> Another interesting thing is that, using EXACTLY the same parameters, the > compilation of glpsol 4.9 I downloaded from http://winglpk.sourceforge.net/ > is unable to get the gap below 4.6% (the optimum is 0%), whereas it seems to > solve fairly easily in the GUSEK 0.2 Windows IDE. This is exactly the > opposite to another problem I have also been working on, in which I used > GUSEK for development only because it took around 5x longer to resolve in > that program.
It is probably because GUSEK uses a more recent version of GLPK than 4.9 and the newer versions implement cut generations. Try enabling / disabling cut generation, it may lead to significant differences in the computation time. Good luck, Ali _______________________________________________ Help-glpk mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-glpk
