>>From what I understand by looking at the GLPK manuals, it does provide you > with the capability to specify a bound on the objective function value > which avoids extra branching. For my problem, I have formulated it as a > MIP optimization problem. Currently, it takes too long to solve it and I > have tried almost all relevant options. The other approach that I think > might work is to employ a heuristic to get a feasible solution and then > feed the > entire solution to GLPK. THe heuristic I have currently is very fast so if > GLPK accepts an initial solution of the MIP and optimizes it furthur, one > would expect the runtime to go down. > > Does GLPK let you accomplish this in some direct/indirect way?
Such feature is not implemented yet. However, it is easy to write the initial incumbent objective value directly into the b&b data structure. If you are interested in hacking, I can explain how to do that. Andrew Makhorin _______________________________________________ Help-glpk mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-glpk
