Re: [rules-users] OptaPlanner - best strategy for variable number of planning entities?
ge0ffrey wrote
Without a construction heuristic, local search waste a lot of time
doing
easy work.
No CH is like swimming 25 meter with LS.
CH with LS is like jumping 10 meters into the pool and then swimming the
remaining 15.
I've seen cases were it takes minutes for LS to make up for the work of
CH.
Oh, I completely agree, that's how it's supposed to theoretically work.
It's not that I'm not using a construction heuristic; I just can't seem to
make one actually function. I've tried FIRST_FIT, FIRST_FIT_DECREASING,
BEST_FIT, and BEST_FIT_DECREASING. In each case, it basically does
absolutely nothing. With the localsearch turned off, and a construction
heuristic section of:
<constructionHeuristic>
<constructionHeuristicType>BEST_FIT_DECREASING</constructionHeuristicType>
</constructionHeuristic>
I get:
OptaPlanner solver test...
2014-03-05 11:54:46,431 [main] INFO Solving started: time spend (11), score
(0hard/0soft), new best score (0hard/0soft), random seed (0).
2014-03-05 11:54:46,435 [main] TRACE Created cachedEntityList with size
(100) in entitySelector(Sorting(FromSolutionEntitySelector(Item))).
2014-03-05 11:54:46,436 [main] TRACE Sorted cachedEntityList with size
(100) in entitySelector(Sorting(FromSolutionEntitySelector(Item))).
2014-03-05 11:54:46,450 [main] INFO Phase (0) constructionHeuristic ended:
step total (0), time spend (32), best score (0hard/0soft).
2014-03-05 11:54:46,450 [main] INFO Solving ended: time spend (32), best
score (0hard/0soft), average calculate count per second (62).
Still trying to figure out why...
--
View this message in context:
http://drools.46999.n3.nabble.com/OptaPlanner-best-strategy-for-variable-...
Sent from the Drools: User forum mailing list archive at Nabble.com.