[gecode-users] Optimizing the constraint

Jeffrey jeffreym9 at yahoo.com.hk
Sun Apr 5 04:47:28 CEST 2009


I'm optimizing my constraints with Gist at the moment.
I've found that there are lots of pruned tree on the left hand side.
These trees may span up to 4-5 levels.
Unfortunately, when I trigger gist to print out the dead node, I didn't find any values that is making any constraints failed. I had tried to add implied constraints which had already improve the speed for some cases. 


I'm currently using: INT_VAR_SIZE_MAX, INT_VAL_MIN   for branching an IntVarArray.
 
(1) What problems may it be? Is there any dynamic implied constraint in Gist?
(2) If I found that many dead node of a big pruned tree is found by the last assignment of values, How can I make such assignment comes earlier? e.g. Assign variables with more constraints first? Is it possible for Gist?
(3) Do you mind if I send up my little program for you to understand more about the Gist tree span?
(4) Although i use "INT_VAR_SIZE_MAX", it doesn't assign all values at once. Is that a bug or i'm misunderstanding its use? "INT_VAR_RND" is not suitable to me cause I want the search to be deterministic.
(5) Is there any other tricks I could use to sort out the problem? e.g. Deleting the constraint one by one to see which constraint is making it goes so long?
(6) Is there another symthoms i should try to look at the gist tree to get hint about the problem? e.g. what propagation setting should it be.

Big thanks again.


      Yahoo!香港提供網上安全攻略,教你如何防範黑客! 請前往 http://hk.promo.yahoo.com/security/ 了解更多!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.ps.uni-sb.de/pipermail/users/attachments/20090404/89fd8b81/attachment.htm>


More information about the gecode-users mailing list