Skip to main content

JACOBI AND GAUSS-SIEDEL SCHEMES


##The formulas and the iteration procedure
##below are taken from the Devies Book
##Eqn 5.89
##Constants and Initializations
dx=1E-3; ##Increment in x.
c1=1-2.5*dx; ## c1,c2,c3,c4 are constants in Eqn.5.89
c2=1+2.5*dx;
c3=-10*dx*dx;
c4=1/(2+c3);
## Set up the initial profile
x=0.0:0.01:1.0;
yold=0:100;
hold on
plot(x,yold);
xlabel('x');
ylabel('y');
##Impose the boundary conditions
Nsteps=1000;
yoldJac=yold; ## old y values for Jacobi Iteration Scheme
yoldGauSied=yold; ## old y values for Gauss-Siedel Scheme
ynewJac=yoldJac; ## new y values for Jacobi Iteration Scheme
ynewGauSied=yoldGauSied; ## new y values for Gauss-Siedel Scheme
for n=1:Nsteps
for i=2:100
ynewJac(i)=c4*(c1*yoldJac(i+1)+c2*yoldJac(i-1)+c3*x(i));
ynewGauSied(i)=c4*(c1*yoldGauSied(i+1)+c2*ynewGauSied(i-1)+c3*x(i));
endfor
yoldJac=ynewJac;
yoldGauSied=ynewGauSied;
endfor
plot(x,ynewJac,'g*',x,ynewGauSied,'m@');
title('y vs x');
xlabel('x');
ylabel('y');
legend('y(initial)','Jacobi','Gauss-Siedel');
axis([0,1.5]);
hold off
save -text JACOBIGAUSSSIEDEL.dat
print ('-dpsc','JACOBIGAUSSSIEDEL.ps')

Comments

Post a Comment

Popular posts from this blog

Simple Euler Method

##Usage:Call Octave from terminal ##and then call EulerMethodUmitAlkus.m ##from octave and finally ##press enter. That's all. ##Simple Euler Method ##Constants and initializations x=[]; ## initial empty vector for x y=[]; ## initial empty vector for y x(1)=1; ## initial value of x y(1)=1; ## initial value of y h=1E-3; ## increment in x dery=[]; ## 1st derivative of y wrt x n=1; ## inital loop index for while ## enter the while loop for the interval x=[1,2] while (x(n)<=2) x(n+1)=x(n)+h; dery(n+1)=x(n)*x(n)-2*y(n)/x(n); ##given y(n+1)=y(n)+h*dery(n); ##Euler method n++; endwhile ##exit from the 1st while loop ##Modified Euler Method ##Constant and initializations ymid=[]; ## empty vector function evaluated at x midpoint xmid=[]; ## empty vector func. of midpoints in x ymid(1)=1; ## inital value for ymid. derymid=[]; ## derivative of y at midpoints ##Enter the 2nd while loop n=1; while (x(n)<=2) xmid(n)=x(n)+h/2; derymid(n)=xmid(n)*xmid(n)-2*ymid(...
Post a Comment
Read more

FACTORIAL

## Function that calculates the factorial of a number ## Usage : f=factorial(n) function f=factorial(n) ## Initialize the output f=1; ## Check whether the input is correct if ( (n<0) || (rem(n,1)~=0) ) printf("n cannot be a negative number. Exiting...\n"); return endif for num=1:n f*=num; endfor endfunction
Post a Comment
Read more

REALISTIC AND NON-REALISTIC STRINGS

## A 'realistic', 'non-elastic' string, which responses to any ## bending and has stifness. This script takes in the previous ## and the present profiles and iterates to find ## the profile in the next time step. The ratio 'r' is not 1 ## like in the 'non-realistic' string since the speed of the wave ## always less than the speed of the string it should be less than 1 ## for best and most stable solution ##constants dx=1e-2 ## Spatial increment (m) L=2 ## Length of the string (m) M=L/dx ## Dimensionless partition E=1e-4 ## Dimensionless stiffnes function ynext=propagate_stiff(ynow,yprev,r) ## Quick and dirty way to fix boundary conditions -- for each step ## they are the same as the previous step. ynext=ynow; ynow(1)=ynow(2)=0; ##Entering the loop for i=3:length(ynow)-1 ## boundary condition ynow(length(ynow)-1)=ynow(length(ynow)-2)=0; ## Divide the ynext with many terms into three parts for easiness ynext(i)=(2−(2*r^2)−(6*E*(r^2)*...
Post a Comment
Read more