I also tried to use Glpk 1) as a standalone solver from the command line, and 2) as a library called from my C++ code (after modelling an ILP with FlopCpp and using Osi), and I also experienced different behaviours. In particular, I need to call it from my C++ code, but the command line solver has the best performances on my problems. It is not only a matter of default values, it seems that two versions of Glpk code are being called with the two alternatives. So, my question is: is there a way of creating an ILP model with FlopCpp, and then choosing algorithmic options (cuts, presolve, branch, etc.) and calling Glpk from my C++ code as it would be from the command line? (without saving my model as lp on disk and then reading it by calling the standalone solver with a syscall.) I tried calling glp_main() with the list of algorithmic options as arguments, but I did not find how to pass my model to it. Do you have an example? Thanks a lot.
-- View this message in context: http://www.nabble.com/different-solutions-for-same-code-tp25245249p25946722.html Sent from the Gnu - GLPK - Help mailing list archive at Nabble.com. _______________________________________________ Help-glpk mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-glpk
