RE: [Help-glpk] Outputting dual cost from GLPSOL

[D'Agostino, Larry - TX]

Xypron,

How do you propose the patch is incorporated?  Do you think maybe as a switch 
on the glpsol binary?

-----Original Message-----
From: help-glpk-bounces+larry.d'address@hidden 
[mailto:help-glpk-bounces+larry.d'address@hidden On Behalf Of Andrew Makhorin
Sent: Friday, June 12, 2009 8:33 AM
To: Xypron
Cc: address@hidden
Subject: Re: [Help-glpk] Outputting dual cost from GLPSOL

Hi Xypron,

> the question of outputting dual cost from GLPSOL has been raised
> repeatedly:

> http://lists.gnu.org/archive/html/help-glpk/2007-02/msg00024.html
> http://lists.gnu.org/archive/html/help-glpk/2008-01/msg00060.html
> http://lists.gnu.org/archive/html/help-glpk/2008-12/msg00076.html
> http://lists.gnu.org/archive/html/help-glpk/2009-02/msg00065.html

> A patch is available:
> http://glpk.dyndns.org/viewvc/svn/glpk/glpk/branches/glpk-4.38-dot/
> svn://glpk.dyndns.org/svn/glpk/glpk/branches/glpk-4.38-dot/

Thank you for your efforts.

> Do you see a chance to integrate this feature?

It is still unclear to me in which context the dual values and
statuses could be used in the model, because there is no feature to
change and reoptimize the model. Only for reporting purposes?

(I noticed that in the patch to glpapi16.c you obtain the row/column
statuses with glp_get_xxx_stat even for GLP_IPT and GLP_MIP. For GLP_IPT
all statuses should be assigned to GLP_BS, because formally in the
interior-point solution all primal and dual variables are "basic". For
GLP_MIP it would be possible to fix all integer variables at their
optimal values and then find corresponding basic solution; this would
provide "correct" statuses and dual values.)

Best regards,

Andrew Makhorin



_______________________________________________
Help-glpk mailing list
address@hidden
http://lists.gnu.org/mailman/listinfo/help-glpk