[Scip] Fractional graph coloring - accessing solution of LP relaxation

[Timo Berthold]

Hi.

For nearly all methods, which have a SCIP_SOL as input, the comment for this 
parameter says:
/**< primal solution, or NULL for current LP/pseudo solution */
In particular, this works for SCIPgetSolVals() and SCIPprintSol(). You do not 
need to change the variables types.
More solution related methods you find in the SCIP online documentation of 
scip.h, under the point "Primal Solution Methods".
http://scip.zib.de/doc/html/scip_8h.html

Best,
Timo

Am Saturday 05 December 2009 14:35:12 schrieb Dimitri Sverdlov:
> Hello again,
>
> I am interested to see the solution of the LP relaxation of a graph
> coloring problem.
>
> For that I modified the Coloring example by replacing the type of
> variables that are created
> from SCIP_VARTYPE_CONTINUOUS to SCIP_VARTYPE_INTEGER. I also removed any
> preprocessing to keep things more simple.
>
> Strangely, the solution value that I get with this fractional coloring
> application is not the same as the dual bound from the LP relaxation in
> the root node from the original coloring application. So I guess there
> is still something wrong with my modified application.
>
> But I guess there must be a more simple way to get the solution of the
> LP relaxation. Is there a call similar to
> SCIPgetBestSol(scip)
> for that? I have already found the SCIPgetRootNode call, but I do not
> see how to get the LP solution from the SCIP_Node structure.
>
> Thank you, best regards,
> Dimitri
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> http://listserv.zib.de/mailman/listinfo/scip