Hi Malcolm, yes you can add constraints dynamically. Actually you can add anytime! The only thing you need is to have access to the space you want to add constraints to.
One way to get started would be to branch only over the subset of the variables. Then when search stops (it will stop if the branching is done), you can add more constraints and additional branchings. Cheers Christian -- Christian Schulte, www.ict.kth.se/~cschulte/ -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Malcolm Ryan Sent: Monday, August 25, 2008 8:25 AM To: gecode list Subject: [gecode-users] Incremental constraints I'm having efficiency problems with my constraint system. It is slow compared to a standard search algorithm with much more limited deductive ability. I think that too much time is being spent propagating constraints in ways that are ultimately not improving the search. Rather than abandon CSPs completely, I'd like to experiment with modulating the amount of propagation to see if there is some point at which it becomes advantageous. Is there any generic way to do this? If I remove constraints completely, it will effect the correctness of my solution, which I don't want to do. I just want to delay constraint checking/propagation. One way that occurs to me would be to add constraints incrementally, during the search. Could Gecode handle this? If so, which phase of the search process should it occur in? I'm guessing that it should be part of commmit(). Would that be appropriate? Thanks, Malcolm _______________________________________________ Gecode users mailing list [EMAIL PROTECTED] https://www.gecode.org/mailman/listinfo/gecode-users _______________________________________________ Gecode users mailing list [EMAIL PROTECTED] https://www.gecode.org/mailman/listinfo/gecode-users
