Originally posted by: SystemAdmin
By definition phase II of feasopt works with the
optimal solution of phase I. So CPLEX will never start phase II if phase I did not finish with a solution that is optimal or optimal within tolerances.
If your stopping criterion for phase I is relative or absolute gap then you can achieve that using setting parameters and phase II should kick in as expected. If you need a different stopping criterion in phase I then the only way to get phase II I can see right now is to explicitly create the phase II model (with the solution that CPLEX found in phase I) and solve that.
#CPLEXOptimizers#DecisionOptimization