# LU decomposition code, don't know what it's doing. Can someone explain what this code is doing line-by-line?

63 views (last 30 days)
Nathan Nguyen on 26 Jan 2020 at 0:21
Commented: Harsha Priya Daggubati on 29 Jan 2020 at 4:38
Hi,
I'm trying to modify this code that performs an LU decompositon for a matrix A via column operations. The default code is written for row operations.
I don't understand what the code is doing though for the row operations. Does anyone mind explaining what is going on line-by-line? I know how to do LU decomposition by hand, but I don't have a really good idea what the script is doing.
I'll include the programme for you.
Thank You!
function [L,U] = mylu(A)
n = size(A,1);
for k = 1:n
if A(k,k)==0
warning('LU factorization fails');
L = []; U = []; return;
end
i = k+1:n;
A(i,k) = A(i,k)/A(k,k);
A(i,i) = A(i,i)-A(i,k)*A(k,i);
end
L = tril(A,-1)+eye(n); U = triu(A);

Harsha Priya Daggubati on 28 Jan 2020 at 12:34
Edited: Harsha Priya Daggubati on 28 Jan 2020 at 12:34
Hi, In LU Decomposition method we try to convert A matrix to echleon form by using gauss elimination method. The code starts from the first row, tries to find the factor by which we need to multiply the current row and subtract it from the rows below it, to make the elements in the Lower Triangular Matrix as zeros.
i = k+1:n; %To access rows below the current row
A(i,k) = A(i,k)/A(k,k); % To get the factor by which we need to multiply the current row and subtract it from row present below
A(i,i) = A(i,i)-A(i,k)*A(k,i); % To subtract from each row elementwise
The above code iterates till all the elements in Lower Triangular matrix becomes zero. Later we get the tril and triu of A, which gives L and U matrices.
I suggest using breakpoints, and work on this program. To know the updation of A for each iteration.
Hope this helps!

Nathan Nguyen on 29 Jan 2020 at 3:41
Is doing the breakpoint thing that you are suggesting easy to implement..? I have never used a breakpoint before.
I read that clicking on the line setsa breakpoint, but that doesn't really accomplish much.
Harsha Priya Daggubati on 29 Jan 2020 at 4:38
Using a breakpoint and step through each line of the code further to know the value of A at each step and comparing it with manual solving of LU Decomposition Method helps you to know the code better.
Take an example matrix A and try to solve it manually and then execute the program for the same matrix A. I hope this helps you.