Originally posted by: SystemAdmin

[bistra said:]

Hi,
from the documentation of stateFunction it seems that the transition times should be enforced only between consecutive states.
However, the results I am getting suggest that the transitions are imposed between any pair of states one followin gthe other even if there are other states between them.

e.g. if transition between state 1 and 3 is 6000, between 1 and 2 is 0 and between 2 and 3 is 0, then a sequence of states 1,2,3 each of length 1 will still result in makespan 6002.

how can one impose sequence dependent changeover times in OPL so that the transitions are applied only between states that immediately follow each other?

thank you!
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[JR said:]

Hi

The transition distance for state function must satisfy triangular inequality (see the concept in the documentation in chapter state function and transition distance). In such a case, only the transition distance between immediate successor matters.

Having the behavior you search for without enforcing the triangular inequlaity requires modeling in a way that depends of the actual model

Cheers
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[bistra said:]

i was hoping that OPL might have a prebuild constraint that in fact works with changeovers that do not follow the triangle inequality (obviously there are explicit models that can handle it but such explicit modeling requires more variables).

It seems that it would be trivial to implement such a global constraint under the hood (of CP optimizer).


-- bistra
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[JR said:]

Hi Bistra

I would not say trivial mainly because of performances issues. Anyway I do agree with you not requiring triangular inequality for state resource is a good point.
#DecisionOptimization
#OPLusingCPOptimizer