Decision Optimization

Decision Optimization

Delivers prescriptive analytics capabilities and decision intelligence to improve decision-making.


#Analytics
#DecisionOptimization
#DecisionOptimization
 View Only
  • 1.  Solution Pool

    Posted 02/20/12 11:53 AM

    Originally posted by: kc78


    Hi I have a MILP problem and i am trying to interate all the possible solutions to the problem using the solution pool feature and at each iteration i add the solutions obtained as integer cut constraints to the problem and resolve it again. I have set the intensity to 4 and the relative and absolute gaps to 0. It works well for the first 2 iterations and then i keep getting repeats of the optimal obj value. For example:

    iteration 1 => obj value: 26, # solutions = 6
    iteration 2 => obj value: 27, # solutions = 2
    iteration 3 => obj value: 28, # solutions = 23
    iteration 4 => obj value: 28, # solutions = 21
    iteration 5 => obj value: 29, # solutions = 20
    iteration 6 => obj value: 29, # solutions = 26
    iteration 7 => obj value: 29, # solutions = 19

    I thought it could be due to scaling issues between my variables but after i converted all my continuous variables so my problem only has integer and binary variables i still get this problem. I was wondering what other parameters i might have to set to make sure at each iteration i capture all the solutions.

    Thank you.
    #CPLEXOptimizers
    #DecisionOptimization


  • 2.  Re: Solution Pool

    Posted 02/21/12 02:39 AM

    Originally posted by: SystemAdmin


    First of all, a problem may have many different solutions that have the same objective function value but different x vectors. So in general it is not unexpected to find several solutions with the same objective function value.

    How do the cuts look like that you specify?
    Did you consider using cutoff parameters instead of cuts (CPX_PARAM_CUTLO or CPX_PARAM_CUTUP? Or range filters for the solution pool (see here)? They might work better than your cuts.
    #CPLEXOptimizers
    #DecisionOptimization


  • 3.  Re: Solution Pool

    Posted 02/21/12 09:04 AM

    Originally posted by: kc78


    Hi Daniel

    Thank you for your suggestion! How can i implement diversity filters in the cplex MATLAB class? Is there a cplex.readFilter method to call on?
    #CPLEXOptimizers
    #DecisionOptimization


  • 4.  Re: Solution Pool

    Posted 03/04/12 07:50 AM

    Originally posted by: SystemAdmin


    I'm afraid I have to confess that solution pool filters are not (yet) supported in the matlab connector for CPLEX.
    #CPLEXOptimizers
    #DecisionOptimization