Lecture13

35
Lecture1 3 Numerica l Analysis

description

Numerical Analysis. Lecture13. Chapter 3. Solution of Linear System of Equations and Matrix Inversion. Introduction Gaussian Elimination Gauss-Jordon Elimination Crout’s Reduction Jacobi’s Gauss- Seidal Iteration Relaxation Matrix Inversion. Gauss–Seidel Iteration Method. - PowerPoint PPT Presentation

Transcript of Lecture13

Page 1: Lecture13

Lecture13Lecture13

Numerical Analysis

Numerical Analysis

Page 2: Lecture13

Chapter 3Chapter 3

Page 3: Lecture13

Solution of Linear System of Equationsand Matrix Inversion

Solution of Linear System of Equationsand Matrix Inversion

Page 4: Lecture13

IntroductionIntroduction

Gaussian Elimination Gaussian Elimination

Gauss-Jordon EliminationGauss-Jordon Elimination

Crout’s Reduction Crout’s Reduction

Jacobi’s Gauss- Seidal Iteration Jacobi’s Gauss- Seidal Iteration

RelaxationRelaxation

Matrix InversionMatrix Inversion

Page 5: Lecture13

Gauss–Seidel Iteration Method

Gauss–Seidel Iteration Method

Page 6: Lecture13

It is another well-known It is another well-known iterative method for solving a iterative method for solving a system of linear equations of system of linear equations of the formthe form

11 1 12 2 1 1

21 1 22 2 2 2

1 1 2 2 1

n n

n n

n n nn n

a x a x a x b

a x a x a x b

a x a x a x b

Page 7: Lecture13

In Jacobi’s method, In Jacobi’s method, the (the (rr + 1) + 1)thth approximation to the approximation to the above system is given above system is given by Equationsby Equations

Page 8: Lecture13

( 1) ( ) ( )11 121 2

11 11 11

( 1) ( ) ( )22 212 1

22 22 22

( 1)( 1) ( ) ( )11 1

r r rnn

r r rnn

n nr r rn nn n

nn nn nn

ab ax x x

a a a

ab ax x x

a a a

ab ax x x

a a a

Page 9: Lecture13

Here we can observe Here we can observe that no element of that no element of replaces replaces entirely entirely for the next cycle of for the next cycle of computation.computation.

( 1)rix

( )rix

Page 10: Lecture13

In Gauss-Seidel method, In Gauss-Seidel method, the corresponding the corresponding elements of elements of replaces those of replaces those of as as soon as they become soon as they become available. available.

( 1)rix

( )rix

Page 11: Lecture13

Hence, it is called the method Hence, it is called the method of successive displacements.of successive displacements. For illustration considerFor illustration consider

11 1 12 2 1 1

21 1 22 2 2 2

1 1 2 2

n n

n n

n n nn n n

a x a x a x b

a x a x a x b

a x a x a x b

Page 12: Lecture13

In Gauss-Seidel In Gauss-Seidel iteration, the (r + 1)th iteration, the (r + 1)th approximation or approximation or iteration is computed iteration is computed from:from:

Page 13: Lecture13

( 1) ( ) ( )11 121 2

11 11 11

( 1) ( 1) ( )22 212 1

22 22 22

( 1)( 1) ( 1) ( 1)11 1

r r rnn

r r rnn

n nr r rn nn n

nn nn nn

ab ax x x

a a a

ab ax x x

a a a

ab ax x x

a a a

Page 14: Lecture13

1( 1) ( 1) ( )

1 1

i nij ijr r ri

i j jj j iii ii ii

a abx x x

a a a

Thus, the general procedure can be written in the following compact form

1,2,...,i n 1,2,...r for all for all and and

Page 15: Lecture13

To describe system in the To describe system in the first equation, we substitute first equation, we substitute the r-th approximation into the r-th approximation into the right-hand side and the right-hand side and denote the result by denote the result by In the In the second equation, we second equation, we substitute substitute and and denote the result by denote the result by

( 1)1 .rx

( 1) ( ) ( )1 3( , ,..., )r r r

nx x x

( 1)2rx

Page 16: Lecture13

In the third equation, we In the third equation, we substitute substitute and denote the result by and denote the result by

and so on. This and so on. This process is continued till we process is continued till we arrive at the desired result. arrive at the desired result. For illustration, we consider For illustration, we consider the following examplethe following example :

( 1) ( 1) ( ) ( )1 2 4( , , ,..., )r r r r

nx x x x

( 1)3 ,rx

Page 17: Lecture13

Relaxation Method

Relaxation Method

Page 18: Lecture13

This is also an iterative method This is also an iterative method and is due to Southwell. and is due to Southwell.

To explain the details, consider To explain the details, consider again the system of equationsagain the system of equations

11 1 12 2 1 1

21 1 22 2 2 2

1 1 2 2

n n

n n

n n nn n n

a x a x a x b

a x a x a x b

a x a x a x b

Page 19: Lecture13

LetLet

be the solution vector be the solution vector obtained iteratively after p-th obtained iteratively after p-th iteration. If denotes the iteration. If denotes the residual of the i-th equation residual of the i-th equation of system given above , that of system given above , that is ofis of

( ) ( ) ( ) ( )1 2( , ,..., )p p p p T

nX x x x

1 1 2 2i i in n ia x a x a x b

( )piR

Page 20: Lecture13

defined bydefined bydefined bydefined by( ) ( ) ( ) ( )

1 1 2 2p p p pi i i i in nR b a x a x a x

we can improve the solution we can improve the solution vector successively by vector successively by reducing the largest residual reducing the largest residual to zero at that iteration. This to zero at that iteration. This is the basic idea of relaxation is the basic idea of relaxation method. method.

Page 21: Lecture13

To achieve the fast convergence of the To achieve the fast convergence of the procedure, we take all terms to one side procedure, we take all terms to one side and then reorder the equations so that the and then reorder the equations so that the largest negative coefficients in the largest negative coefficients in the equations appear on the diagonalequations appear on the diagonal..

Page 22: Lecture13

Now, if at any iteration, is Now, if at any iteration, is the largest residual in the largest residual in magnitude, then we give an magnitude, then we give an increment to being the increment to being the coefficient ofcoefficient of xxii

iR

ii

ii

Rdx

a

;ix iia

Page 23: Lecture13

In other words, we change In other words, we change to to

to relax to relax

that is to reduce to zero. that is to reduce to zero.

.ix( )i ix dx

iRiR

Page 24: Lecture13

Example Example Solve the system of equations Solve the system of equations

by the relaxation method, by the relaxation method, starting with the vector (0, 0, 0).starting with the vector (0, 0, 0).

1 2 3

1 2 3

1 2 3

6 3 11

2 8 15

7 10

x x x

x x x

x x x

Page 25: Lecture13

SolutionSolutionAt first, we transfer all the At first, we transfer all the terms to the right-hand side terms to the right-hand side and reorder the equations, so and reorder the equations, so that the largest coefficients in that the largest coefficients in the equations appear on the the equations appear on the diagonal. diagonal.

Page 26: Lecture13

Thus, we getThus, we get

after interchanging the 2after interchanging the 2ndnd and 3and 3rdrd equations. equations.

1 2 3

1 2 3

1 2 3

0 11 6 3

0 10 7

0 15 2 8

x x x

x x x

x x x

Page 27: Lecture13

Starting with the initial Starting with the initial solution vector (0, 0, 0), that is solution vector (0, 0, 0), that is taking taking

we find the residualswe find the residuals

of which the largest residual in of which the largest residual in magnitude is Rmagnitude is R33, i.e. the 3, i.e. the 3rdrd equation equation

has morehas more error and needs immediate error and needs immediate attention for improvement. attention for improvement.

1 2 3 0,x x x

1 2 311, 10, 15R R R

Page 28: Lecture13

Thus, we introduce a change, Thus, we introduce a change, dxdx33in in xx33 which is obtained which is obtained

from the formulafrom the formula

33

33

151.875

8

Rdx

a

Page 29: Lecture13

Similarly, we find the new Similarly, we find the new residuals of large residuals of large magnitude and relax it to magnitude and relax it to zero, and so on. zero, and so on.

We shall continue this We shall continue this process, until all the process, until all the residuals are zero or very residuals are zero or very small.small.

Page 30: Lecture13

Iteration Residuals Maximum Difference VariablesIteration Residuals Maximum Difference Variables

numbernumber RR11 RR22 RR33 xx11 xx22 xx33

00 1111 1010 -15-15 -15-15 1.8751.875 00 00 00

11 9.1259.125 8.1258.125 00 9.1259.125 1.52881.5288 00 00 1.8751.875

22 0.04780.0478 6.59626.5962 -3.0576-3.0576 6.59626.5962 -0.9423-0.9423 1.52881.5288 00 1.8751.875

iR idx

Page 31: Lecture13

Iteration Residuals Maximum Difference VariablesIteration Residuals Maximum Difference Variables

numbernumber RR11 RR22 RR33 xx11 xx22 xx33

00 1111 1010 -15-15 -15-15 15/815/8=1.875=1.875

00 00 00

11 9.1259.125 8.1258.125 00 9.1259.125 -9.125/(-6)-9.125/(-6)=1.5288=1.5288

00 00 1.8751.875

22 0.04780.0478 6.59626.5962 -3.0576-3.0576 6.59626.5962 -6.5962/7-6.5962/7=-0.9423=-0.9423

1.52881.5288 00 1.8751.875

33 -2.8747-2.8747 0.00010.0001 -2.1153-2.1153 -2.8747-2.8747 2.8747/(-6)2.8747/(-6)=-0.4791=-0.4791

1.04971.0497 -0.9423-0.9423 1.8751.875

44 -0.0031-0.0031 0.47920.4792 -1.1571-1.1571 -1.1571-1.1571 1.1571/81.1571/8=0.1446=0.1446

1.04971.0497 -0.9423-0.9423 1.8751.875

iR idx

Page 32: Lecture13

Iteration Residuals Maximum Difference Variables

number R1 R2 R3 x1 x2 x3

55 -0.1447-0.1447 0.33460.3346 0.00030.0003 0.33460.3346 -.3346/7-.3346/7=-0.0478=-0.0478

1.04971.0497 -0.9423-0.9423 2.01962.0196

66 0.28810.2881 0.00000.0000 0.04750.0475 0.28810.2881 -.2881/(-6)-.2881/(-6)=0.0480=0.0480

1.04971.0497 -0.9901-0.9901 2.01962.0196

77 -0.0001-0.0001 0.0480.048 0.14350.1435 0.14350.1435 =-0.0179=-0.0179 1.00171.0017 -0.9901-0.9901 2.01962.0196

88 0.01780.0178 0.06590.0659 0.00030.0003 -- -- 1.00171.0017 -0.9901-0.9901 2.00172.0017

iRidx

Page 33: Lecture13

At this stage, we observe At this stage, we observe that all the residuals that all the residuals RR11, R, R22

and and RR33 are small enough are small enough

and therefore we may take and therefore we may take the corresponding values the corresponding values of of xxii at this iteration as the at this iteration as the

solution. solution.

Page 34: Lecture13

Hence, the numerical solution Hence, the numerical solution is given by is given by

The exact solution isThe exact solution is

1 2 31.0017, 0.9901, 2.0017,x x x

1 2 31.0, 1.0, 2.0x x x

Page 35: Lecture13

Lecture13Lecture13

Numerical Analysis

Numerical Analysis