Originally posted by: AshishS

Hi,

I am working with an OPL model in CPLEX Optimization Studio 12.2, being solved with the CP Optimizer. I would like to customize search using the search phase functionality provided in OPL, to something like the following:

variables: x[1] through x[10], each with domain {0..5}.

variable selection: smallest to largest

value selection: select 0 if it is in the domain; select a random value otherwise

Looks like f.selectLargest(f.varIndex(x)), where f is cp.factory, will achieve the desired variable selection if I add it to an execute block before the model constraints. However, I don't see how to define a simple value selection heuristic such as the above. f.explicitValueEval(...) appears to be useful, but the documentation doesn't quite explain it.

Any help would be greatly appreciated!

Ashish
#CPOptimizer
#DecisionOptimization

Hello,

could the following help you?

int values[i in 0..5]=i; 

int valueEval[i in 0..5]=(i==0)?1000:0;     dvar 

int x[1..10] in 0..5;   execute 
{ var f = cp.factory;   var phase1 = f.searchPhase(x, f.selectSmallest(f.varIndex(x)), f.selectLargest(f.explicitValueEval(values, valueEval, 0))); cp.setSearchPhases(phase1); 
}

Alex
#CPOptimizer
#DecisionOptimization