Originally posted by: UMQX_Olfat_ELMahi

Hello

i was worknig on my scheduling problem with interval:
dvar interval aplAct2j in TAp2Pkt ;

and i faced the following :

the following two commands as i understand shoul be the same but when i switch between them i found :

with the first one:
forall ( j in 0 .. nbPkt2*nbTsk2 - nbTsk2 -1 )
startOf(aplAct2[j]) <= startOf(aplAct2j+ nbTsk2);

error:
Model found inconsistent because of: forall(j in 0..2739) startOf(aplAct2[j]) <= startOf(aplAct2j+10).
with the second one:
forall (j in 0 .. nbPkt2*nbTsk2 - nbTsk2 -1 )
startBeforeStart(aplAct2[j], aplAct2j+ nbTsk2, 0);

results:
Problem found infeasible at the root node.

are the two constarint different ??

Thanks
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: UMQX_Olfat_ELMahi

correcting the commends apperant:
forall ( j in 0 .. nbPkt2*nbTsk2 - nbTsk2 -1 )
startOf(aplAct2[j]) <= startOf(aplAct2/j+ nbTsk2/);

forall (j in 0 .. nbPkt2*nbTsk2 - nbTsk2 -1 )
startBeforeStart(aplAct2[j], aplAct2/j+ nbTsk2/);
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: UMQX_Olfat_ELMahi

Sorry
forall ( j in 0 .. nbPkt2*nbTsk2 - nbTsk2 -1 )
startOf(aplAct2[j]) <= startOf(aplAct2[j+ nbTsk2]);

forall (j in 0 .. nbPkt2*nbTsk2 - nbTsk2 -1 )
startBeforeStart(aplAct2[j], aplAct2[j+ nbTsk2]);
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

Hello,
There are two points:

1) Generally speaking, the two constraints are not equivalent.

When you write startOf(a)<=startOf(b), if interval variable b is absent, this will enforce startOf(a)<=0 (as the value of startOf(b) is 0 if no other value is specified). When you write startBeforeStart(a,b), if b is absent it won't have any effect on the start value of a. Of course similar difference applies when a is absent.

That's why for scheduling, we advise using the form startBeforeStart that handles the optionality status (and that should be more efficient in the engine).

If both interval variables a and b are necessarily present (not declared optional) in the model, then the two constraints are semantically equivalent.

2) This being said, in your case, both models seem to be infeasible. The first one reports an inconsistency at extraction time whereas the second one reports inconsistency in initial propagation. But in both cases this means that the model is infeasible it is just that the infeasibility is not detected at the same moment.

Philippe
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: UMQX_Olfat_ELMahi

Thanks that was helpfull
but i wanted to state that although using the startBeforeStart make the engine much more faster but the first commant is finding the infeasibility more efficent maybe only with my program??

thanks
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

I suppose that in your case, the infeasibility is trivial and one of the constraints startOf(a)<=startOf(b) is infeasible because of the initial bounds of interval variables a and b.

Philippe
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: UMQX_Olfat_ELMahi

yes it was thanka ALOT
#DecisionOptimization
#OPLusingCPOptimizer