Dear all, I was working on interfacing NLOpt in the finite element PDE solver freefem++ in order to provide the nlopt optimization algorithms to the PDE software's users. All the interfaced algorithms work well except StoGO (and its randomized version), i always catch a std::runtime_error at the beginning of the algorithm (before even a single cost function evaluation). I noticed it worked when I used it in the NLopt test.cc file, i was then wondering if there was something special to specify to the StoGO optimizer (a particular sopping criterion? etc...), nothing is mentioned in the online documentation except that upper and lower bounds are needed (which I have provided, as well as the gradient expression of course), i tried to set some parameters tolerance related sopping criteria, or function value, elapsed time, max number of fitness evaluation, but none of these seems to make it do the work.
Thanks for any help! Best regards, Sylvain Auliac Laboratoire Jacques-Louis Lions Université Pierre et Marie Curie mail: [email protected] tel : 01 44 27 71 70 _______________________________________________ NLopt-discuss mailing list [email protected] http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/nlopt-discuss
