Scribd
Upload
25 views11 pages

Power Method

The Power Method is an iterative technique used to find the largest eigenvalue and corresponding eigenvector of a matrix. The process involves making initial guesses, normalizing results, and iterating until the estimates converge, with specific termination criteria based on error percentage or calculated values. The document also discusses characteristics of eigenvalues and their relationships with matrix transformations.

Uploaded by

b21102
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
25 views11 pages

Power Method

The Power Method is an iterative technique used to find the largest eigenvalue and corresponding eigenvector of a matrix. The process involves making initial guesses, normalizing results, and iterating until the estimates converge, with specific termination criteria based on error percentage or calculated values. The document also discusses characteristics of eigenvalues and their relationships with matrix transformations.

Uploaded by

b21102
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 11

Power Method

Power method is an iterative approach employed to obtain the largest eigen


value for the system.
Eigen vector too is obtained in the process as a by product.

Consider the following system of equations

Note that the system of equations has been expressed in the form
[A] [X] =  [X]

We begin by considering [X] = [1 1 1]T as the first guess


ME504: Linear Equations: Vishal S Chauhan
Using the value of x’s as 1 in the left hand side or equations

Next we normalize the result of this substitution such that largest value
is 1. This step is summarized in the matrix form as given below

The eigen value estimate after first iteration is 1.778. For the
next iteration [X] = [1 0 1]T is used
ME504: Linear Equations: Vishal S Chauhan
Using [X] = [1 0 1]T for the next iteration, we have

Thus eigen value estimate after second iteration is 3.556. Also


x’s are [1 -1 1]T for next iteration.

ME504: Linear Equations: Vishal S Chauhan


The termination criterion in general can be the estimate of error

It is customary to use error percentage of variation of value obtained in


2 iterations. This avoids calculation through the equation, at the same
time if value stops changing significantly in successive iterations then
it is not of much use continuing the iterations.

However for our case, calculations are not too time taking and we may
use value of Y = [A] [X] -  [X] as the termination criterion. That is the
limiting condition can be on the maximum value of Y

ME504: Linear Equations: Vishal S Chauhan


Iterations

1 = 1.778 2 = 3.556 3 = -7.112

1 1 -0.75
X1 = 0 X2 = -1 X3 = 1
1 1 -0.75

4 = 6.223 5 = 6.096

-0.714 -0.708
X4 = 1 X5 = 1
-0.714 -0.708

ME504: Linear Equations: Vishal S Chauhan


AX = X

X1=[1,1,1]T
Algorithm for
Power method X’1 = AX1

l = largest of X’1

X2=X’1/ 

X’2= AX2

 = largest of X’2
ME504: Linear Equations: Vishal S Chauhan
Use Power method to obtain the dominant eigenvalue of the symmetric square
matrix given below

A= 1 1 1
1 2 2
1 2 3

ME504: Linear Equations: Vishal S Chauhan


1 = 6.0 2 = 5.1667 3 = 5.0645

0.5000 0.4516 0.4459


X1 = 0.8333 X2 = 0.8065 X3 = 0.8025
1 1 1

4 = 5.051 5 = 5.0492 6 = 5.0490

0.4451 0.4451 0.4450


X4 = 0.8020 X5 = 0.8019 X6 = 0.8019
1 1 1

ME504: Linear Equations: Vishal S Chauhan


Characteristics of eigen values which are useful in obtaining them

If λ1, λ2,…. λn are eigenvalue of matrix A

AT will have same eigenvalues λ1, λ2,…. λn .

A-1 will have eigenvalues λ1-1, λ2-1,…. λn-1 .

A-αI will have eigenvalues λ1 - α, λ2 - α,…. λn - α .

Ak will have eigenvalues λ1k, λ2k,…. λnk.

ME504: Linear Equations: Vishal S Chauhan


[A] [X] =  [X] With the rearrangement
described we can find the
[A]-1[A] [X] = [A]-1  [X] highest eigen value ’ for
the matrix B. Which will
be the reciprocal of the
(1/ ) [X] = [A]-1 [X] smallest eigen value of
matrix A.
[A]-1 [X] = -1 [X]

[B] [X] = ’ [X]

ME504: Linear Equations: Vishal S Chauhan

You might also like