> I #39;m trying to solve a series of LP problems, the former > solution is a feasible and suboptimal solution to the latter one, so > in each time, start optimization problem from a suboptimal solution > sounds to be faster than from zeros. So does anyone know about it > please share me the way. Thank you very much!
If you need to reoptimize lp, you should do nothing. By default glp_simplex starts the search from the current basis stored in glp_prob object. So if you have obtained some optimal basis, you can modify the problem object and then just call glp_simplex. Note, however, that in this case you should not remove active (non-basic) rows and/or basic columns in order to keep the current basis valid. _______________________________________________ Help-glpk mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-glpk
