Need a help with the suboptimal solution obtained with the barrier-method ?

Answered

Comments

1 comment

  • Jaromił Najman

    Hi Gayan,

    It seems that your model is on the edge of (in)feasibility. This can happen even if the coefficient ranges are OK, e.g., if the feasible region is extremely small. You can see this behavior by the primal and dual residuals suddenly increasing by a lot.

    Sub-optimal termination is printed when the barrier algorithm does not make any (or only very small) progress in the solution value and cannot decrease the dual residual, meaning that although the final convergence criterion is met, the dual infeasibility cannot be reduced. In your case, the residuals increase rapidly and Gurobi reports the last good enough solution. This is often the case for shaky numerics.

    You can try to overcome this issue by either reformulating your problem and/or experimenting with the following parameters:
    - set NumericFocus to 2 or 3
    - experiment with the value of BarCorrectors, e.g., set it to 10 or 1000
    - turn off Presolve
    - decrease the tolerances BarConvTol, BarQCPConvTol, FeasibilityTol

    Best regards,
    Jaromił

    0
    Comment actions Permalink

Please sign in to leave a comment.

Powered by Zendesk