> I attach a picture of a > 100% gap on integer programming on GLPK. > Some minutes later, the gap even increased to > 200%. > > > Is it normal? > > > If you need, I can send you the model.
Yes, this can happen if the best integer feasible solution (incumbent value) and the best solution to lp relaxation have different signs. See the formula for computing mip gap in the glpk reference manual: http://en.wikibooks.org/wiki/GLPK/Terminal_output _______________________________________________ Help-glpk mailing list [email protected] https://lists.gnu.org/mailman/listinfo/help-glpk
