Scribd
Upload
91 views

Lect 5 PDF

This document provides an introduction to numerical methods for solving systems of linear equations. It discusses direct methods like Gaussian elimination, LU factorization, and Cholesky factorization. It also covers iterative methods like Jacobi and Gauss-Seidel iteration. Special matrices like tridiagonal, banded, and sparse matrices that arise in applications are also discussed. Efficient storage schemes and solvers for these special matrices are described.

Uploaded by

Luis Inga Marquez
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
91 views

Lect 5 PDF

This document provides an introduction to numerical methods for solving systems of linear equations. It discusses direct methods like Gaussian elimination, LU factorization, and Cholesky factorization. It also covers iterative methods like Jacobi and Gauss-Seidel iteration. Special matrices like tridiagonal, banded, and sparse matrices that arise in applications are also discussed. Efficient storage schemes and solvers for these special matrices are described.

Uploaded by

Luis Inga Marquez
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

Introduction to Numerical Analysis for Engineers

Systems of Linear Equations Mathews

Cramers Rule
Gaussian Elimination 3.3-3.5
Numerical implementation
3.3-3.4
Numerical stability
Partial Pivoting
Equilibration
Full Pivoting
Multiple right hand sides
Computation count
LU factorization 3.5
Error Analysis for Linear Systems 3.4
Condition Number
Special Matrices
Iterative Methods 3.6
Jacobis method
Gauss-Seidel iteration
Convergence
13.002 Numerical Methods for Engineers Lecture 5
Linear Systems of Equations
Tri-diagonal Systems

Forced Vibration of a String


Finite Difference
f(x,t)

xi y(x,t)
Discrete Difference Equations
Harmonic excitation
Matrix Form
f(x,t) = f(x) cos(Zt)

Differential Equation

Boundary Conditions

Tridiagonal Matrix

Symmetric, positive definite: No pivoting needed

13.002 Numerical Methods for Engineers Lecture 5


Linear Systems of Equations
Tri-diagonal Systems
General Tri-diagonal Systems

LU Factorization

13.002 Numerical Methods for Engineers Lecture 5


Linear Systems of Equations
Tri-diagonal Systems
LU Factorization
Reduction

Forward Substitution

Back Substitution

LU Factorization: 2*(n-1) operations


Forward substitution: n-1 operations
Back substitution: n-1 operations
Total: 4(n-1) ~ O(n) operations

13.002 Numerical Methods for Engineers Lecture 5


Linear Systems of Equations
Special Matrices

General, Banded Coefficient Matrix


p

0
q

Banded Symmetric Matrix

b is half-bandwidth
p super-diagonals
q sub-diagonals
w = p+q+1 bandwidth

13.002 Numerical Methods for Engineers Lecture 5


Linear Systems of Equations
Special Matrices
Banded Coefficient Matrix
Gaussian Elimination
No Pivoting
p

0
q
0
= =
0

13.002 Numerical Methods for Engineers Lecture 5


Linear Systems of Equations
Special Matrices
Banded Coefficient Matrix
Gaussian Elimination
With Pivoting

q
q
0
=

13.002 Numerical Methods for Engineers Lecture 5


Linear Systems of Equations
Special Matrices

Banded Coefficient Matrix


Compact Storage

Diagonal
p q p q

q
0
0

i i

0 0

j n2 j-i n(p+2q+1)
13.002 Numerical Methods for Engineers Lecture 5
Linear Systems of Equations
Special Matrices

Sparse and Banded Coefficient Matrix


Skyline Systems

..
Skyline
0

Storage ..
..

Pointers 1 4 9 11 16 20 ..
0

Skyline storage applicable when no pivoting is needed, e.g. for banded,


symmetric, and positive definite matrices: FEM and FD methods. Skyline
solvers are usually based on Choleski factorization
13.002 Numerical Methods for Engineers Lecture 5
Linear Systems of Equations
Special Matrices

Symmetric, Positive Definite Coefficient Matrix


No pivoting needed

Choleski Factorization

where

13.002 Numerical Methods for Engineers Lecture 5

You might also like