Decision Optimization

Decision Optimization

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

 View Only
  • 1.  objective value precision

    Posted Thu May 31, 2012 05:08 AM

    Originally posted by: Sothea


    Dear,

    I would like to know if there is a parameter in cplex-matlab which we can define objective value precision. I mean, for exemple an objective value found 1.1296, how to make it equal to 1.1200 or 1.1300. I think this parameter can reduce calculating time if we dont need high precision. Do you have any idea?

    Thanks in advance,
    Sothea HONG
    #CPLEXOptimizers
    #DecisionOptimization


  • 2.  Re: objective value precision

    Posted Thu May 31, 2012 11:34 PM

    Originally posted by: John Cui


    Which kind problem are you solving? MIP?
    John Cui
    #CPLEXOptimizers
    #DecisionOptimization


  • 3.  Re: objective value precision

    Posted Fri June 01, 2012 02:53 AM

    Originally posted by: Sothea


    Yes, it is MIP problem.
    #CPLEXOptimizers
    #DecisionOptimization


  • 4.  Re: objective value precision

    Posted Fri June 01, 2012 03:06 AM

    Originally posted by: John Cui


    You can consider adjusting the MIP gap parameter.
    If you know your objective values are on the order of say 1e+1, and you are only interested
    in the first two decimal places, you can set the MIP gap to .01/1e+1 = .001.

    While this may yield a performance improvement on a few models, but
    in most cases such parameter changes won't yield a significant improvement.
    You'd better spend your time looking for other ways to improve performance.

    Such as tune MIP related parameters(heuristics, cuts, etc.) or improve your model formulation.

    John Cui
    #CPLEXOptimizers
    #DecisionOptimization