Decision Optimization

Decision Optimization

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


#Analytics
#DecisionOptimization
#DecisionOptimization
 View Only
  • 1.  getFeasibility() and eliminated columns

    Posted 10/08/14 10:13 AM

    Originally posted by: VKV7_Anulark_Naber


    Is there any relationship between getFeasibility(binVar) and a variable eliminated by PreSolve? What I understand is that a binary variable that attains an integer value (o or 1), regardless of whether (1) it is fixed by branching, (2) its value in a relaxed solution is integer, or (3) it is fixed by PreSolve, will be "Feasible" by using getFeasibility(). For (1), I store the fixed values in nodedata. However, I cannot distinguish (2) from (3). Without knowing all fixed variables at that node, my cuts will not be tight.

    1. How can I get the information of which variables are eliminated or fixed by PreSolve, using C++ API? I saw in other thread that I won't be able to get this information except from the callable library. Is there any other way around it?

    2. Can I presume that all "Feasible" variables will remain integers at their current values in the subsequent (children, grand..., etc.) nodes?

    Thanks in advance.


    #CPLEXOptimizers
    #DecisionOptimization


  • 2.  Re: getFeasibility() and eliminated columns

    Posted 10/09/14 07:02 AM

    1. getFeasibility() returns a value from the IntegerFeasibility enumeration and the reference documentation for that enumeration specifies that variables that were presolved out have a value of ImpliedFeasible.

    2. Of course you can not assume that. That a variable is integer in the solution to the current relaxation does not imply that it is integer in all relaxations in the subtree rooted at the current node.


    #CPLEXOptimizers
    #DecisionOptimization


  • 3.  Re: getFeasibility() and eliminated columns

    Posted 10/09/14 07:15 AM

    Originally posted by: VKV7_Anulark_Naber


    Great, that was what I missed. Thank you very much.


    #CPLEXOptimizers
    #DecisionOptimization