ht');
ht.feature('init1').set('Tinit', 'T0');
ht.feature('hf1').set('Text', 'Text');

Add a domain point probe plot:

pdom = comp1.probe.create('pdom', 'DomainPoint');

pdom.model('comp1');
pdom.set('coords3',[0 0 1.1e-2]);

Run the study and create a plot group to display the probe:

std.run;
pg1 = model.result.create('pg1', 'PlotGroup1D');
glob1 = pg1.create('glob1', 'Global');

COMMANDS GROUPED BY FUNCTION | 351

 glob1.setIndex('expr', 'comp1.ppb1', 0);
Extract the state-space system matrices of the model with power, Temp, and Text as
input and the probe evaluation comp1.ppb1 as output:

M = mphstate(model,'sol1','out',{'A','B','C','D'},...
'input',{'power','Temp','Text'},'output','comp1.ppb1');

Plot the sparsity of the matrix A:

subplot(1,2,1); spy(M.A); subplot(1,2,2); spy(abs(M.A)>1e-2)

Set the input power parameter and the reference temperature:

power = 30; Temp = 300; Text = 300; T0 = 293.15;

Compute the system solution:

input = [power Temp-T0 Text-T0];

func = @(t,x) M.A*x + M.B*input';
[t,x] = ode45(func,0:1:50,zeros(size(M.A,1),1));
y = M.C*x';
y = y+T0;

Plot the result:

plot(t,y,'r');
hold on;
mphplot(model,'pg1');

Evaluate the steady-state temperature value:

G = M.D-M.C*(inv(M.A))*M.B;
y_inf = full(G*input');
y_inf = y + T0

mphsurf
Create plot data structure from surf data.

SYNTAX
pd = mphsurf(x,y,z)
pd = mphsurf(z)

DESCRIPTION
pd = mphsurf(x,y,z) creates the plot data structure pd from surf data x, y, and z.

pd = mphsurf(z) creates the plot data structure pd from surf data z. A unit scale is
assumed for the x and y coordinates.

352 | CHAPTER 6: COMMAND REFERENCE

EXAMPLE
Create random height data

[x,y] = meshgrid(-0.1:0.2:1.1,-0.4:0.2:0.4);
z = cumsum(0.1*randn(size(x)));

Create 3D surface plotdata structure

pd = mphsurf(x,y,z);
mphplot(pd)

SEE ALSO
mphquad2tri, mphreadstl, mphwritestl

mphtable
Get table data.

SYNTAX
info = mphtable(model,tabletag)

DESCRIPTION
info = mphtable(model,tabletag) returns the structure info containing the data
with the tabletag tag and its headers.

The returned value info is a structure with the following content

TABLE 6-40: FIELDS IN THE INFO STRUCT

FIELD CONTENT

headers Headers of the table

tag Tag of the table
data Data of the extracted table
filename Filename when table exported to file

EXAMPLE
Load model_tutorial_llmatlab.mph, add a stationary study and compute the
solution for different power values:

model = mphopen('model_tutorial_llmatlab');
std = model.study.create('std');
stat = std.feature.create('stat','Stationary')
stat.setIndex('pname','power',0);
stat.setIndex('plistarr','30 60 90',0);
std.run;

COMMANDS GROUPED BY FUNCTION | 353

 Evaluate the maximum temperature in the model and set the results in a table:

max = model.result.numerical.create('max','MaxVolume');
max.selection.all;
tbl = model.result.table.create('tbl','Table');
tbl.comments('Volume Maximum (T)');
max.set('table','tbl');
max.setResult;

Extract the table data:

str = mphtable(model,'tbl');
tbl_data = str.data

SEE ALSO
mpheval, mphevalpoint, mphglobal, mphint2, mphinterp, mphmax, mphmean,
mphmin

mphtags
Get tags and names for nodes in a COMSOL Multiphysics model.

SYNTAX
mphtags(model)
mphtags(node)
mphtags(model, type)
[tags,labels,displaystrings] = mphtags(...)
mphtags
mphtags -show
[tags,filename,fullfilename] = mphtags

DESCRIPTION
mphtags is used to retrieve tags from nodes in a COMSOL Multiphysics model or tags
from models that are loaded on the server.

When mphtags is called with a model or node variable the tags are returned form the
model. mphtags also be called using a model variable and a type, where type can be
one of these strings: result, dataset, table, numerical, and export to give
easy access to the nodes under the result node. It is sufficient to use the first letter of
the types.

If mphtags is called with output arguments it is possible to get both the tags as well as
labels and display names used for the nodes. For example,

[tags,labels,displaystrings] = mphtags(model.geom)

354 | CHAPTER 6: COMMAND REFERENCE

If mphtags is called with the root model node as argument, the filename of the model
can be returned:

[tag,filename,displaystring] = mphtags(model)

mphtags can be used to return a list of files currently loaded on the server. For
example,

[tags,filename,fullfilename] = mphtags

In order to see this information quickly it is possible to call mphtags like this:

mphtags -show

that just produces output that is useful viewing on screen.

mphthumbnail
Set or get model thumbnail.

SYNTAX
mphthumbnail(model,filename)
mphthumbnail(model,image)
mphthumbnail(model,fig)
mphthumbnail(model,'')
[image,imagefilename] = mphthumbnail(model)

DESCRIPTION
mphthumbnail sets or gets the model thumbnail for model loaded on the server. In
order to update the model thumbnail on disk the model must be saved.

mphthumbnail(model,filename) sets the thumbnail for the model to the image

contained in filename. The file must be a PNG- or JPG-file.

mphthumbnail(model,image) sets the thumbnail using the image data image. image
is either a NxM or a NxMx3 matrix. The preferred size of the image is 280 by 210
pixels.

mphthumbnail(model,fig) sets the thumbnail using the image in the figure with
handle fig.

mphthumbnail(model, '') clears the thumbnail from the model.

[image,imagefilename] = mphthumbnail(model) gets the image data image and

the image filename imagefilename for the thumbnail stored in the model model.

COMMANDS GROUPED BY FUNCTION | 355

 EXAMPLE
Load model_tutorial_llmatlab.mph:

model = mphopen('model_tutorial_llmatlab');

Get the thumbnail image:

im = mphthumbnail(model);

Show the thumbnail in a MATLAB figure:

imshow(im)

Save the current figure image as an image file:

filename = fullfile(tempdir,'imagefile.png');
print(filename,'-dpng','-r48')

Set the thumbnail for the model:

mphthumbnail(model,filename)

SEE ALSO
mphload, mphsave

mphversion
Return the version number for COMSOL Multiphysics.

SYNTAX
v = mphversion
[v,vm] = mphversion(model)

DESCRIPTION
v = mphversion returns the COMSOL Multiphysics version number that MATLAB
is connected to as a string.

[v,vm] = mphversion(model) returns the COMSOL Multiphysics version number

that MATLAB is connected to as a string in the variable v and the version number of
the model in the variable vm.

EXAMPLE
Load model_tutorial_llmatlab.mph:

model = mphopen('model_tutorial_llmatlab');

Get the version numbers:

356 | CHAPTER 6: COMMAND REFERENCE

 [version, model_version] = mphversion(model)

SEE ALSO
mphload, mphsave

mphviewselection
Display a geometric entity selection in a MATLAB® figure.

SYNTAX
mphviewselection(model,geomtag,number,entity,...)
mphviewselection(model,seltag,...)

DESCRIPTION
mphviewselection(model,geomtag,number,entity,...) displays the geometric
entity number of type entity in MATLAB figure including the representation of the
geometry geomtag.

mphviewselection(model,seltag,...) displays the geometric entity selection

seltag in a MATLAB figure including the representation of the geometry.

The function mphviewselection accepts the following property/value pairs:

TABLE 6-41: PROPERTY VALUE/PAIRS FOR THE MPHVIEWSELECTION FUNCTION

PROPERTY VALUE DEFAULT DESCRIPTION

edgecolor Char | RGB array k Color for edges

edgecolorselected RGB array [1,0,0] Color for selected
edges
edgelabels on | off off Show edge labels
edgelabelscolor Char | RGB array g Color for edge labels
edgemode on | off on Show edges
entity Domain | Set the selected entity
boundary | edge | type
point
facealpha Double 1 Set transparency value
facecolor RGB array [0.6,0.6, Color for face
0.6]
facecolorselected RGB array [1,0,0] Color for selected
face
facelabels on | off off Show face labels
facelabelscolor Char | RGB array b Color for face labels

COMMANDS GROUPED BY FUNCTION | 357

 TABLE 6-41: PROPERTY VALUE/PAIRS FOR THE MPHVIEWSELECTION FUNCTION

PROPERTY VALUE DEFAULT DESCRIPTION

facemode on | off on Show faces

geommode on | off on Show entire geometry
marker . Vertex marker
markercolorselected Char | RGB array r Color for selected
vertex marker
markersize Int 12 Font size of marker
parent Double Parent axis
renderer Opengl | zbuffer opengl Set the rendering
method
selection String | Positive Set selection name or
integer array entity number
selectoralpha Double 0.25 Set selector
transparency value
selectorcolor RGB array [0,0,1] Color for selected
marker
showselector on | off on Show Selector
vertexlabels on | off off Show vertex labels
vertexlabelscolor Char | RGB array r Color for vertex labels
vertexmode on | off off Show vertices

EXAMPLE
Plot boundary 6 using yellow color:

model = mphopen('model_tutorial_llmatlab');
mphviewselection(model,'geom1',6,'boundary',...
'facecolorselected',[1 1 0],'facealpha',0.5)

Plot edges 1 to 8 using green color:

mphviewselection(model,'geom1',1:8,'edge',...
'edgecolorselected',[0 1 0])

Add an explicit selection for boundaries 7 to 12 and plot the selection in a figure:

model.selection.create('sel1','Explicit').geom(2).set(7:12);
mphviewselection(model,'sel1');

Add a selection to get the vertex indices with the box delimited with the coordinates
[-1e-3 11e-3;-1e-3 11e-3;9e-3 11e-3] and plot both the selected entities and the
selector:

358 | CHAPTER 6: COMMAND REFERENCE

 box = model.selection.create('box1', 'Box');
box.set('entitydim', '0');
box.set('xmin', '-1e-3').set('xmax', '11e-3');
box.set('ymin', '-1e-3').set('ymax', '11e-3');
box.set('zmin', '10e-3').set('zmax', '11e-3');
mphviewselection(model,'box1','facemode','off');

SEE ALSO
mphgeom, mphselectbox, mphselectcoords

mphwritestl
Export plot data as an STL file.

SYNTAX
mphwritestl(filename, pd)
mphwritestl(filename, pd, '-binary')

DESCRIPTION
mphwritestl(filename, pd) exports data in the plot data structure pd as the STL
file filename.

mphwritestl(filename, pd, '-binary') exports data in the plot data structure

pd as the STL file filename using the binary file format.

pd is a structure with fields pd and t.

• The field p contains node point coordinate information.

• The field t contains the indices to columns in p of a simplex mesh, each column in
t representing a simplex.

Other fields in the plot data structure are not considered to generate the surface mesh.

EXAMPLE
.Generate a surface mesh from solution plot:

model = mphopen('vacuum_flask_llmatlab')
pd = mphplot(model, 'pg1')
pd2stl = pd{2}{1};
mphwritestl('vacuum_flask.stl', pd2stl)

Generate a surface mesh from a volume mesh

model = mphopen('model_tutorial_llmatlab');
model.component('comp1').mesh('mesh1').run;
[s,d] = mphmeshstats(model, 'mesh1');

COMMANDS GROUPED BY FUNCTION | 359

 idx = strcmp(s.types, 'tri');
pdmesh.p = d.vertex;
pdmesh.t = d.elem{idx};
mphwritestl('mesh2geom.stl', pdmesh);

SEE ALSO
mphquad2tri, mphreadstl, mphsurf

mphxmeshinfo
Extract information about the extended mesh.

SYNTAX
info = mphxmeshinfo(model, ...)

DESCRIPTION
info = mphxmeshinfo(model,...) extracts extended mesh information from the
active solution object.

The function mphxmeshinfo accepts the following property/value pairs:

TABLE 6-42: PROPERTY VALUE/PAIRS FOR THE MPHVIEWSELECTION FUNCTION

PROPERTY VALUE DEFAULT DESCRIPTION

soltag String Active solution Solution object tag

object
studysteptag String Study step node tag
meshcase Positive integer | String 1 Mesh case tag

The function xmeshinfo returns a structure with the fields shown in the table below
TABLE 6-43: FIELD IN THE RETURNED STRUCTURE FROM MPHXMESHINFO

FIELD DESCRIPTION

soltag Tag of the solution object

ndofs Number of DOFs
fieldnames Names of the field variables
fieldndofs Number of DOFs per field name
meshtypes Types of mesh element
dofs Structure with information about the degrees of
freedom
nodes Structure with information about the nodes
elements Structure with information about each element type

360 | CHAPTER 6: COMMAND REFERENCE

The extended mesh information provide information about the numbering of
elements, nodes, and degrees of freedom (DOFs) in the extended mesh and in the
matrices returned by mphmatrix and mphgetu.

EXAMPLE
Extract xmesh information:

model = mphopen('model_tutorial_llmatlab.mph');
std = model.study.create('std');
std.feature.create('stat', 'Stationary');
std.run;
info = mphxmeshinfo(model)

Get the number of degrees of freedom and the nodes coordinates:

dofs = info.ndofs
coords = info.dofs.coords;

Get the DOFs indices connected to the tetrahedron:

idx = info.elements.tet.dofs

Retrieve the xmesh information with several physics

model = mphopen('model_tutorial_llmatlab.mph');
comp1 = model.component('comp1');
ec = comp1.physics.create('ec','ConductiveMedia','geom1');
ec.feature.create('gnd1','Ground',2).selection.set(3);
pot = ec.feature.create('pot','ElectricPotential',2);
pot.selection.set(7);
pot.set('V0',1,'50[mV]');
hs = comp1.physics('ht').feature('hs1');
hs.set('heatSourceType',1,'generalSource');
hs.set('Q_src',1,'root.comp1.ec.Qh');
std = model.study.create('std');
std.feature.create('stat', 'Stationary');
std.run;
info = mphxmeshinfo(model)

Get the index of the nodes for element with the index 100:

idx_nodes = info.elements.tet.nodes(:,100)

Get the index of the DOFs for element with the index 100:

idx_dofs = info.elements.tet.dofs(:,100)

Get the index of the variables names corresponding to the DOFs with the index
idx_dofs:

idx_names = info.dofsinds(idx_dofs);

COMMANDS GROUPED BY FUNCTION | 361

 Find the dofnames index corresponding to the variable V:

idx_dofnames = find(strcmp(info.dofs.dofnames,'comp1.V'))-1;

Get the list of DOFs that correspond to the variable V:

idx = find(idx_names==idx_dofnames)

Get the coordinates of the DOFs corresponding to the dependent variable V that
belong to element 100:

info.dofs.coords(:,idx_dofs(idx))

SEE ALSO:
mphgetu, mphmatrix, mphsolinfo, mphsolutioninfo

362 | CHAPTER 6: COMMAND REFERENCE

I n d e x
A adding model objects 40
animations 150 client/server mode 24
ball selections 126 cluster computing 137
box selections 128 color display, selections 132
geometry operations 50 color tables 145
global equations 120 combining meshes 94
interpolation functions 122 compact history 41
job sequences 137 compose operation 55
MATLAB feature node 257 composite object, creating 54
mesh sequences 76 COMSOL API 38
parametric sweeps 137 COMSOL exceptions 242
physics interfaces 66, 115 COMSOL Multiphysics binary files 103
plot groups 140 COMSOL Multiphysics text files 103
study nodes 134 COMSOL server 24
adjacent selections 130 connect to server 47
advancing front method 82 connecting MATLAB 28
animation export 150 constructor name 114
animation player 151 converting
Application Libraries window 20 curve segments 58
ASCII format 150 image data 71
average of expressions 170 image file to data 69

B ball selections 126, 132 mesh elements 102

batch jobs 137 copying

batch mode 190 boundary meshes 98

boundary layer meshes 95 mphnavigator properties 236

boundary meshes 98 creating

boundary modeling 57 1D geometries 53

box selections 128, 132 2D geometries 54, 57

building 3D geometries 59

geometry sequences 51 composite objects 54

mesh sequences 77 geometry from image data 69

meshes 87 materials 118

mesh information 109
C calling MATLAB functions 36, 157
model objects 39
clearing
curve interpolation, example 68
functions 259
D data export 151

INDEX| 363
 data, extracting 186 data 159, 186
datasets syntax 148 eliminated matrices 196
defining matrices 205
materials 118 mesh information 109
MATLAB functions 254 plot data 143
selections 125 solution vectors 225
settings 122 system matrices 192
Delaunay method 82 extruding meshes 90, 92
derivative recovery 163
F floating network license (FNL) 29
difference operation 55
free meshing 94
directory path, MATLAB function 257
free quad mesh, example 84
disabling model history 246
free triangle mesh, example 82
disconnecting MATLAB 30
function derivatives 259
displaying
function inputs/outputs 258
geometries 51
functions
meshes 77
interpolation 122
plot groups 141
MATLAB 157
selections 131
MATLAB, adding 252
documentation 18
G geometry
dofs, xmesh 229
creating 67
E element, xmesh 229
displaying 51
emailing COMSOL 21
parameterized 64
enabling model history 246
parametrization, example 188
entity, geometry 52
retrieve information 62
equations, modifying 119
sequence 50
errors 242
global equations 120
evaluating
global expressions 180
data 149
global matrix 184
expressions 173
H Hankel function 256
global expressions 180
history, model 42
global matrix 184
integrals 168 I image data conversion, example 71
explicit selections 125 image data, create geometry 69
exporting importing
data 151 meshes 103
expression average 170 imread (MATLAB function) 69
extended mesh 228 inner solution 222
extracting integrals, evaluating 168

364 | I N D E X
 internet resources 18 methods 38
interpolation curve 67 methods, mphnavigator 237
interpolation functions 122 Microsoft Windows 26
minimum of expressions 75, 163
J Java 38
model examples 17
Java heap size 244
model expressions 237
job sequences 137
model features 245
K knowledge base, COMSOL 21
model history 42, 246
L linear matrix 198 model object
linearization points 193, 206 calling 157
Linux 26 create custom GUI 247
list model object 40 information 237
load model 41 methods 39
loops 187, 246 navigating 231
Model Tree 235
M Mac OS X 26
models, running in loops 187
mass matrix 204
ModelUtil method 39
materials 118
modifying equations 119
MATLAB desktop 24
mpheval 173–176
MATLAB feature node 257
mphevalglobalmatrix 184
MATLAB functions 157, 252
mphevalpoint 176–178
MATLAB functions, plot 255
MPH-files 20
matrices, state-space 204
mphgetexpressions 241
maximum of expression 165
mphgetproperties 239–240
measuring, mesh quality 105
mphgetselection 241
memory requirements 246
mphgetu 225–226
mesh
mphglobal 180, 182
boundary layers 95
mphinputmatrix 198–199
converting 102
mphint2 168–170
copying 98
mphinterp 159–161, 163
data 109
mphmatrix 192–193, 196
displaying 77
mphmax 165–167
element size, controlling 78
mphmean 170–172
importing 103
mphmin 75, 163–165
quality 105
mphmodel 237
refining 98
mphnavigator 231, 233, 235–237
resolution 79
mphparticle 179–180
sequence 76
mphray 179
statistics 106

INDEX| 365
 mphsearch 237 running, models in loops 187
mphshowerrors 242
S save model object 44
mphsolinfo 220–221
selecting, linearization points 193
mphsolutioninfo 222–224
selections
mphstate 204–207, 212
defining 125
mphtable 186
displaying 131
mphxmeshinfo 201, 228, 230
sequences of operations 38
myscript 190
sequences, solvers 136
N NASTRAN 103 set method 156
node points 173 set operations 55
nodes, xmesh 229 set the feature property 51
numerical node syntax 149 setindex method 157
setting
O ODE problem, example 121
linear matrix system 198
outer solution 222
linearization points 206
P parameterized geometries 64
simplex elements 98
parametric jobs 137
solid modeling 59
parametric sweep 137
solution information 220, 222
particle trajectories 179
solution object 220
Physics Builder 123
solution vector 225
physics interfaces 114–115
solutions, specifying 226
plot data, extracting 143
solver configurations syntax 135
plot groups 140–141
solving, ODE problems 121
plot while solving 138
squeezed singleton 178
plotting data, example 146
state-space export 204
port number 24
statistics, mesh 106
preferences 32
structured meshes 85
prism mesh 92
study syntax 134
progress bar 40
sweeping meshes 90
Q quadrilateral mesh, example 85 swept meshing 94
syntax
R ray trajectories 179
datasets 148
refining meshes 98
materials 118
remove model object 40
numerical node 149
resolution, mesh 79
physics interfaces 114
results evaluation 149
plot groups 140
revolved prism mesh, example 90
solver configurations 135
revolving face meshes 90
studies 134
run solver sequences 136

366 | I N D E X
 system matrices 192

T table data 186

technical support, COMSOL 21
tolerance radius 127
transparency, selections 132
triangular mesh, example 79

U updates, disable 245

user-defined physics interface 123

W warnings 242
weak form equation, example 119
websites, COMSOL 21

X xmesh 228
xterm 190–191

INDEX| 367
368 | I N D E X