[Ipopt] sparse matrix additions

[Kelly, Jeff (ON0F)]

Rodrigo;

 

I use TOMS601 and SPARSKIT (Y. Saad) for sparse matrix and vector manipulations which usually require you to present the coordinate, ordered-pair, non-zero method or “triplet” storage format in a row or column sorted arrangement.

 

To sort a row or column in a triplet storage sparse matrix format the most efficient method to do this is the “distribution counting sort” found in TOMS636 as SRTDAT() and is embed in TRSMGU() of TOMS601 – do not use a quicksort to do this.

 

Jeff

 

From: ipopt-bounces at list.coin-or.org [mailto:ipopt-bounces at list.coin-or.org] On Behalf Of Rodrigo Lopez-Negrete
Sent: Friday, October 22, 2010 11:21 AM
To: Ipopt - Mailing list
Subject: [Ipopt] sparse matrix additions

 

Dear all,

 

I'm using the C++ interface of Ipopt, and I'm wondering if any of you out there have suggestions for codes that will allow me to do fast sparse matrix additions using the triplet format used in Ipopt. I've seen some codes that use compressed row or column formats, but I'm not using those conventions. For now, I'm trying to not change my code that much. I need this to add the Hessian of the objective function with the Hessian of the constraints, among others. 

 

Any suggestions will be appreciated.

 

Thanks,

 Rodrigo

-- 
 <http://www.cs.cmu.edu/~aml/logos/wordmark.jpg> 
Rodrigo López-Negrete de la Fuente
Doctoral Candidate 
Email: rln at cmu.edu <mailto:rln at cmu.edu> 

WWW: http://rodrigo.lopez-negrete.org <http://rodrigo.lopez-negrete.org/> 

Department of Chemical Engineering
Carnegie Mellon University 
5000 Forbes Ave. 
Pittsburgh, PA 15213-3890
USA

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.coin-or.org/pipermail/ipopt/attachments/20101022/71645e01/attachment.html