[gecode-users] Watching Propagators

Max Ostrowski ChaosAngel at gmx.de
Mon Jun 20 16:11:44 CEST 2011


Thanks a lot for the paper.
I was not used to think in "non-trailing" systems, as i use a SAT-like solver.
But different to the approach from the paper i do not want to optimize search in Gecode.
So it would be enough to restrict myself to the propagation phase (Space->status()).

In my SAT system i added the CSP propagation to my unit-propagation.
So i do not do search (not here) with Gecode.
In my old implementation i checked which variables have changed after calling status(). All constraints that have these variables involved might have propagated something. The problem is, that i do not know the order of propagation.
Given that i decided (in my SAT solver) a reified constraint "A" is true and afterwards i call Gecode propagation. Another reified constraint "B" is made false by Gecode (just through propagation).
The reason for this might be "A", but also all other constraints that might got involved and whose domains have changed.
Given the order of propagation i could simply reduce the number of constraints that are the reason for constraint "B".

So, the problem doesn't consider search in Gecode, as i only start to search after i have a full propositional assignment.

Thanks a lot,
Max Ostrowski


-------- Original-Nachricht --------
> Datum: Mon, 20 Jun 2011 13:31:53 +0200
> Von: Guido Tack <tack at gecode.org>
> An: Max Ostrowski <ChaosAngel at gmx.de>
> CC: Mikael Zayenz Lagerkvist <zayenz at gmail.com>, users at gecode.org
> Betreff: Re: [gecode-users] Watching Propagators

> On 20 Jun 2011, at 12:04, Max Ostrowski wrote:
> 
> > I think there was once a work implementing learning in Gecode, i will
> contact the author to find out what he did. (But i suppose that he extended
> Gecode or did wrote own propagators, which i want to avoid).
> 
> Raphael Reischuk and I worked on SAT-style conflict clause learning in
> Gecode for Raphael's Bachelor thesis:
> http://www.infsec.cs.uni-saarland.de/~reischuk/publications/pdfs/Reischuk08-Reconciling.pdf
> 
> But this involved quite some modifications of the Gecode kernel.
> 
> Cheers,
> Guido
> 
> -- 
> Guido Tack, http://people.cs.kuleuven.be/~guido.tack/
> 
> 
> 
> 
> 
> _______________________________________________
> Gecode users mailing list
> users at gecode.org
> https://www.gecode.org/mailman/listinfo/gecode-users

-- 
Empfehlen Sie GMX DSL Ihren Freunden und Bekannten und wir
belohnen Sie mit bis zu 50,- Euro! https://freundschaftswerbung.gmx.de



More information about the users mailing list