[gecode-users] different nr of solutions with and without GIST

[Tias Guns]

Hey,

I did indeed used a custom propagator. I refactored it and now the problem  
doesn't occur anymore... I guess the fault was totaly mine, sorry for the  
overhead...

g,T

On Wed, 07 Jan 2009 15:00:00 +0100, Christian Schulte <cschulte at kth.se>  
wrote:

> Hi Tias,
>
> what constraints are used (in particular anything you implemented  
> yourself)?
> Does it happen if you switch off recomputation (passing -c-d 1 on the
> command line or setting c_d in the search options to 1)?
>
> Could you run it against 2.2.0?
>
> Otherwise, we would need your code to have a look.
>
> Cheers
> Christian
>
> --
> Christian Schulte, www.ict.kth.se/~cschulte/
>
>
> -----Original Message-----
> From: users-bounces at gecode.org [mailto:users-bounces at gecode.org] On  
> Behalf
> Of Tias Guns
> Sent: Wednesday, January 07, 2009 2:45 PM
> To: users at gecode.org
> Subject: [gecode-users] different nr of solutions with and without GIST
>
> Hello,
>
> I am running my code against the latest SVN version (r8036) and have the
> following bizar problem:
> I am looking for all solutions that adhere to a set of constraints. The
> search happens over one array of boolean decision variables, and all the
> constraints are put on different subsets of this array.
>
> For a certain input datadaset, running the model returns 500 results. If  
> I
> run the same model using GIST (-mode gist), it returns 508 results !  
> After
> manual verification, there are 508 results that satisfy all the
> constraints. The 8 results that are missing form a complete branch in the
> GIST search tree.
>
> I can't create a small example to reproduce it, but it happens everytime
> on my model for that dataset... Do you have any idea how this is  
> possible ?
>
>
> Greetings,
> Tias
>
> _______________________________________________
> Gecode users mailing list
> users at gecode.org
> https://www.gecode.org/mailman/listinfo/gecode-users