Originally posted by: Rusman

I conduct the experiment using ILOG CPLEX to execute my problem formulation,
but I found error message CPLEX Error 5002 : Q is not positive semi-definite
any one can help how to change my model into positive definite,
for consideration attached my proposed model and data
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

I did not look very closely, but I think your MIQCP can be transformed into an equivalent MILP.

For example, variables "Flow_aP" and "Ass_zP" always come in pairs as a product. What you wanted to model is probably "there can only be flow if the assignment variable is zero".

In such cases, instead of using the product of Flow_aP * Ass_zP everywhere, you should just use Flow_aP in the constraints and the objective, and then enforce

Ass_zP = 0 -> Flow_aP = 0

as an additional constraint. You can either use indicator constraints to do this, or a big-M formulation.
Tobias
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Using Tobias's suggestion of using indicator constraints, I was able to convert your model from the MIQCP to a MILP. Please find the mod/dat file attached.

This model however is infeasible in its current form. Going through the constraints that may cause this, it seems that the maxCap variable is set too low- the sum of the maxCap for the 1st and 2nd Relay station is lower than the total demand making the model infeasible. Once you increase the maxCap for the 1st or 2nd Relay station enough to meet the demand, the model becomes feasible.

I hope this helps.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: Rusman

AbhishekRaman and Tobias, Thank you very much for excellent explanation and help
Best regards
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: WaiLin

Hi, I am facing the same error too.
Can anyone please help me? I could not continue next step for my optimization problem.
THANK YOU !
#CPLEXOptimizers
#DecisionOptimization