I am solving a nonlinear numerical optimization problem in Pyomo using Scip as a solver. The goal is to minimize a certain objective function. For certain input conditions, I notice that the solver yields final optimal value of the objective function that is higher than the initial objective function value. I am wondering as to why this could be the case. Any comments would be helpful.

  • 11
    $\begingroup$ you should edit your question to include a minimal reproducible example $\endgroup$ – Michael Feldmeier Jul 8 '19 at 11:30
  • $\begingroup$ Is your problem convex or non-convex? Can you describe the cases where you get the strange results, do they share certain features? How about numerical stability of your problems (try to print some statistics of your models)? $\endgroup$ – JakobS Jul 8 '19 at 12:07
  • 3
    $\begingroup$ Are you sure your variables are initialized to a feasible point when you're checking the initial objective function value? $\endgroup$ – Bethany Nicholson Jul 8 '19 at 13:30
  • 2
    $\begingroup$ There may also be numerical problems at play. A minimal reproducible example would indeed be helpful. $\endgroup$ – Kevin Dalmeijer Jul 8 '19 at 20:50
  • 1
    $\begingroup$ Some people desire to close this as "too broad", I do not share that opinion. Is a better choice: "Unclear - Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking."? --- Comments request clairification, yet others (whom did not comment) seem to agree it is too broad. The closure reason on a question should be based on the contents of the question, not on what might be envisioned as a potential answer. --- your_boy_gorja, please improve your question; if you need an answer. $\endgroup$ – Rob Aug 12 '19 at 18:49

Browse other questions tagged or ask your own question.