Posts by aqms

aqms · Help Me !

Hi all,

I am losing my mind trying to figure out to perform modified incomplete cholesky factorization.

What I am trying to do is to use this method to create a preconditioner for the conjugate gradient method in order to solve Ax = b.

As A is a very large sparse matrix (of the size millions), the only possible way of solving this problem (practically) is to either use iterative method such as the conjugate gradient or the multigrid method.

Most of the reference books I read says that it is basically a modification of the incomplete cholesky factorization, by subtracting the fill values from the diagonal elements.

I did that, and I was not able to get the correct result.

Please, if anybody knows anything about this topic, do reply and share some thoughts.

My interest specifically would be on how to perform modified incomplete cholesky factorization with no fill - ins (MIC(0)).

Aznul

aqms · Help Me !

Hi all,

I would be really thankful if someone can explain to me on how to perform modified incomplete cholesky fatorization with no fill - in (MIC(0)).
I've got the idea that Instead of simply discarding the fill elements, what we need to do is to subtract the fill value from the corresponding diagonal element.
However, I have been following the algorithm provided in the book Matrix Computation (Golub Van loan), and am trying to modify the algorithm from standard incomplete cholesky factorization to the modified version, however, to no avail.

It would be nice if I can get the algorithm(pseudocode) or the layout of steps that I need to do to perform the factorization.

Regards,
Aznul

Math Is Fun Forum

#1 Help Me ! » Please Help Me! Modified incomplete cholesky factorization » 2009-03-24 15:21:36

#2 Help Me ! » Help:Modified incomplete cholesky factorization (no fill in) » 2009-03-19 23:44:21

Board footer