|
|
| Author |
Message |
cop04yh@shef.ac.uk
science forum beginner
Joined: 25 Jun 2006
Posts: 1
|
 Posted: Sun Jun 25, 2006 8:08 pm Post subject:
Can anyone help me?
|
|
|
Can anyone help me?
I am currently doing research on sparseness matrix issues. As I am
carrying on this topic, I had a problem that don't know what to do.
Basically the problem is as follows:
A and B are originally both sparse matrix, and I have to calcuate
P=(inv(B)*A)^(-2), when the order of P is small say 32 by 32, I still
can apply P into my algorithm and find a good solution, but when the
order is large, say 72 by 72 , and I applied it in Matlab@, I got a
warning like below
Warning: Matrix is close to singular or badly scaled.
Results may be inaccurate. RCOND = 2.781682e-021.
If the order of A and B are larger, say 1000 by 1000, then my system is
just not responding at all no matter how long I wait and matlab is
switch off automatically.
Can anyone help me to find a way slove the P matrix in large sparse
case in Matlab, which can avoid the matrix becomes singular?
Many thanks.
Khan. |
|
| Back to top |
|
 |
Gordon Sande
science forum beginner
Joined: 11 May 2005
Posts: 48
|
| Posted: Sun Jun 25, 2006 8:28 pm Post subject:
Re: Can anyone help me?
|
|
|
On 2006-06-25 17:08:22 -0300, cop04yh@shef.ac.uk said:
| Quote: | Can anyone help me?
I am currently doing research on sparseness matrix issues. As I am
carrying on this topic, I had a problem that don't know what to do.
Basically the problem is as follows:
A and B are originally both sparse matrix, and I have to calcuate
P=(inv(B)*A)^(-2), when the order of P is small say 32 by 32, I still
can apply P into my algorithm and find a good solution, but when the
order is large, say 72 by 72 , and I applied it in Matlab@, I got a
warning like below
Warning: Matrix is close to singular or badly scaled.
Results may be inaccurate. RCOND = 2.781682e-021.
If the order of A and B are larger, say 1000 by 1000, then my system is
just not responding at all no matter how long I wait and matlab is
switch off automatically.
Can anyone help me to find a way slove the P matrix in large sparse
case in Matlab, which can avoid the matrix becomes singular?
Many thanks.
Khan.
|
For sparse matrices 1000 x 1000 is not that large. And you are
trying out things on 32 x 32 and 72 x 72 which are not that large
for numerical computation. Better to find the problems with small
examples.
So use some tried and proven dense matrix techniques.
Namely find the singular values and vectors of A and B.
Look at the singular values to judge what the conditioning issue is.
Whether the matrix is singular is a property of the problem and not of
the solution technique. If it is singular, or poorly conditioned, then you
need to understand the problem. Noone will be able to offer an alternate
solution technique. Perhaps there might be other matrices that are not
so difficult but that is your research problem. |
|
| Back to top |
|
|
Google
|
|
| Back to top |
|
|
|
|
The time now is Sat Jan 10, 2009 4:20 am | All times are GMT
|
|
Mobile Phones | Remortgages | Xbox Mod Chip | Personal Car Finance | Free Advertising
|
|
Copyright © 2004-2005 DeniX Solutions SRL
|
|
Other DeniX Solutions sites:
Electronics forum |
Medicine forum |
Unix/Linux blog |
Unix/Linux documentation |
Unix/Linux forums
|
Powered by phpBB © 2001, 2005 phpBB Group
|
|
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
Forum index
»
Science and Technology
»
Math
»
num-analysis
