Originally posted by: SystemAdmin

[golopang said:]

Hi, is there anyone know how to write an objective function, which is a summation with index value as a decision variable?? Something like below:

tuple Arc {
  int FromNode;
  int ToNode;
}
{Arc} arcs = ...;
float ArcCost[arcs] = ...;

dvar int+ successor[0..2*numJobs] in 1..2*numJobs+1;

minimize
    forall (a in 0..2*numJobs) sum ArcCost[<a, successor&#91;a&#93;>];

Thanks.
Anthony
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[alain.chabrier said:]

Hi,

try something like :

minimize
    sum (a in 0..2*numJobs) ArcCost[<a, successor&#91;a&#93;>];

Hope this helps,

Alain
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[golopang said:]

Dear Alain,

Thanks for your prompt reply, but still prompted to error message. I have sent you my model and data files, with almost all constraints disabled.

Thanks.
Anthony

#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[Didier Vidal said:]

With the datastructure you have, you can write the following which seems to be equivalent to your statement:


minimize
    forall (a in 0..2*numJobs, t in arcs : t.FromNode == a) sum ArcCost[t]*(t.ToNode == successor[a]);


If your problem is a job scheduling problem (just saying this in case your graph is a temporal network), maybe you can also consider using the detailed scheduling feature of OPL 6 (with time intervals).
#DecisionOptimization
#OPLusingCPOptimizer