For some reason every time I try to test my forward elimination function there is an error in the sum part, I cannot figure out why or what else I need to ad

function [ y ] = forward_elim( L, b )
%performs gaussian foward elimination solving Ly=b using specific
%algorithm given
%   Detailed explanation goes here

n = length(b);
y= zeros (n,1);

for i = 1:n

        i = i+1;

      %check to see that L is a valid matrix

        if size (L) == size(L')                    % testing to see that L is a square matrix

            if isnan(L)                           % if there is a not real number terminate
            disp('error: invalid inputs in L')
            break

            elseif diag(L) == 0                   % if a diagnol element of the matrix is 0 terminate
            disp ('error: A diagnol element of L = 0')
            break
            end
        else
            disp('error: L is not a suqare matrix')
                break
        end

        sum = 0;
        for j = 1:i-1
            sum = sum + L(i,j)*y(j);
        end

        y(i) = (b(i)- sum)/L(i,i) ;  
  end

end

[Walter Roberson]

What is the error message?

You should firmly avoid using sum as the name of a variable, as it is the name of a key MATLAB function -- a function you might want to use in that code.

By the way:

L(i,1:i-1) * y(1:i-1)

will give you the sum without any for loop.

Answer 1

[James Tursa]

Looks like you are incrementing the i index variable twice ... once as a for-loop index and once explicitly:

for i = 1:n
        i = i+1;  % <-- You should probably delete this line

Also, you might consider moving your L checks outside of your for-loop. There is no need to do these checks over and over again for each iteration. Just once prior to the for-loop would be sufficient. And some of these tests should change:

            if isnan(L)

should be

            if any(isnan(L(:)))

and

            elseif diag(L) == 0

should be

            elseif any(diag(L) == 0)

Finally, I would advise against using just a message with a break when you encounter an error, as this will still return a garbage value to the caller. Instead, actually force an error. E.g., change this:

            disp('error: invalid inputs in L')
            break

to this

            error('invalid inputs in L')

And same for all of your other error catches.