0
I have a solver called Solvopt and it is good to solve my problem of Soft and unrestricted minimizing, but usually it stops at minimal locations. On the university computer I have all the toolboxes installed, but I want to use some good heuristics, like the VNS, already implemented in MATLAB to globally optimize my specific problem using Solvopt, preferably in parallel. Is there any way to do that?
Your question is a little generic, you want to use something to optimize variables? If yes a common option is fminch . Also, you can take a look at the Matlab page for nonlinear optimization. Give a comment if it’s close to what you’re looking for.
– Guto
So... yes, Matlab has these heuristics already programmed, but they do not allow me to use Solvopt in these heuristics, but only those that already exist programmed in MATLAB. The idea of using Solvopt within some other heuristics is that Solvopt is an excellent solver of my particular problem and I do not lose the efficiency of my method. Imagine if, by switching Solver, I start to get worse results? It would be sad, no?
– R. W. Prado
Nonlinear problems sometimes have non-linear answers :) . You know how to integrate Solvopt into your code then?
– Guto
My question is for me, after having several approaches, compare them and see which one is best. It turns out that my problem is already well solved with Solvopt, so it would be worth investigating if there are any heuristics that allow me to use my Solver internally in it so that I can get the most out of my Solver. After that, I compare with the other heuristics that Matlab provides. This step of comparing with those already made of MATLAB will happen. = ) The hope is that using my software things get better, because it already solves well.
– R. W. Prado
Hello, Guto, yes, I know. My problem is already "solved" (theoretically it is possible to improve, but in practice it is the best there is so far) with Solvopt, but it is common for Solvopt to fall into local minimums, precisely because my model has many local minimizers.
– R. W. Prado