Download the full version of the textbook now at textbookfull.

com

Numerical Methods for Engineering: An

introduction using MATLAB® and computational
electromagnetics examples (Electromagnetic
Waves) 2nd Edition Karl F. Warnick
https://textbookfull.com/product/numerical-
methods-for-engineering-an-introduction-using-
matlab-and-computational-electromagnetics-
examples-electromagnetic-waves-2nd-edition-karl-f-
warnick/

Explore and download more textbook at https://textbookfull.com

 Recommended digital products (PDF, EPUB, MOBI) that
you can download immediately if you are interested.

Numerical Methods for Engineers and Scientists Using

MATLAB Second Edition Esfandiari

https://textbookfull.com/product/numerical-methods-for-engineers-and-
scientists-using-matlab-second-edition-esfandiari/

textbookfull.com

An introduction to numerical methods: a MATLAB approach

4th Edition Guenther

https://textbookfull.com/product/an-introduction-to-numerical-methods-
a-matlab-approach-4th-edition-guenther/

textbookfull.com

An introduction to numerical methods : a MATLAB® approach

Fourth Edition Guenther

https://textbookfull.com/product/an-introduction-to-numerical-methods-
a-matlab-approach-fourth-edition-guenther/

textbookfull.com

Design of CMOS phase locked loops from circuit level to

architecture level 1st Edition Behzad Razavi

https://textbookfull.com/product/design-of-cmos-phase-locked-loops-
from-circuit-level-to-architecture-level-1st-edition-behzad-razavi/

textbookfull.com
Time Limited Adolescent Psychodynamic Psychotherapy A
Developmentally Focussed Psychotherapy for Young People
Stephen Briggs
https://textbookfull.com/product/time-limited-adolescent-
psychodynamic-psychotherapy-a-developmentally-focussed-psychotherapy-
for-young-people-stephen-briggs/
textbookfull.com

The Wisconsin Oneidas and the Episcopal Church A Chain

Linking Two Traditions L Gordon Mclester Iii Laurence M
Hauptman Kenneth Hoyan House
https://textbookfull.com/product/the-wisconsin-oneidas-and-the-
episcopal-church-a-chain-linking-two-traditions-l-gordon-mclester-iii-
laurence-m-hauptman-kenneth-hoyan-house/
textbookfull.com

New Media Spectacles and Multimodal Creativity in a

Globalised Asia Art Design and Activism in the Digital
Humanities Landscape Sunny Sui-Kwong Lam
https://textbookfull.com/product/new-media-spectacles-and-multimodal-
creativity-in-a-globalised-asia-art-design-and-activism-in-the-
digital-humanities-landscape-sunny-sui-kwong-lam/
textbookfull.com

World of Art The Photograph as Contemporary Art Charlotte

Cotton

https://textbookfull.com/product/world-of-art-the-photograph-as-
contemporary-art-charlotte-cotton/

textbookfull.com

Fit at Any Age It s Never Too Late 1st Edition Susan

Niebergall

https://textbookfull.com/product/fit-at-any-age-it-s-never-too-
late-1st-edition-susan-niebergall/

textbookfull.com
The Puzzle of Peace: The Evolution of Peace in the
International System 1st Edition Gary Goertz

https://textbookfull.com/product/the-puzzle-of-peace-the-evolution-of-
peace-in-the-international-system-1st-edition-gary-goertz/

textbookfull.com
Numerical Methods for
Engineering
 The ACES Series on Computational and
Numerical Modelling in Electrical Engineering
Andrew F. Peterson, PhD – Series Editor

The volumes in this series encompass the development and application of numerical
techniques to electrical and electronic systems, including the modelling of electro-
magnetic phenomena over all frequency ranges and closely related techniques for
acoustic and optical analysis. The scope includes the use of computation for engi-
neering design and optimization, as well as the application of commercial modelling
tools to practical problems. The series will include titles for senior undergraduate and
postgraduate education, research monographs for reference, and practitioner guides
and handbooks.

Published Books in the ACES Series:

W. Yu, X. Yang, and W. Li, “VALU, AVX and GPU Acceleration Techniques for
Parallel FDTD Methods,” 2014.
A.Z. Elsherbeni, P. Nayeri, and C.J. Reddy, “Antenna Analysis and Design Using
FEKO Electromagnetic Simulation Software,” 2014.
A.Z. Elsherbeni and V. Demir, “The Finite-Difference Time-Domain Method in
Electromagnetics with MATLAB® Simulations, 2nd Edition,” 2015.
M. Bakr, A.Z. Elsherbeni, and V. Demir, “Adjoint Sensitivity Analysis of High
Frequency Structures with MATLAB® ,” 2017.
O. Ergul, “New Trends in Computational Electromagnetics,” 2019.
D. Werner, “Nanoantennas and Plasmonics: Modelling, design and fabrication,” 2020.
Numerical Methods for
Engineering
An introduction using MATLAB® and
computational electromagnetics examples
2nd Edition

Karl F. Warnick

The Institution of Engineering and Technology

Published by SciTech Publishing, an imprint of The Institution of Engineering and Technology,
London, United Kingdom

The Institution of Engineering and Technology is registered as a Charity in England & Wales
(no. 211014) and Scotland (no. SC038698).

© The Institution of Engineering and Technology 2020

First published 2011

Second Edition published 2020

This publication is copyright under the Berne Convention and the Universal Copyright
Convention. All rights reserved. Apart from any fair dealing for the purposes of research
or private study, or criticism or review, as permitted under the Copyright, Designs and
Patents Act 1988, this publication may be reproduced, stored or transmitted, in any
form or by any means, only with the prior permission in writing of the publishers, or in
the case of reprographic reproduction in accordance with the terms of licences issued
by the Copyright Licensing Agency. Enquiries concerning reproduction outside those
terms should be sent to the publisher at the undermentioned address:
The Institution of Engineering and Technology
Michael Faraday House
Six Hills Way, Stevenage
Herts, SG1 2AY, United Kingdom

www.theiet.org

While the author and publisher believe that the information and guidance given in this
work are correct, all parties must rely upon their own skill and judgement when making
use of them. Neither the author nor publisher assumes any liability to anyone for any
loss or damage caused by any error or omission in the work, whether such an error or
omission is the result of negligence or any other cause. Any and all such liability
is disclaimed.

The moral rights of the author to be identified as author of this work have been
asserted by him in accordance with the Copyright, Designs and Patents Act 1988.

British Library Cataloguing in Publication Data

A catalogue record for this product is available from the British Library

ISBN 978-1-83953-073-9 (hardback)

ISBN 978-1-83953-074-6 (PDF)

Typeset in India by MPS Limited

Printed in the UK by CPI Group (UK) Ltd, Croydon
 Contents

About the Author xiii

Preface xv

1 Introduction 1
1.1 Scientific Computation, Numerical Analysis, and Engineering 2
1.2 Computational Electromagnetics 3
1.2.1 Applications of CEM Tools 5
1.2.2 Types of CEM Methods 5
1.2.3 Mesh and Grid Generation 6
1.3 Accuracy and Efficiency 7
1.4 Programing Languages 8
1.5 Writing and Debugging Numerical Codes 8
References 10

2 Fundamentals of Electromagnetic Field Theory 11

2.1 Electromagnetic Field and Source Quantities 11
2.2 Maxwell’s Equations 12
2.2.1 Constitutive Relations 12
2.2.2 Impressed and Induced Currents 13
2.2.3 Magnetic Currents 13
2.3 Coordinate Systems 14
2.3.1 Rectangular Coordinates 14
2.3.2 Cylindrical Coordinates 15
2.3.3 Spherical Coordinates 16
2.4 Gradient, Curl, and Divergence 16
2.5 Laplacian 18
2.6 Wave Propagation 18
2.7 Electromagnetic Boundary Conditions 19
2.8 Time- and Frequency-Domain Representations 20
2.9 Plane Waves 21
2.9.1 Wave Vector 22
2.9.2 Position Vector 22
2.9.3 Plane Wave in Vector Notation 23
2.9.4 Characteristic Impedance 24
2.10 Propagating, Standing, and Evanescent Waves 24
2.11 Bessel Functions 25
vi Numerical Methods for Engineering

2.11.1 Hankel Functions 26

2.11.2 Cylindrical Waves 27
2.12 Power and Energy 28
2.13 Initial Value Problems and Boundary Value Problems 29
2.13.1 Modes 30
2.13.2 1-D, 2-D, and 3-D Boundary Value Problems 30
2.13.3 Radiation and Scattering Problems 33
2.13.4 Inverse Problems 34
2.13.5 Green’s Functions and Radiation Integrals 34
2.13.6 Formulating and Solving Boundary Value Problems 35
2.13.7 The Equivalence Principle 36
2.14 Other Topics 37
Problems 38
References 38

3 Basic Numerical Tasks 41

3.1 Introduction to MATLAB Programing 41
3.1.1 Vectors and Arrays 42
3.1.2 Working with Plots 43
3.1.3 Scripts 45
3.1.4 Functions 48
3.1.5 Arguments and Structure Arrays 49
3.1.6 Speeding Up MATLAB Codes 49
3.1.7 Other MATLAB Commands 49
3.2 Numerical Differentiation 50
3.2.1 Code Example: Central Difference Rule 51
3.3 Numerical Integration 52
3.3.1 Code Example: Midpoint Integration Rule 53
3.4 Interpolation 54
3.5 Curve Fitting 54
3.5.1 Code Example: Polynomial Fitting 56
3.6 Newton’s Method 57
Problems 58
References 59

4 Finite Difference Methods 61

4.1 Basic Components of Finite Difference Solvers 61
4.1.1 Grid 62
4.1.2 Stencil 63
4.1.3 Boundary Conditions 63
4.1.4 Sources 65
4.1.5 Solution Method 65
4.2 Wave Equation: 1-D FDTD Method 66
4.2.1 1-D Grid 66
4.2.2 Update Equation for the 1-D Wave Equation 67
 Contents vii

4.2.3 Initial Condition 67

4.2.4 Boundary Conditions for the 1-D FDTD Method 68
4.2.5 Hard and Soft Sources 69
4.2.6 Source Turn-On Functions 70
4.2.7 Code Example: 1-D FDTD Algorithm 71
4.2.8 Stability 75
4.2.9 Accuracy 77
4.3 Laplace’s Equation: 2-D Finite Difference Method 80
4.3.1 Example: 2-D FD Method on a 4-by-4 Grid 83
4.3.2 Waveguide Modes 83
4.3.3 Spectrum of the Laplacian 85
4.3.4 Numerical Implementation 87
4.3.5 2-D FD for Transmission Lines with Dielectric Materials 87
4.4 2-D Finite Difference Time-Domain (FDTD) Method 89
4.4.1 2-D EM Problems 89
4.4.2 Yee Cell and 2-D FDTD Method for TM Polarized Fields 91
4.4.3 Dielectric and Conductive Materials 93
4.4.4 Anisotropic Materials 94
4.4.5 Stability Criterion 94
4.4.6 Boundary Conditions for the 2-D FDTD Method 95
4.4.7 Preprocessing 101
4.5 FDTD Modeling for Scattering Problems 102
4.5.1 Incident Field 103
4.5.2 Scattered Field 104
4.5.3 Scattered Field Formulation 104
4.5.4 Scattering Amplitudes, Scattering Width, and
Radar Cross Section 106
4.5.5 Bistatic and Monostatic Scattering 107
4.6 Postprocessing the FDTD Solution 108
4.6.1 Frequency-Domain (Phasor) Fields 108
4.6.2 Near Field to Far Field Transformation 110
4.6.3 Other Types of Postprocessing 116
4.7 Code Verification 118
4.7.1 Scattering by a Circular Cylinder 118
4.8 3-D FDTD Method 120
4.8.1 PEC Cavity 122
4.8.2 Stability Criterion 123
4.9 Summary 123
Problems 124
References 128

5 Numerical Integration 131

5.1 Types of Integration Rules 132
5.2 Newton–Cotes Quadrature Rules 133
5.2.1 Error Analysis of the Midpoint Rule 133
viii Numerical Methods for Engineering

5.2.2 Higher Order Newton–Cotes Rules 137

5.2.3 Extended Newton–Cotes Rules 138
5.2.4 Romberg Integration 140
5.3 Gaussian Quadrature 140
5.3.1 Orthogonal Polynomials and Gaussian Quadrature 142
5.3.2 Gauss–Legendre Quadrature (w(x) = 1) 145
5.4 Nonclassical Gaussian Quadrature Rules 146
5.4.1 Lanczos Algorithm 147
5.4.2 Weights and Nodes 149
5.5 Implementation 150
5.6 Other Methods for Singular Integrands 150
5.6.1 Singularity Subtraction 151
5.6.2 Transformation Rules 152
5.7 Multidimensional Integrals 154
5.7.1 Monte Carlo Integration 155
5.8 MATLAB’s Built-in Numerical Integration
Functions 156
Problems 157
References 159

6 Integral Equations and the Method of Moments 161

6.1 Integral Operators 161
6.1.1 First and Second Kind Integral Equations 162
6.1.2 Solving Integral Equations Numerically 162
6.1.3 Smooth Kernels and Operator Conditioning 163
6.1.4 Singular Kernels and Conditioning for Non-Fredholm
Operators 164
6.2 Integral Equations in Electromagnetics 164
6.2.1 Electric Field Integral Equation, 2-D Transverse Magnetic
Polarization (TM-EFIE) 166
6.2.2 Electric Field Integral Equation, 2-D Transverse Electric
Polarization (TE-EFIE) 168
6.2.3 Magnetic Field Integral Equation 169
6.2.4 Combined Field Integral Equation 169
6.2.5 Thin-Wire Integral Equations 170
6.3 Method of Weighted Residuals 173
6.3.1 Basis Functions 175
6.3.2 MoM Implementation 176
6.3.3 Mesh Types 177
6.4 Method of Moments for the TM-EFIE 178
6.4.1 1-D Mesh Generation for Simple Geometries 178
6.4.2 Path Integrals 178
6.4.3 Testing Integration 179
6.4.4 Source Integration 180
6.4.5 Incident Field 181
6.4.6 Physical Interpretation of the Method of Moments 181
 Contents ix

6.4.7 Mesh Element Density 182

6.4.8 MoM Code Overview 183
6.4.9 1-D Mesh Generation for Complex Geometries 184
6.4.10 Postprocessing 187
6.4.11 Scattering Amplitudes 187
6.4.12 MoM Implementation 190
6.5 Accuracy and Efficiency of the Method of Moments 192
6.5.1 Computational Cost 192
6.5.2 Error Analysis 194
6.5.3 Sources of Error 196
6.6 Dielectric Structures 204
6.6.1 2-D Volume Method of Moments 205
6.7 2.5-Dimensional Methods 207
6.8 3-D Electric Field Integral Equation 208
6.8.1 Rooftop Functions 209
6.8.2 Rao–Wilton–Glisson (RWG) Basis 209
6.8.3 Method of Moments 212
Problems 215
References 218

7 Solving Linear Systems 219

7.1 Linear Spaces and Linear Operators 219
7.1.1 Linear Spaces 220
7.1.2 Norms on Linear Spaces 221
7.1.3 Linear Operators 222
7.1.4 Operator Norms 222
7.1.5 Range, Null Space, and Rank 223
7.1.6 Operator Inverse and Adjoint 223
7.1.7 Classes of Operators 224
7.1.8 Eigenvalues and Eigenvectors 226
7.1.9 Field of Values 226
7.1.10 Matrix Decompositions 226
7.1.11 Other Matrix Formulas 229
7.2 Direct and Iterative Solution Methods 229
7.3 LU Decomposition 231
7.4 Iterative Methods 232
7.4.1 Stationary Iterations 232
7.4.2 Implementation of Iterative Algorithms 235
7.5 Krylov Subspace Iterations 236
7.5.1 Conjugate Gradient Method 236
7.5.2 Residual Error 238
7.5.3 Condition Number 239
7.5.4 CGNE Algorithm 240
7.5.5 Other Krylov Subspace Methods 240
7.5.6 Convergence of Krylov Subspace Iterations 241
7.5.7 Preconditioners 245
Visit https://textbookfull.com
now to explore a rich
collection of eBooks, textbook
and enjoy exciting offers!
x Numerical Methods for Engineering

7.6 Multiscale Problems 246

7.6.1 Fast Algorithms 247
7.6.2 Reduced-Order Representations 249
Problems 250
References 252

8 Finite Element Method 253

8.1 Variational Principles in Mathematical Physics 254
8.1.1 Operators and Functionals 254
8.1.2 Variational Principles 255
8.1.3 Variational Calculus 257
8.1.4 Euler–Lagrange Equation 258
8.1.5 Variational Principles for PDEs 260
8.1.6 Variational Principles for Self-Adjoint, Positive Definite
Operators 260
8.1.7 Functionals in Mathematical Physics 260
8.1.8 Rayleigh–Ritz Method 262
8.2 Overview of the Finite Element Method 264
8.2.1 Mesh Types and Mesh Generation 264
8.2.2 Basis Functions 265
8.2.3 Variational Principle and Rayleigh–Ritz Procedure 265
8.2.4 Linear System Solution 265
8.3 Laplace’s Equation: 1-D FEM 266
8.3.1 Functional Form of the Generalized
Laplace Equation 266
8.3.2 Mesh Representation 267
8.3.3 Rayleigh–Ritz Procedure 267
8.3.4 Element Stiffness Matrix 268
8.3.5 Basis Functions and Shape Functions 269
8.3.6 Evaluating the Element Stiffness Matrix 271
8.3.7 Assembly of the Global Stiffness Matrix 272
8.3.8 Example: Five-Element Mesh 273
8.3.9 Comparison of FEM and FD 274
8.3.10 Sparse Matrix and Dense Matrix Methods 275
8.4 Helmholtz Equation: 2-D FEM 275
8.4.1 Boundary Conditions for FEM 276
8.4.2 Rayleigh–Ritz Method for the Helmholtz Functional 278
8.4.3 Eigenvalue Problems (Unknown k) 279
8.4.4 Scattering Problems (Known k) 279
8.4.5 FEM Formulation of the 2-D Scattering Problem 280
8.4.6 Triangular Mesh 281
8.4.7 Basis Functions and Shape Functions 282
8.4.8 Evaluating Element Matrices 284
8.4.9 Matrix Assembly 286
 Contents xi

8.5 Finite Element Method–Boundary Element Method 287

8.5.1 Boundary Element Method 289
8.5.2 Implementation 293
8.6 Numerical Results 296
Problems 298
References 302

9 Optimization Methods 303

9.1 Introduction 303
9.1.1 Optimization Problems 303
9.1.2 Local and Global Optimization 305
9.2 Classes of Optimization Problems 305
9.2.1 Convex Optimization 306
9.2.2 Types of Optimization Algorithms 307
9.2.3 Common Optimization Algorithms 308
9.2.4 Gradient and Gradient-Free Methods 310
9.3 One-Dimensional Optimization 310
9.3.1 Golden Section Search 310
9.3.2 Tolerance Parameter 311
9.3.3 Inverse Quadratic Interpolation 312
9.3.4 Brent’s Method 312
9.4 Nelder–Mead Simplex Method 312
9.4.1 Initial Simplex 313
9.4.2 Simplex Transformations 313
9.4.3 Termination 315
9.4.4 Implementation Details 315
9.4.5 Numerical Example 316
9.5 Practical Considerations for Optimization 317
Problems 318
References 319

10 Inverse Problems 321

10.1 Introduction 321
10.2 Types of Inverse Problems 321
10.2.1 Inverse Scattering 321
10.2.2 Imaging 322
10.2.3 Inverse Source Problems 322
10.2.4 Design Synthesis 322
10.2.5 Applications 324
10.3 Ill-Posed Problems 325
10.3.1 Regularization 326
10.3.2 Resolution 328
10.3.3 Types of Inverse Scattering Methods 328
10.4 Born Approximation Methods 330
10.4.1 Iterated Born Approximation 333
xii Numerical Methods for Engineering

10.4.2 Diffraction Tomography 334

10.4.3 Holographic Backpropagation Tomography 335
10.4.4 Simulating Forward Scattering Data 337
10.4.5 Implementing the Holographic Backpropagation Tomog-
raphy Method 338
10.4.6 Numerical Examples of Holographic Backpropagation
Tomography 338
10.5 Regularized Sampling 340
10.5.1 Discretization of the Regularized Sampling Equation 342
10.5.2 Implementing the Regularized Sampling Method 343
10.5.3 Numerical Examples of the Regularized Sampling Method 344
Problems 346
References 347

Index 349
 About the Author

Karl F. Warnick is a professor in the Department of Electrical and Computer Engi-

neering at BrighamYoung University, USA, and a fellow of the IEEE. He has published
widely on electromagnetics theory, numerical methods, antenna applications, and
high sensitivity phased arrays for satellite communications and radio astronomy.
This page intentionally left blank
 Preface

This book evolved out of my experience that well-prepared undergraduate engineer-

ing students, as well as all beginning graduate students, can learn to understand and
implement powerful numerical techniques if the topic is presented with clear explana-
tions, good examples, and problem sets that move from simple algorithms to complex
codes with real-world capabilities. I undertook the challenge of developing a good
set of class notes and refined them based on significant feedback from students with
a variety of specializations, research interests, and career objectives. After nearly a
decade of iterations on the course notes followed by external reviews from instructors
who class-tested advance versions of the book, I feel confident that the finished book
will be a welcome self-study tool and course text.
The book has two major objectives:
1. To impart the skill of taking a mathematical prescription for a numerical method
and translating it into a working, validated software code.
2. To help users of numerical modeling and design software understand the rel-
ative merits, ranges of validity, domains of applicability, expected accuracy,
and underlying properties of several popular classes of numerical algorithms for
engineering problems.
Since design work often requires the creation of specialized software tools, engineers
must have the ability to implement numerical solution algorithms based on a mathe-
matical model for a system or physical problem. Even when using commercial design
packages, it is important to understand how to use the software to obtain sufficiently
accurate results within an acceptable amount of computation time.
The examples used in the book are chosen primarily from wave propagation and
electromagnetics for a reason. Of all the engineering fields, electromagnetic analysis,
and design tends to use the most sophisticated algorithms. This focus ensures that
the book touches the state-of-the-art in numerical methods for engineering. In keep-
ing with the emphasis on computational electromagnetics, a secondary goal of the
book is to help students better understand the electromagnetic theory by implement-
ing numerical solution methods for the fundamental equations of electromagnetics.
One of the best ways to become competent in working with the laws and principles
of electromagnetics, as well as to gain insight into the behavior of field and wave
solutions, is to translate the governing equations of electric and magnetic fields into
a computational algorithm and then to use the algorithm as a tool to visualize fields
and waves in conjunction with practical examples.
xvi Numerical Methods for Engineering

Although the emphasis is on computational electromagnetics, numerical tools

are developed in this book for basic building block tasks such as solving differential
equations, evaluating integrals, and solving linear systems. This means that much
of the material can be applied to all branches of engineering. General principles
of numerical analysis such as efficiency, accuracy, and ill-posed problems will be
treated within the context of computational electromagnetics, but these also have
broad application in computational science and engineering.
As a textbook, the material works best at the senior year after a first course in
basic electromagnetics or as a first-year graduate course. This book bridges the gap
between typical one-term introductory electromagnetics and the advanced computa-
tional electromagnetics textbooks that are not very accessible to seniors or beginning
graduate students. The balance of electromagnetics-specific content with general top-
ics is intended to serve engineering students specializing in electromagnetics, wireless
communications, optics, and related fields as well as students concentrating in other
areas who desire an exposure to numerical methods.
The electromagnetics content of the book is fairly self-contained. To avoid too
much distraction from the central topic of numerical methods, topics from electromag-
netic theory are presented in a condensed form without lengthy development. Chapter
2 includes a review of the basic concepts of electromagnetics. More advanced material
on wave propagation, radiation integrals, Green’s functions, propagation in materials,
waveguides, and scattering theory are later introduced as needed throughout the book.
Chapters 3–8 cover finite difference methods, numerical integration, the method
of moments for integral equations, linear system solution algorithms, variational
methods, and the finite element method. Chapters 9–10 on optimization and inverse
problems may be omitted or briefly surveyed in a senior undergraduate numerical
methods course. These topics are included to enrich the material for more advanced
graduate students.
Homework problems in the book are programmed, in the sense that groups of con-
secutive problems often build incrementally toward a completed code. It is common
for the core of an algorithm to be easy to implement in relation to the preprocessing
required to create a grid or mesh and postprocessing used to compute-derived physi-
cal quantities. Consequently, one or two homework problems typically deal with the
core algorithm. Several successive problems add additional pre- and postprocessing
capability. This incremental approach eases the development of complex codes and
implicitly teaches the vital debugging skill of verifying a small portion of the code
before adding another major functional block.
Another important technique used in the homework problems is cross-code com-
parison. Many of the physical problems are repeated from one chapter to the next, and
homework problems are included to compare numerical results from multiple algo-
rithms for the same physical system. This helps students to understand the relative
merits of different algorithms in terms of accuracy, efficiency, and ease of imple-
mentation. It is possible to explain in words why the method of moments for surface
integral equations can, in some cases, be more efficient than the finite difference
time-domain method. It is a more powerful learning experience to run codes for
 Preface xvii

both methods and see results from one of them that are more accurate and take less
computation time.
Looking beyond the immense practical value of numerical methods for engi-
neering work, numerical codes, and algorithms are wonderful tools for learning
engineering electromagnetics. Assessment data at my university indicate that our
beginning electromagnetics course is one of the most difficult in the electrical engi-
neering program. This can be attributed to the large amount of content that must be
packed into a single semester course and to the abstract nature of electromagnetic the-
ory in relation to, say, circuit analysis. I have observed many students who leave the
beginning electromagnetics course with a marginal understanding of the principles
of electromagnetics make their way through a course on computational electromag-
netics based on the material in this book. After implementing numerical algorithms
on their own and seeing the results appear in graphical form, the light turns on, and
they finally feel that they understand something of what the equations and theory of
electromagnetics are all about.
Some of the numerical methods covered in the book require considerable math-
ematical detail. To help the student distinguish supporting material from the primary
formulas needed for homework solutions, in this second edition, key results required
to implement the methods in code are set off using colored boxes. To further aid the
student, implementation helps; summaries of algorithmic steps, code fragments, and
examples are also included in colored boxes.
I appreciate the many outstanding students who have used the earlier versions
of this textbook in courses and have helped in many ways to refine the material,
including particularly Clayton Davis and Derek Hudson. I thank my colleagues who
class-tested preliminary drafts and the reviewers for insightful comments, corrections,
and valuable suggestions. Jakob Kunzler did a remarkably thorough reading of an early
draft of the second edition, even locating errors that had escaped many readers over
years of using the first edition. I welcome and appreciate the readers’ and instructor’s
feedback to continue improving the text.
Karl F. Warnick
Brigham Young University
Provo, Utah, USA
This page intentionally left blank
 Chapter 1
Introduction

This book is about numerical methods that allow the differential equations, inte-
gral equations, and other mathematical relationships that govern wave propagation,
electromagnetic (EM) fields, and other physical systems to be solved using com-
putational algorithms. The progress of numerical methods has followed the rise of
low-cost, widely available computational power. Numerical methods are now funda-
mentally important in engineering work, since almost all branches of engineering rely
on software tools, design packages, and numerical models. Circuit design, antenna
design, microwave systems, optical systems, remote sensing, control theory, signal
processing, and microelectronics are just a few of the fields that are heavily impacted
by software tools and computer-aided engineering. The ability to understand, use, and
create numerical methods is already crucial for engineers and will continue to grow
in importance.
In this book, we will survey general numerical methods as well as specialized
algorithms, with a focus on examples from computational electromagnetics. General
numerical tools and techniques to be covered include numerical integration, differen-
tiation, the solution of large linear systems, and methods for discretizing and solving
partial differential equations and integral equations. Finite difference methods, the
method of moments, and the finite element method will be treated in detail. While the
emphasis is on wave propagation and electromagnetics, the principles and techniques
of numerical analysis that are illustrated throughout the book can be applied to many
branches of engineering.
The goals are to teach the reader how to design algorithms that solve a given
set of equations or analyze a particular system and to help users of computer-aided
engineering software tools understand how to choose the best analysis technique,
interpret numerical results, and optimize the efficiency and accurate of the solu-
tion method. Another expected learning outcome is a deeper understanding of the
characteristic behaviors of solutions to the differential and integral equations of
electromagnetic field theory. Studying numerical methods has a double benefit for
engineering students, since they not only become competent at creating and using
numerical techniques but also gain insight into the fundamental laws and equations
that govern the systems used by engineers in creating new technologies.
The development of numerical methods has become an academic area in its own
right. When the emphasis is on modeling physical systems, the field is referred to
as scientific computation, and the theoretical study of the properties and behavior of
computational algorithms using rigorous mathematics is numerical analysis. We will
Visit https://textbookfull.com
now to explore a rich
collection of eBooks, textbook
and enjoy exciting offers!
2 Numerical Methods for Engineering

begin our thinking about numerical methods by considering the relationships between
these fields.

1.1 Scientific Computation, Numerical Analysis,

and Engineering

Science has historically been divided into two major areas of inquiry: theory and
experiment. In the last few decades, a third area of effort has become prevalent. This
new area is the application of numerical algorithms implemented in computer code
to model physical systems and is commonly referred to as scientific computation.
Scientific computation combines aspects of theoretical analysis, since it involves the
solution of differential equations or other descriptive frameworks, and experimental
science, since a system can be modeled under a variety of stimuli as would be done
when conducting an empirical study. Computational modeling is a growth industry in
nearly all fields of science, from simulation of quantum field theories for subatomic
particles to numerical weather prediction.
Since engineering is the application of physical principles to the design of new
technologies, the methods and techniques used to analyze physical systems from a
scientific perspective become the tools used by engineers. Numerical methods have
become vital in engineering analysis and design, and in recent years engineering
applications have driven a flowering of the power, capability, and breadth of algorithms
and commercially available software packages for analysis and design of uncounted
types of technologies.
As scientific computation and computer-aided engineering have increased in
importance, a new branch of mathematics dealing with the development and improve-
ment of solution algorithms has grown and matured. This field is referred to as
numerical analysis. Numerical analysis is concerned with creating algorithms that
are coupled with rigorous theoretical proofs of theorems that bound the computation
time required to obtain a solution for a given class of problems and that guarantee the
accuracy of computed solutions.
Historically, there has been a fair amount of interplay between engineering appli-
cations of computational algorithms and numerical analysis, but for the most part the
gap between the two areas has been unfortunately wide. Developers working in appli-
cation areas are driven by market forces to develop new numerical methods more
rapidly than rigorous theory can keep up, so the tools used in engineering work typi-
cally can be quite sophisticated yet unsupported by careful theoretical proofs of valid-
ity and accuracy of solutions. Rigorous mathematical analysis often leads to insights
that drive the state-of-the-art forward, but theoretical solution convergence proofs
often apply only to simplistic problems and some mathematically sophisticated numer-
ical methods can be inadequate for complex, real-world applications. Another barrier
between numerical analysis and the engineering world is abstract mathematical nota-
tion, which can disguise simple problems and methods that should be familiar to the
engineer and hinders the dissemination of new methods to application-oriented users.
To bridge this gap, this book attempts to balance theoretical ideas from numer-
ical analysis with practical examples from scientific computation and engineering.
 Introduction 3

Interestingly, the more the engineer is focused on applications, the more the study of
numerical methods should focus on theory. An engineer doing design work typically
uses a commercial software package, so the basic work of implementing mathemat-
ical equations as a computer algorithm has already been done. What the practicing
engineer needs is a feel for the expected accuracy of the results and insight into the
best way to use the software to get optimal results. Gaining this kind of understanding
requires only a modest amount of experience in writing numerical codes, and the
emphasis should be on accuracy analysis, efficiency, and comparisons of the perfor-
mance of different algorithms for the same problem. For most engineers, the right
mix is an equal combination of numerical analysis (theory) and practical algorithms
(applications).

1.2 Computational Electromagnetics

The development and application of numerical methods for problems involving
electromagnetic fields and waves is referred to as the field of computational elec-
tromagnetics (CEMs). To place CEM techniques in a historical context, we will
briefly survey the development of solution methods for wave propagation and
electromagnetic systems.
As with most areas of physics, engineering, and applied science, the basic trend
in solution methods for electromagnetics applications over the last century is from
“pencil and paper” analysis techniques to powerful, sophisticated software pack-
ages. In the early days of radar and wireless communications, engineers relied on
the analytical techniques—series expansions, canonical problems, and perturbation
methods—developed in the late 1800s and the early 1900s. In spite of the great clev-
erness of researchers in applying these methods to an expanding range of problems
with increasing complexity, eventually the wide range of EM applications required
the development of more flexible and powerful techniques.
Asymptotic methods based on the high- and low-frequency limits provided a
way to analyze problems involving structures and geometries that are too complex for
analytical solutions. Prominent among these developments was Ufimtsev’s creation
of the uniform theory of diffraction [1], which enabled advances in low-observable
stealth technology for aircraft. Asymptotic methods are more flexible than analytical
solutions and are vital for many engineering applications, but the accuracy of the
approximation is difficult to improve for structures that are geometrically complex or
otherwise fall outside the range of validity of the asymptotic approach.
More recently, the explosion of wireless technologies, medical imaging, optics,
terahertz devices, local area networks, mobile handheld devices, and a host of other
applications have required still more powerful methods for analysis and design, and
analytical and asymptotic methods have been augmented or replaced by CEM tools.
In the mid-1960s, the three pillars of numerical analysis in EMs were introduced: the
method of moments (MoM), the finite difference time-domain (FDTD) method, and
the finite element method (FEM). These breakthrough algorithms precipitated the
present era of computational electromagnetics and are part of the broad trend toward
joining simulation with theory and experiment throughout science and engineering.
4 Numerical Methods for Engineering

Analytical, asymptotic, and numerical methods that have been developed for
electromagnetic modeling include the following:

Early 1900s: Analytical Methods

● Canonical problems and separable geometries (e.g., spheres, cylinders,

rectangles, and cubes)
● Modal solutions, orthogonal functions, and series expansions
● Restricted to a few classes of geometries

Mid-1900s: Asymptotic Methods

● Low-frequency regime (quasistatic approximations)

● High-frequency regime
Geometrical optics and ray tracing
Physical optics
Geometrical theory of diffraction (GTD), physical theory of diffraction
(PTD), uniform theory of diffraction (UTD)
Wentzel–Kramers–Brillouin (WKB) approximation
Incremental length diffraction coefficients (ILDCs)

Late 1900s–Present: Numerical Methods

● Mode matching (“piecing together” analytical solutions)
● Finite difference time-domain method (FDTD), Yee, 1966 [2]
● Method of moments (MoM), Richmond, 1966 [3]
● Finite element method (FEM), 1968 [4,5]
● Hybrid methods that combine “full-wave” numerical methods like FDTD and
MoM with asymptotic methods

Numerical methods are highly flexible and can be used to analyze and design
an unlimited variety of microwave systems, antennas, components, and structures,
but methods of all types, including analytical techniques, are very important and
widely used today, and there is much overlap among these three broad areas. Even
with powerful computational methods, analytical and asymptotic solutions are very
important. They provide qualitative physical insight for complex problems, lead to
simple formulas that are useful in design work, and, as will be seen in later chapters,
are used as test cases to validate numerical methods.
 Introduction 5

1.2.1 Applications of CEM Tools

Software analysis and design tools based on CEM algorithms are useful for reducing
product development costs by replacing costly device testing with computational
simulation, design optimization, and integration of electromagnetic phenomena with
other system aspects in a multiphysics environment. A key goal is to replace the
design–build–test cycle with a design–simulate cycle, to reduce fabrication costs,
make more efficient use of finite resources, and shorten the time to market for product
development.
CEM methods are used in application areas such as the following:

Applications of Computational Electromagnetics

● Microwave circuit and system design

● Antenna analysis and design
● Wireless communications
● Propagation of electromagnetic waves in various environments (e.g., indoor,
outdoor, urban locations, underground, over the ocean surface, in atmospheric
layers, embedded in biological tissues or in the body)
● Optics and photonics
● Design and analysis of metamaterials
● Radar systems
● Stealth and low observable technology
● Remote sensing of oceans, ice caps, land, and other planets
● Electromagnetic compatibility and interference (EMC/EMI)

Since most CEM tools are essentially solution methods for systems of differential
equations or integral equations, the basic techniques can be applied to many other
fields beyond those directly associated with electromagnetic theory and its appli-
cations. CEM methods are related to those used in such broad-ranging fields as
structural analysis, mechanics, fluid dynamics, acoustics, microelectronics, weather
and climate models, galactic dynamics, simulations of colliding black holes, and
financial modeling.

1.2.2 Types of CEM Methods

There are four main categories of numerical methods used in electromagnetics:
differential equation methods, integral equation methods, mode matching, and
asymptotic approximations. Differential equation methods solve Maxwell’s equations
directly by transforming the continuous differential equations into discrete differ-
ence equations. Integral equation methods use the equivalence principle, Green’s
functions, and boundary conditions to transform Maxwell’s equations into an integral
equation for an unknown current. Asymptotic approximations are based on the
6 Numerical Methods for Engineering

high- or low-frequency limiting behaviors of fields and waves, and mode match-
ing is based on piecing together analytical solutions for simple, canonical problems
to analyze more complex structures.
Prominent algorithms in these classes of methods are outlined as follows:

CEM Algorithms

Differential equation methods

• Finite difference (FD) method
Finite difference time-domain (FDTD) method
• Finite element method (FEM)
• Finite integration method
• Transmission line matrix (TLM) method
• Method of lines
Integral equation methods
• Method of moments (MoM)
Mode matching

Numerical methods based on asymptotic approximations

• Ray tracing (geometrical optics)
• GTD, PTD, and UTD codes and hybridizations with other algorithms
such as MoM

The methods that we will focus on in this book are emphasized in color.
Numerical methods are also specialized to solving single-frequency (time-
harmonic) problems and broadband (time-domain) problems. The major methods
in these categories are given in Table 1.1. The MoM–time-domain integral equa-
tion (MoM-TDIE) approach is a relatively new development, because only recently
have instability problems been overcome sufficiently to allow for useful implemen-
tations. In this book, we will consider both time-domain methods (FDTD) and
frequency-domain methods (MoM, FEM).

1.2.3 Mesh and Grid Generation

CEM methods require some kind of discrete computational representation of the
problem to be solved. Typically this is a regular grid of points or mesh of triangular
patches, cubes, or other simple shapes, together with parameter values that specify
the physical properties in the material in each mesh element. As will be seen in

Table 1.1 Algorithms for time- and frequency-domain problems

Algorithm/Domain Frequency domain Time domain

Differential equation methods FD, FEM FDTD

Integral equation methods MoM MoM-TDIE
Random documents with unrelated
content Scribd suggests to you:
 Yearling moose.

Rutting activities begin by the first of September and continue 17

into October. The antlers of the bulls have reached full size and
hardened by the end of August, at which time the bulls begin to rub
off the velvet, the skin that has covered the growing antlers. Saplings
and brush are thrashed with great vigor, and this activity continues
long after the antlers have been cleaned. Apparently it serves as one
of the outlets for the strong rutting emotions. The bulls soon begin to
spar and to determine who is boss over whom. And they begin to
seek the cows. A successful bull usually has but a single cow, and he
follows her closely as she moves about in her feeding. During the rut
he utters at intervals a deep grunt. The cow, apparently when in an
emotional state, utters a drawn-out wailing call.

The one or two calves are generally born in late May or early June.
They are reddish without spots. The mother must sometimes protect
her calf from prowling grizzlies and this she generally seems fully
capable of doing, judging from incidents in which the bear is chased
away by an infuriated mother. A large male grizzly, however, is
apparently not easily discouraged. By autumn the calves have made a
surprising growth and have a new coat that resembles that of the
adults. They remain with the mother until near the time for a new
calf, when she no longer tolerates their presence.

The moose is a wilderness animal, requiring for his haunts big

country. The picturesque bull, silhouetted on a hill or on a lake shore,
adds repose and serenity to the wilderness.

Dall Sheep
Ovis dalli

The Dall or white sheep is one of the outstanding wildlife features of

the park. The north side of the greater part of the Alaska Range is
excellent sheep habitat. Within the park the most extensive sheep
country extends from the Nenana River to the Muldrow glacier, a
distance of about 70 miles by road.

Most of the sheep spend the winter north of the road. This is
favorable winter range because the snowfall is relatively light and
strong winds keep the exposed ridges free of snow.

Many sheep remain on the winter range all year, but more of them
migrate toward the heads of the rivers in May and June. In making
the migration, the sheep must in places cross 2 or 3 miles of low
country. They are fully aware of their vulnerability to grizzlies and
wolves in these crossings. Before venturing away from a safe take-off
ridge, they may scrutinize the low country for a day or two, until they
feel that no danger lurks along the way. A band of 60 or 70 sheep
may move across slowly in a rather compact group; at other times
urgency replaces caution and they frequently break into a hurried
gallop. Having reached the safety of rough country again, the sheep
may gambol about as though the weight of tension has suddenly
been lifted. The return migration is made in August and September.

The large amber-colored horns of the rams with transverse 18

ridges and sweeping outward curl have a rugged, graceful
beauty. They may spread widely at the tips or curl rather close to the
head. The ewes are less imposing. Their horns are slender spikes that
extend upward in a slight curve, resembling those of the mountain
goat but they lack the shiny jet black color and are not as sharp. The
horns are never shed and continue to grow throughout the sheep’s
life span of 11 to 14 years. The growth is slight during the later
years. Growth takes place during the summer when food is highly
nutritious. In winter only a groove or ridge encircling the horn is
formed. By counting these annual rings the age of a sheep can be
determined.

For detecting danger the sheep depend on their sharp eyes. They
appear to disregard scent which for many animals is the final decisive
word on any situation. But this seems quite logical, because the
sheep generally have a strategic view, and in the varying air currents
no dependence can be placed on getting scent messages. Noises are
considered rather unimportant unless the sheep have already caught
a glimpse of movement nearby.

To approach sheep for photography it is usually best to move slowly

toward them from below with no attempt to hide. They at once
become suspicious if they glimpse someone stalking. However, I have
at times stalked sheep where the opportunity for undetected close
observation was obvious. On one occasion, from a ragged rocky ridge
top, I spent most of an afternoon watching a band of rams some 50
yards away without being discovered. Some bands are wilder than
others and the same band does not always behave uniformly. A band
that has rested and is ready to move may take your approach as an
excuse for a romp.

The food of sheep consists of grasses, herbaceous species and

browse, chiefly willow. Scattered over the range are a number of salt
licks which the sheep seek for minerals.

The most active mating period extends from about the middle of
November to the middle of December. The rams who have been
fraternizing on friendly terms for many months, now and then
showing mating behavior such as gentle joustings, begin to take
greater notice of the ewes. The old rams continue to associate, but
now serious battling takes place. The fighting follows rather uniform
conventional rules. The two matched battlers move apart several
yards, then, as though by a signal, they turn and face each other and
at the same time rise up on their hind legs, then charge full speed at
each other, their horns crashing together with a loud thud. If the
joust is even, they may repeat the performance until the superiority
of one of the combatants is evident. But there is some tolerance
among the rams, for two or more may breed with ewes in a band
indiscriminately.

The numbers of sheep on a range under natural conditions may vary

considerably. In the park, a very high population, possibly as high as
5,000 or more, suffered severe losses during 1929 and again in 19
1932, due to extremely deep snow conditions, and an icy crust
in the latter year. In 1945 the population was down to about 500.
Since 1945 there has been a steady increase. The numbers in 1959
were up to about 2,000.
 Dall sheep; rams on the Alaska Range in summer.

Sheep are subject to wolf predation, especially when the numbers are
so high that part of the population must graze on hills too gentle for
safety. Sheep legs are strong and sturdy but for their effective
functioning steep country is needed. The steep terrain is, so to speak,
part of their legs. In my studies in the park, the losses showed that it
was the very old, the ailing, and the lambs in their first winter that
were most vulnerable to predation. The lynx (when rabbit numbers
have crashed and these animals have become scarce), wolverine, and
grizzly may capture an occasional sheep but their effect is
unimportant. The golden eagle may capture an occasional young
lamb, but all my observations and food-habit studies indicate that any
eagle predation that takes place is insignificant. If the park is large
enough to support the sheep and their predators (natural conditions),
we have a situation ideal for the future of the sheep.

Mountain sheep have a high esthetic appeal. In part this may 20

be due to their setting, for we associate them with their
beautiful haunts, the precipitous cliffs and ledges intermingled with
green slopes spangled with flowers. This is idyllic country in which to
hike and climb. Here we encounter the golden eagle who shares the
ridge tops with the sheep; the wheatear, who comes all the way from
Asia to nest; the gray-crowned rosy finches; the flashing black and
white snowbirds nesting in rock crevices; and the surfbird that has
left the ocean beaches to nest in these remote mountains. And up
high, the saxifrages, delicate yellow poppies, forget-me-nots and
spring beauties add color to it all.

Mountain Goat
Oreamnos kennedyi

On May 27, 1955, a goat was discovered on Igloo Mountain on the

slope directly above the cabin I was occupying. It remained on the
mountain for 3 weeks before wandering away. It has not been seen
since. This is the only verified record for the park. But two road men
reported seeing a goat cross the road at Mile 3, on August 8, 1950. I
believe this to be a good record because both men are reliable
observers. In the fall of 1950 a goat was shot at Cantwell, not far
from the park boundary.

The nearest known goat range is about 60 miles from the park in the
Talkeetna Mountains. The goats that reached the park may have
been sporadic wanderers for it is not unusual for goats to
occasionally wander 25 or 30 miles from their known ranges. On the
other hand, it is possible that the goats are expanding their range
toward the park. In 1959, I was told that a band of a dozen goats
had been reported at the head of Jack River where they had not
been reported before. Jack River lies between the goat range and the
park.

Identification is not difficult. The goat’s horns are short, slightly

curved spikes, similar to the horns of the female sheep, but shiny
black and smooth rather than grayish and rugose. The goat’s chin
whiskers are identifying, as is the shoulder hump, and the knee
length pantaloons of long hair. Also the goat’s face is noticeably
longer than that of sheep. The goat sexes are similar.

It is not unlikely that goats will continue to be occasionally seen in

the park. Any lone “sheep” might turn out to be a goat.

21

Alaska wolf.
From a color sketch by William J. Berry.
 Wolf
Canis lupus pambasileus

Wolves vary considerably in size and color. The average male weighs
about 100 pounds and the female somewhat less, about 85 pounds.
Their color may be almost white, black, gray, or brown. Most wolves
in interior Alaska are either black or brownish like a coyote. The facial
markings show some variation and there may occasionally be
noticeable patterns over the rest of the body. A few wolves have a
blackish saddle; one that I knew had a black robber-mask across the
eyes. Individual disposition and behavior also varies. A handsome
male had an extra touch of spirit in his gallop; a male parent had a
dour expression and seemed, to my imagination, weighted with care.
Wolves raised in captivity from puppyhood are extremely friendly.

For a den, the wolf considers an enlarged fox burrow both convenient
and suitable. Dens have been found in a variety of situations. One
was located on a wooded rock bluff, another was beyond timber near
the top of a bluff bordering a river, and one was on a wooded island
between old river channels. The four to six young, probably the
average size of litter, are born the early part of May. The mother
remains at home with the pups and the male provides the victuals.

At one den that I observed closely there were two extra males and
an extra female with the pair. These wolves all fraternized in the most
friendly manner. Before departing for the night hunt, the five would
sometimes assemble in a close group, wag tails and frisk about, and
sitting on haunches sing in chorus. Later in the season this group of
five adults was joined by two additional males.

The following year the same pair returned to the den. They were
accompanied by one of the extra males that had been at the den the
previous year. The extra female and one of the bachelors set 22
up their own housekeeping farther down the river. But when
their pups were large enough to travel, they all came up the river and
joined the original pair. Young and old combined added up to 15
wolves. This wolf pack was composed of two pairs, an extra adult,
and 10 pups. Some of the extra wolves of the previous year were not
seen—they may have been trapped or poisoned beyond the park
boundaries during the winter months.

The wolf’s food varies with the seasons and the prey species
available. When voles and lemmings are plentiful, the wolves may
spend hours in the grass and sedge areas pouncing on them. During
the summer months the ground squirrel has at times been one of the
more important food items. Occasionally an unfortunate marmot is
surprised and in years when rabbits are plentiful, the snowshoe rabbit
becomes a food source. I have found remains of several porcupines
eaten by wolves, the spine-covered hide neatly inverted.

But the wolf also, and primarily, feeds on the ungulates—the

mountain sheep, caribou, and moose. Under natural conditions the
relationship between the wolf and these prey species is old and tried.
There is the aphorism, “nothing in nature offends nature.” In the
hunting of these animals the wolf appears to be an evolutionary force
in that there is a tendency for the weaker individuals to succumb.

In spring the wolf hunts the caribou calves, which early develop
surprising speed, so that when a wolf chases a group of caribou, the
calves race along with the adults. But after a time a weak calf, one
not up to the others in endurance, may begin to drop behind, and it
is this weak individual that is overtaken, an example of the
elimination of the weak, the survival of the fittest. In the winter
hunting, the old and weak animals are the most susceptible. It is a
struggle, a testing for all, but through the ages, the sheep, moose,
and caribou have survived and come down to us adapted to their
particular way of life, with the wolf as one of the environmental
factors.

At McKinley we have an opportunity to preserve a northern flora and

fauna. But the future of the wolf is precarious because the home
range of the park wolves extends beyond park boundaries into
territory where the wolf has no protection, where there is a bounty
on his head. The silencing of the longdrawn call of the wolf would be
a tragic loss to the human spirit.

Coyote
Canis latrans incolatus

So far as known, the coyote has always been rare out in the park.
Along the Nenana River, however, I frequently have heard his song.
Here he seems to find conditions more favorable for his way of life.
Perhaps it is the presence of the snowshoe rabbit in this low brushy
country that attracts him.

23

Coyote.
 Red fox.

The coyote weights about 25 pounds on the average. His color 24

is brownish with black-tipped hairs intermingled. Color variation
in coyotes is so slight it is not noticeable in the field; he does not
have the black, whitish and various color patterns that are present in
the wolf. The muzzle is long and pointed, ears well developed, eyes
sharp.

As a field biologist I have had an opportunity to observe coyotes in

many regions. In Yellowstone I made a 2-year study of its
relationships with other animals because it had been feared by some
that he would destroy the antelope, bighorn, and deer, if not
controlled. The study showed that the coyote there lived chiefly on
meadow mice and pocket gophers in summer, and carrion in winter,
and that he had no harmful effect on the large ungulates.
On the San Carlos Indian Reservation in Arizona the coyote was
blamed for cattle losses. Here a study showed that the basic cause of
losses was over-use of the range and that where grazing was good all
losses were insignificant. Cattlemen are finding this true and are
beginning to appreciate the usefulness of the coyote as a curbing
influence on rodent depredations. In addition to a meat diet, which
includes great quantities of grasshoppers in season, the coyote feeds
extensively on fruit. On Isle Royale, in Lake Superior, I found it
feeding on sarsaparilla berries; in Jackson Hole, Wyoming, on
silverberry and quantities of haw; in Arizona, on manzanita and
juniper berries, the latter being the winter staff of life.

The coyote is best known for his song, which in all its variations,
symbolizes the spirit of wildness and remote country. J. Frank Dobie
in his The Voice of the Coyote expresses the sentiment of many when
he writes: “I confess to a sympathy for the coyote that has grown
until it lives in the deepest part of my nature.”

Red Fox
Vulpes fulva alascensis

The fabled red fox is abundant, widely distributed over the park and
frequently seen. Silver, cross, and red color phases, along with some
intermediate variants, are well represented, and two or three of these
types frequently show up in a single litter. The prominent white tip on
the tail distinguishes the fox from the coyote and wolf.

Hundreds of dens are scattered over the countryside, many more

than are used in any one season. They are located indiscriminately in
spruce woods and out on the open tundra miles from the nearest
tree. Each pair has a selection of old dens to choose from, but often
they occupy favorite sites year after year. During a season, a family
sometimes moves from their first choice to a den nearby. The 25
connecting burrows of a den usually have 5 or 6 entrances and one I
examined had 19.

The young are born the early part of May. By June the blackish, blue-
eyed, chubby pups may be seen walking about clumsily. At this age
they are nursed in the open. One mother that I often watched almost
always nursed her five pups from a standing position. Only twice did I
see her lying on her side to nurse. As the pups grow they become
slimmer and the eyes turn brown and the coat color changes so that
the different color types can be identified. Although nursing seems to
cease toward the end of June, the cubs remain at the den into
September.

While the male travels far to hunt, the vixen remains at home to
watch over the pups. Most of the time she is curled up at the den or
perhaps a hundred yards away on a prominence. When she wishes to
nurse the pups or give them food, she puts her head in the mouth of
a burrow and calls softly “mmmmm,” “mmmmm.” If they do not
come forth, she may go to another entrance and call. But usually
they respond at once. A sharp, guttural “klung” has the opposite
meaning; when the pups hear this warning they scurry into a burrow.
This command is often given after the pups have nursed and the
mother wants to go off a short distance to lie down.

The female exhibits extreme friendliness toward her mate. When he

returns to the den after an absence she greets him with tail-wagging,
face-licking and much wriggling of her body. He is less demonstrative
and acts tired, which he probably is after several hours of hunting.
She picks up his offerings—usually mice or ground squirrels—eats her
fill or, if not hungry, carries the booty to the burrows and calls to the
young. He moves off to one side to rest. She sometimes watches for
his return from various points. One evening a vixen impatiently
moved from one lookout to another for over 2 hours before the male
arrived and received her warm welcome.

Mice and lemmings are the staple all-year food, but in summer the
ground squirrel may make up about half the diet. When snowshoe
hares and ptarmigan are plentiful, they become prominent in the
diet. Carrion is especially attractive in winter, and the fox attends
carcasses and curls up on the snow to wait until the wolves and
wolverine have eaten. He robs wolf and wolverine caches, and he
sometimes has his own caches robbed.

In the latter part of July the foxes go berrying, for they are fond of
fruit. Blueberries and crowberries are everywhere available for the
picking. The berries are also eaten in winter sometimes, it is said,
quite extensively when mice are scarce.

Foxes appear to be well able to take care of themselves. They can

outrun the grizzly, wolf, and wolverine. When the golden eagle
swoops at him he stands on watch with his bushy tail erect and
straight as a ramrod. The eagle dares not strike.

26

Lynx
Lynx canadensis

The lynx manner is one of independence, confidence and

complacency. He walks through the woods with dignity, looking
neither right nor left. Of course, he is not as oblivious as he appears
to be. He may even stop to watch you, but only briefly, and then he
continues sedately on his way.

His long legs are thicker than seem necessary to support the lean
body, but they are no doubt valuable for long jumps and pouncing.
The large, widespreading feet serve him well as snowshoes, and
strong, curved claws enable him to scramble readily up a tree. The
eyes are startlingly big and yellow and the throat ruff gives to the
face a squarish look. Long, glistening black tassels adorn the ears.
The stub tail, about 4 inches long and tipped with black, serves to
register emotions. The winter fur is soft and grayish, with few
markings except for the facial pattern; the summer coat is more
tawny.

Nature has bestowed on the lynx a snowshoe fixation so that he

spends his nights and days thinking and dreaming of rabbit dinners.
So dependent has he become on the rabbit for his main course that
his numbers flourish and wane in the wake of rabbit statistics.

During the period between 1954 and 1956, when lynx were abundant
in the park, I made a study of their food habits by analyzing several
hundred lynx scats. In addition to rabbits, the lynx had fed
considerably on ptarmigan and in summer on ground squirrels. This
part of the diet increased as the rabbits decreased. But with the
decline of the rabbits, the supplemental foods did not suffice to
maintain the population, and the lynx became scarce.

Lynx.

In the winter of 1907-08, Charles Sheldon noted two instances 27

of lynx preying on sheep. The rabbit population had crashed
and the lynx had turned to other sources for survival. One lynx that
made its attack on a sheep from ambush found the prey rather large,
for in the ensuing struggle he received some severe bruises. He
apparently was driven to hunting animals out of his class. About 2
years after rabbits disappeared in the Kuskokwim River region a
number of years ago, lynx did some preying on reindeer in winter by
leaping on their backs and biting the neck. The lynx were said to
have attacked the reindeer only that one winter. During periods of
food scarcity, lynx have also been observed to prey on each other.

The young are born in May in a cave, or perhaps more often, under a
windfall. The gestation period is about 60 days.

In early June, 1955, I saw a lynx in the spruce woods near Savage
River. As I stood watching I heard crying sounds up in the woods.
The lynx disappeared in the direction of the crying. I followed and
saw the parent under a windfall as it was departing with a baby in its
mouth, the last of a litter it was moving. Snow and rain had fallen
and the mother was carrying her family, one by one, from under an
inadequate windfall to another about 250 yards away. The new home
was under a brushy spruce that provided a dry shelter in any kind of
weather. So well hidden and secure did the mother feel that she
barely opened her sleepy eyes even when approached within 20 feet.

How empty the woods and willow patches become with the decline of
the rabbits and the departure of the lynx. It is like an empty stage
after the actors have finished their play and departed. Scattered
through the quiet woods are their signs of life and activity, but the
action has stopped. On the tall willows, 6 feet from the ground, is the
gnawed white rabbit-line, where rabbits had sat on the snow and
gnawed the bark within reach. In places the ground is littered with
severed twigs, many of them partially gnawed. And everywhere one
encounters tufts of rabbit fur and hind legs, left on the green moss,
signifying rabbit tragedies and lynx banquets. But the rabbits will
return again to dance in the moonlight, and the lynx will be back in
his rich domain walking with stately and regal step.
 Wolverine
Gulo hylaeus

The fabulous wolverine is a powerful and picturesque member of the

weasel tribe weighing up to 35 pounds or more. Because of his
stocky build and long hair, he resembles a small bear. Frequently the
large hoary marmot is mistaken for him—there is considerable
similarity. But the broad yellowish-tan stripe on the sides of the 28
body is distinctive. A whitish collar, not always visible, extends
across the throat. The tail is short and bushy; the sharp, well-
developed claws are whitish. His range is circumpolar and extends
southward in the mountains to Colorado and California, but he is now
scarce south of Canada.

The wolverine in late years seems to have become more plentiful in

the park; nevertheless, it is always considered something special to
see one. They range from river bottom to ridge top, are found in the
woods or in open country miles beyond timber. Perhaps because of
the open view, he is frequently seen on the low passes, especially on
Sable Pass.

In winter the track of the far-wandering wolverine is frequently seen.

In his usual gait he bounces along with back arched. Each jump
usually leaves a set of three imprints; the one in front is made by a
hind foot; the middle imprint is made by a hind foot falling in the
track of a front foot; the rear imprint is made by the other front foot.
As in the tracks of a hopping rabbit, the hind feet tend to be brought
up ahead of the front feet.

The wolverine readily climbs trees. One winter, near a moose carcass,
tracks in the snow showed that a wolf had chased a wolverine up a
tree on two or three occasions. If the two had met in the open the
threatening posture of the wolverine would, no doubt, be sufficient to
discourage attack. When attacked by a dog, a wolverine has been
seen to lie on its back in a defensive attitude, a position that was
effective. Powerful ripping claws and jaws face the attack.
Not much quantitative information has been gathered on his food
habits. I have watched him pouncing on mice and suspect that mice
(voles) and lemmings are the most important items in his diet. In
summer he captures ground squirrels, sometimes by doing some
digging. Once I noted that he had dug out a wasp nest hidden in the
ground. The calves of caribou and moose, when very young, are no
doubt potential victims. But observations indicate that even a caribou
can ward off an attack on the young calves. Such items would, of
course, be unimportant in the wolverine’s total economy. His wide
wanderings in winter would seem to be helpful in finding carrion. In
rich wild country, considerable carrion probably comes his way. I have
found the wolverine attending a frozen moose carcass for a number
of days. When a carcass is not frozen, he carries away what he can
to cache for later use.

The gestation period is said to be about 9 months. The breeding

apparently takes place in summer. The fertilized eggs, after brief
development, lie unattached and dormant in the uterus for several
months. Some time in midwinter the eggs become attached to the
wall of the uterus and the more usual development takes place. (The
marten and short-tailed weasel have a similar breeding history.)
Females have been found in a nursing condition in early April. Along
Igloo Creek, Mr. and Mrs. Edwin C. Park watched a mother nurse two
young at least two-thirds grown.

29
 Wolverine.

According to Peter Krott, in his fascinating book about the wolverine

in Europe, this fierce animal makes a friendly pet. The author, in the
beginning, made a business of acquiring young wolverines for sale to
zoos. Because the animal was rare and intriguing, the demand was
great and the prices remunerative. But Mr. Krott and his wife became
fond of the wolverines and found it ever harder to dispose of them.
Soon they ceased selling them, and, instead, kept them as pets and
allowed them to roam freely over wild country. Studies were made of
their habits. Their wolverines might wander far and stay away for
several days, but they would return at intervals.

The wolverine is at home in the McKinley wilderness. Here we have

the rare opportunity of seeing him in his natural environment.

Marten
Martes americana actuosa
The marten is long and lithe, and its graceful activity is conspicuous.
The usual color is a rich brown, shading to blackish on the feet and
tail. The face is grayish with a short, dark line extending upward from
the inner corner of each eye. A large orange throat and breast patch
is very striking in most individuals. The fur is soft and long, the tail is
long and well-furred, and serves to register various emotions. The
marten is alert to sounds and this is indicated by its well-developed,
broad ears.

The body is 16 to 17 inches long, and the tail, including hair at tips, 8
or 9 inches. A large male may weigh up to 2½ pounds. The female is
somewhat smaller than the male.

The marten is found in the forested parts of the park along the
northern and eastern boundary. In winter, I have noted a few tracks
in the big spruce woods south of Wonder Lake. In Wyoming, I have
found martens in rock slides beyond timber, the rock crevasses 30
furnishing the desired protection.
Marten.
At one time the marten was thought to depend on the red squirrel for
his daily fare, but recent studies indicate that ordinarily relatively few
red squirrels are eaten. A food-habits study made at Castle Rocks
near the northwest corner of the park showed that the martens there
were living primarily on meadow voles and the red-backed mouse.
Blueberries were eaten in winter as well as in summer. In Wyoming, I
have found martens feeding extensively on blueberry, rhamnus, haw,
and mountain ashberries by choice at a time when voles and other
foods were plentiful. Like the fox and coyote, they have a strong
predilection for berries. In slide rock, they manage to capture an
occasional pika.

The marten breeds in July and August, but the young are not born
until 9 months later. The long gestation period for such a small
animal is due to the delayed attachment of the fertilized egg to the
uterus. Except for the period when the female is followed by young,
and during the breeding period, martens travel alone.

In Grand Teton Park, Wyoming, where I had much opportunity to

observe martens, I found that, although they seldom captured a red
squirrel, in their vagabond life over their home area, they did use red
squirrel homes for sleeping. The marten might spend a few days
resting in a squirrel’s spare nest, then move on to another squirrel
domicile for a few days. The squirrels suffered only the inconvenience
of an unwanted guest, and perhaps the temporary loss of a favorite
bed.

Mink
Mustela vison ingens

The mink is the amphibious member of the weasel family. He lives

along rivers and lakes and probably forages more in the water than
on land. Fish, frogs, insects, snails, crayfish, rabbits, muskrats, and
mice all appear on his bill of fare. In the country between the mouth
of the Yukon and the Kuskokwim River, the mink is said to subsist
largely on Alaska blackfish (Dallia pectoralis). So abundant 31
were the mink in the area that the Eskimo were called “mink
people.” The muddy waters in this watery region apparently
supported enough blackfish for both the mink and the natives. This
fish is said to have been the chief food of the natives. It is very
tenacious of life. Kegs of live fish, packed densely, were kept for food
in the dwellings. A steady slow rotary movement of the mass of fish
brought each fish to the surface at intervals for a gulp of air. When a
frozen blackfish is thawed, it is said to become as lively as ever.

Mink tracks have been noted along the Nenana River, but over most
of the eastern half of the park the mink is rare.

River Otter
Lutra canadensis yukonensis

The otter is rare in the park. It was reported present in Wonder Lake
some years ago and tracks in the snow were reported at Savage
River. It probably occurs in the Nenana River, along the eastern park
boundary.

The otter, a member of the weasel family, has become adapted to life
in the water. His body is about 3 feet long, and his long muscular tail
is over a foot long. His cousin, the sea otter, plentiful in the Aleutian
Islands, is much larger and more specialized for an aquatic life.

I have watched a family of otters in Grand Teton Park fishing for an

hour or longer. They kept diving steadily, and occasionally one would
come up with a small fish which he would proceed to eat, beginning
at the head. Larger fish are taken ashore. Trout, chubs, and suckers
were available but numerous droppings showed that the otter were
feeding chiefly on the chubs and suckers. The fish taken were no
doubt those most easily captured. A few crayfish were also eaten.
This particular family was living in a large beaver house also occupied
by beavers. They entered their chamber by land and apparently lived
upstairs above the beaver’s part of the house with its underwater
entrance.

In winter the otter frequently travels over the snow from one piece of
water to another. In these travels he slides on his belly down all
slopes and sometimes even on the level. In play, a family may
repeatedly climb a mudbank or a snowbank to course down a slide
leading into water.

Short-Tailed Weasel
Mustela erminea arctica

Two species of weasel occur in the park. The larger one with a black-
tipped tail is called the short-tailed weasel, and the smaller one with
an extremely short and all-white tail is the least weasel.

Both weasels are brown in summer and white in winter, a 32

protective coloration no doubt useful in escaping detection. In
some southern parts of their ranges these weasels remain brown all
year, and in intermediate areas part of the population turns white in
winter and part of it remains brown. It is apparent that climate has
an effect on coat color, the specific factor being the presence or
absence of snow on the ground.

It has been pointed out that the short-tailed weasel is much larger in
the north than in the southern part of the range. In Wyoming and
Colorado, where the tiny least weasel is absent, the short-tailed
weasel approaches the least weasel in size and probably fills that
weasel’s niche in the environment.

The food of the short-tailed weasel probably consists chiefly of

various species of meadow mice and lemmings. Observations indicate
that ground squirrels and rabbits may occasionally be captured.
Shrews no doubt are also on the menu.

In winter, weasel tracks form an odd pattern. Their jumps are

alternately long and short, and often they make an erratic trail.
Frequently the tracks show that the weasel disappears and travels
beneath the surface for a stretch before reappearing.

Even though weasels are not very palatable because of their well-
developed musk glands, they nevertheless are often preyed upon. It
is a case of coyote or fox capturing any small animal that moves and
examining the victim afterwards. Weasels are often left uneaten.

Short-tailed weasel.

33

Least Weasel
Mustela rixosa eskimo
The range of the least weasel is circumpolar. In North America it is
found over most of Alaska and Canada, and southward to Montana,
Kansas, North Carolina. It is widely dispersed but apparently nowhere
abundant. This tiny weasel is only 6 to 6½ inches long with a
maximum tail length of 1½ inches. It is the smallest living member of
the carnivores and weighs no more than a meadow mouse. The tail is
pure white, lacking the black tip present in other weasels.

I have a record of four specimens from the park. One captured in a

mousetrap was 5½ inches long, the tail measuring less than 1 inch. I
found a dead one at an eagle perch on a ridge top, and remains of
two others on gravel bars, apparently discarded after being captured.

A sourdough on the Koyokuk River with whom my brother and I

stayed one night, had a least weasel spending the winter with him. It
had the run of the cabin and was very tame.

Apparently the chief food of the least weasel is mice, some of them
about as large as himself.

Snowshoe Rabbit
Lepus americanus macfarlani

Like the ptarmigan and the northern weasel, the snowshoe rabbit, or
varying hare, each autumn changes from a dominantly brown
summer coat to a white winter ensemble. (In Washington where
snow is scarce in its habitat, the snowshoe rabbit remains brown the
year round.) His coat color blends at all seasons with his background,
so all he need do to be fairly sure of escaping visual detection is to
have confidence in his camouflage and sit motionless. The fur is so
long, thick, and warm that he can sit all day in fifty below zero
weather without freezing. His large hind legs are equipped with
snowshoe feet, an obvious advantage in snow country.
The most favorable rabbit habitat is the brushy country along the
east and north boundaries. Here a few may always be found. Out in
the park they are quite scarce except in those years when the
population is at or near a peak.

In winter, the rabbits feed on bark gnawed from various shrubs and
saplings. Willow, dwarf birch and alder, because of their high
palatability and abundance, are especially important winter foods. In
years of rabbit abundance, I have seen patches of willow and dwarf
birch trimmed to the snow line. At such times large willow 34
brush may show a white band 2 feet wide where the rabbits
have gnawed the bark within reach of the snow line. As the snow
deepens, some foods are buried but the change of level brings new
food supplies within reach. A variety of other shrubs are also eaten at
this season. Spruce bark is relished. Porcupines, and also red
squirrels, feeding in a spruce tree inadvertently add to the rabbit
menu many dropped spruce twigs. In summer, the rabbits turn to a
variety of fresh green foods.

The young of the snowshoe rabbit are furred and active when born
and apparently there is no real nest provided. (In the cottontail
branch of the family the young are born hairless, helpless and in a
warm nest.) The litters may vary from one to six. The gestation
period is about 36 days. The young are weaned (in captivity) when
about 4 weeks old. The females breed again soon after a litter is
born. It seems likely that a female may have as many as three or
four litters during a summer. A male is apparently with a female for
only a short time.

A number of animals are subject to cycles of extreme abundance and

scarcity. The pendulum swings from one extreme to the other. A
population, in spite of enemies of all kinds, increases until the
numbers become so large that they threaten the food supply or,
because of congestion, are drastically reduced by diseases. The
length of cycle in a species depends upon annual losses and the rate
of increase. Cycles are relatively short in voles and lemmings which
breed at an early age (a few weeks), breed often, and have large
litters. In these small rodents the cycle may cover a span of about 4
years. In larger species, the cycles are longer.

The snowshoe hare is one of the more obvious examples of a cyclic

species. From acute scarcity the population in about 10 years
pyramids until the country is full of rabbits. The woods are alive with
a variety of activity. Enterprise, lovemaking, and tragedy are at their
peak. Not only have the rabbits multiplied, but their enemies have
flourished, and the lynx, fox, wolverine and birds of prey have all
prospered, and certain enemies such as the lynx, become especially
abundant.

During the high rabbit population peak, between 1953 and 1955, a
few dead rabbits began to appear in the summer of 1954. In early
August a group of tourists on a short walk noted three dead rabbits
in the hotel area. But the rabbits were still numerous in the spring of
1955 and I anticipated the woods alive with young rabbits the
following months. Instead, they decreased. By July, along Igloo
Creek, they had become scarce. The so-called rabbit crash had taken
place.

Nature steps into all situations, and one control or another

automatically appears. Food shortage, disease, predation or
competition enter the picture. Adjusting is a continuous process.
Many people are talking and writing about the human population
explosion in our midst, fearing that space for ourselves and nature is
disappearing alarmingly. Perhaps we should consider the snowshoe
rabbit.

35
 Snowshoe rabbit, or varying hare.

Collared Pika
Ochotona collaris

The pika, cony, or rock rabbit, as he is variously called, makes his

home in rock slides. His way of life, and his physical attributes, are
such, that he would have difficulty surviving away from the labyrinth
of passages in his slide rock home.
The pika is in the same order (different family) as hares and rabbits.
Like rabbits, they have two pairs of upper incisors; back of the
grooved anterior incisors is a pair of very small incisors. The feet are
furred; the ears are not long but are rounded and prominent.
Something has happened to the external tail for there is none. The
tail vertebrae lie under the body skin. The plump body is about 6 or 7
inches long as the cony sits on a rock; the color is gray.

36

Collared pika (cony, rock rabbit)

The call is a single nasal “yank,” usually uttered while perched on a

rock where he can look around. He may be difficult to locate, but a
movement as he disappears in a crevice and reappears on the same