On 06-03-14 03:23, Nadim wrote: > I solved the construction heuristic problem. Someone it didn't dawn on me > right away that primitive types don't work properly as planning variables, > because they're always initialized, thus the construction heuristic quits > right away, thinking there's nothing to do. Good point. OptaPlanner shouldn't surprise you like that, I 've created an issue: https://issues.jboss.org/browse/PLANNER-227 > > However, construction heuristic or no, it looks like it's getting stuck on > local optima most of the time even in brain-dead simple cases (e.g. adding a > constraint to force exactly one window). Not sure if I may need to add some > custom move logic to account for the fact that there are many Boolean > planning variables (one for each potential session to attend), but that most > of the potential toggles don't make much sense (generally you'd want to > either tweak the sessions on the edges of existing windows, or try killing a > window and growing a different one elsewhere. to kill an entire window, it probably makes sense to have PillarChangeMove's (which aren't implemented yet, only PillarSwapMoves are currently available, see docs). > > > > -- > View this message in context: > http://drools.46999.n3.nabble.com/OptaPlanner-best-strategy-for-variable-number-of-planning-entities-tp4028470p4028555.html > Sent from the Drools: User forum mailing list archive at Nabble.com. > _______________________________________________ > rules-users mailing list > [email protected] > https://lists.jboss.org/mailman/listinfo/rules-users >
_______________________________________________ rules-users mailing list [email protected] https://lists.jboss.org/mailman/listinfo/rules-users
