Forum Discussion

FICO's avatar
FICO
Community Manager
10 years ago

General nonlinear mixed integer problems

Hello,

I'm trying to solve a nonlinear mixed integer problem (quadratic objective function, quadratic constrains, non convex, mixed integer) using mmxnlp, but I'm getting a feasible solution that is not optimal, I found a solution that is better that the one Xpress is giving. Can Xpress solve this kind of problems? If yes, why am I getting this wrong solution?

Thank you very much.

fico® blaze advisor®

1 Reply

  • FICO's avatar
    FICO
    Community Manager
    10 years ago

    Hi there,

    if the problem is nonlinear nonconvex it is not possible to solve it to global optimality unless you have a global solver. With nonconvex objective function and/or nonconvex constraints, Xpress uses SLP, a nonlinear solver that returns a local optimum, but not necessarily a global optimum. It is therefore possible, as you report, to have solutions that are better than that found by SLP.

    Hope this helps. Regards,

    Pietro

Related Content

Related Content