| Re: diagonal dominant [message #82948] |
Wed, 30 January 2013 08:05  |
lecacheux.alain
Messages: 325
Registered: January 2008
|
Senior Member |
|
|
Le mercredi 30 janvier 2013 15:26:11 UTC+1, Gompie a écrit :
> Hi,
>
> I have a matrix( 256,256) and I want to convert it into diagonal dominant form.
>
> What is the best way to do it. Does IDL have function to do it?
>
> Gompie
I do not well understand your question. A matrix is diagonal dominant or it is not: you cannot change that by any converting method. Now, if you want to find a space in which your matrix is diagonal, your problem is solved by eigenvalue decomposition. In IDL, you can use EIGENQL and EIGENVEC functions or equivalently their Lapack counterparts: LA_EIGENQL, LA_EIGENVEC and LA_EIGENPROBLEM.
Hoping to have answered.
alain.
|
|
|
|
| Re: diagonal dominant [message #83047 is a reply to message #82948] |
Wed, 30 January 2013 08:23  |
Gompie
Messages: 76
Registered: August 2012
|
Member |
|
|
I basically want to solve AX=B.
When I give determ(A) idl gives 0.0000 with floating point exception. So I am not sure if idl does a good job at getting inverse.
So, I wanted to use gauss elemination on A matrix and apply the routine gs_iter.pro. The routine says that the input matrix should be diagonal dominant. So I thought if by pivoting I can make the array diagonally dominant.
While A is a (256,256) matrix. I have the flexibility of replacing rows in A with new ones if the A is singular with a particular set of rows.
Any ideas.
Gompie
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
