RE : Not enough work space, increase capacity of C

Thanks for the answer. You are right; it was the sparse matrix
multiplication that caused the problem. We had to pre-allocate storage
for the sparse matrix C.

One last question: We are developing a software at my biomedical
research lab and need a math package for linear algebra. We thought
about lapack but MTL seemed more convivial.

Is MTL still an active project or support and development is only based
on the research community. 

Thanks

Benoit Galarneau, jr. eng.

-----Message d'origine-----
De : Paul C. Leopardi [mailto:[email protected]] 
Envoyé : 27 août, 2003 18:58
À : General Matrix Template Library (MTL) list
Objet : Re: MTL: Not enough work space, increase capacity of C

Ben,
You need to allocate enough storage to C to contain the result of the 
multiplication. I commented on this in my "Evaluation of MTL for
GluCat":
http://www.osl.iu.edu/MailArchives/mtl-devel/msg00244.php

"...
3) Need to pre-allocate storage for sparse matrix operations
If you want to create sparse matrices you need to know how much storage
they
 will need before you operate on them. This is in contrast to eg.
Matlab,
 where a sparse array is extended as needed.

Related bugs:
3.1) Sparse matrix multiply assigns 0 when it runs out of space
Sparse multiply outputs a cryptic message and assigns 0 if not
 enough storage has been assigned to sparse array. Either the message
should
 be more informative or there should be an option to tell operations to
 automatically resize, such as an auto-resize policy class.
3.2) Documentation should mention this behaviour of multiply and does
not
You either need to expect this as "normal" behaviour, or discover this
 behaviour by using the debugger and examining the source code. ..."

GluCat now uses uBLAS. See http://glucat.sf.net
Best regards

On Thu, 28 Aug 2003 01:42, Benoit Galarneau wrote:
> When running a program based on MTL, we have to following message:
> Not enough work space, increase capacity of C.
> 
> This happens when doing the following: mult(A,B,C) where A, B and C
are 
400x400
> sparse matrix. C is initialize with zeros.
> 
> After the multiplication, C only contains zeros.
> 
> Any idea why ?
> 
> Thanks
> 
> Ben
> _______________________________________________
> This list is archived at http://www.osl.iu.edu/MailArchives/mtl-devel/
> 

_______________________________________________
This list is archived at http://www.osl.iu.edu/MailArchives/mtl-devel/

_______________________________________________
This list is archived at http://www.osl.iu.edu/MailArchives/mtl-devel/