[rules-users] Difficulty determining why I have score corruption [Planner]

[Geoffrey De Smet]

Op 09-01-13 01:00, megwoods schreef:
> Hi,
>
> I have been at this a couple of days now but I am getting score corruption
> from a rule that does the following:
>
> rule "to bigger gap"
>      when
>        $res: ResourceInstance()
>        $resList : ArrayList(size > 1)
>                from collect( TripAtTimeWithResource(resource  == $res) )
>      then
>       -- Sort the list using custom comparator.
>       -- Measure a series of deltas between elements in the list.
>       -- If the delta is larger than a threshold add a soft constraint.
1) Ideally, this logic would be in the "when" part.
Not that that explains score corruption, but it should be faster 
performance wise because it does better "incremental score calculation" 
(see manual about that concept).

I would also expect that there is no "from collect", but a "from 
accumulate" to detect the max or sum of the delta.

// quick pseudo code
when
   $res: ResourceInstance()
   (intValue > deltaTreshold )from accumulate(
       TripAtTimeWithResource(resource  == $res, $a : timeslot)
       and TripAtTimeWithResource(resource  == $res, timeslot > $a, $b : 
timeslot)
       and not TripAtTimeWithResource(resource  == $res, timeslot > $a, 
timeslot < $b),
       max(... delta of $a and $b ...)
   )
then


Addendum: If needed (because of performance reasons how from's currently 
work),
you might want split it up into 2 rules, similar as nurse rostering's 
drl does regularly (with the Start and End logicalInserts), although 
it's unlikely.
>       insertLogical(new IntConstraintOccurrence("to bigger gap",
> ConstraintType.NEGATIVE_SOFT,delta * 10, ttrL));
2) What's that cause ttrL? I would have expected $res as a cause?
>
> end
>
> The TripAtTimeWithResource is the planning entity.
>
> I have checked the cloning and been through and checked that all equals and
> hashcode methods and they are correct.
>
> If I take this rule out the implementation functions as correctly.
>
> Interestingly though I first prototyped this using a simple score and with
> the same 'when' and 'then' clause (inserting a different fact) and no score
> corruption was observed.
>
> If anyone has any suggestions I would really like to hear from you.
>
> Cheers Megan Woods
>
> Extract from exception:
> The workingMemory has 3 ConstraintOccurrence(s) in excess:
>      to bigger gap/NEGATIVE_SOFT:[TripAtTimeWithResource
> [trip=a.c.t.p.p.m.TripInstance at 43b, resource=All(17), startTime=TimeUnit
> [minute=285]]]=750
>      to bigger gap/NEGATIVE_SOFT:[TripAtTimeWithResource
> [trip=a.c.t.p.p.m.TripInstance at 46c, resource=All(17), startTime=TimeUnit
> [minute=75]]]=1700
>      to bigger gap/NEGATIVE_SOFT:[TripAtTimeWithResource
> [trip=a.c.t.p.p.m.TripInstance at 421, resource=All(17), startTime=TimeUnit
> [minute=75]]]=960
>    The workingMemory has 4 ConstraintOccurrence(s) lacking:
>      overlapping trips with same
> resource/NEGATIVE_HARD:[TripAtTimeWithResource
> [trip=a.c.t.p.p.m.TripInstance at 43b, resource=All(17), startTime=TimeUnit
> [minute=285]]]=100
>      to bigger gap/NEGATIVE_SOFT:[TripAtTimeWithResource
> [trip=a.c.t.p.p.m.TripInstance at 421, resource=All(17), startTime=TimeUnit
> [minute=75]]]=1710
>      to bigger gap/NEGATIVE_SOFT:[TripAtTimeWithResource
> [trip=a.c.t.p.p.m.TripInstance at 43b, resource=All(17), startTime=TimeUnit
> [minute=285]]]=1650
>      to bigger gap/NEGATIVE_SOFT:[TripAtTimeWithResource
> [trip=a.c.t.p.p.m.TripInstance at 46c, resource=All(17), startTime=TimeUnit
> [minute=75]]]=950
>
>
>
> --
> View this message in context: http://drools.46999.n3.nabble.com/Difficulty-determining-why-I-have-score-corruption-tp4021412.html
> Sent from the Drools: User forum mailing list archive at Nabble.com.
> _______________________________________________
> rules-users mailing list
> rules-users at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/rules-users
>