Re: [Help-glpk] [Fwd: About reading the relative mip gap from the glp_prob C struct]

[Heinrich Schuchardt]

Hello Nathan,

your mail had to be forwarded manually because you are not subscribed to
the GLPK help list, cf.
https://lists.gnu.org/mailman/listinfo/help-glpk

The gap of the last solution can be gathered using a callback function.

You can find an example at
http://sourceforge.net/p/glpk-java/code/HEAD/tree/trunk/examples/java/GmplSwing.java

The graph in
http://a.fsdn.com/con/app/proj/glpk-java/screenshots/278301.jpg
shows how the gap changes from one integer solution to the next.
It is created with the data from the callback function.

Best regards

Heinrich Schuchardt


On 26.08.2015 21:09, Andrew Makhorin wrote:
> -------- Forwarded Message --------
> From: Nathann Cohen <address@hidden>
> To: GLPK help <address@hidden>, David Coudert
> <address@hidden>
> Subject: About reading the relative mip gap from the glp_prob C struct
> Date: Wed, 26 Aug 2015 20:12:26 +0200
> 
> Hello everybody,
> 
> David Coudert and I are contributors to a software called Sagemath
> [1], and we currently try to improve its interface with LP solvers,
> among which is GLPK :-)
> 
> We would be very interested by a way to have in GLPK's C API, after a
> call to glp_intopt, the value of the relative gap reached by the best
> integer solution found so far.
> 
> Our problem is that this information, stored in the search tree
> object, is erased when glp_intopt returns [2]. Would it be possible to
> change the behaviour of glp_intopt and store it in the glp_prob C
> struct before the tree is destroyed?
> 
> Of course, we would be happy with any information that would be
> sufficient to deduce this gap, like the best lower bound (for a
> minimization problem).
> 
> Thank you very much,
> 
> Nathann Cohen
> 
> [1] http://www.sagemath.org/
> [2] http://lists.gnu.org/archive/html/help-glpk/2011-07/msg00068.html
> 
> 
> 
> 
> _______________________________________________
> Help-glpk mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/help-glpk
>