Hi, thank you for adding the parameter for the mip gap tolerance. I just tried it and it make the search stop faster with a already very good solution. :-) However, as return code from lpx_intopt I get LPX_E_TMLIM (time limit exceeded), which I do not really understand. It is probably caused by the "ret = GLP_ETMLIM;" in glpios03.c line 1346.
Wouldn't it be better to return LPX_E_OK, since the user requested to stop with this relative tolerance, so this is no error. And probably lpx_mip_status should report only LPX_I_FEAS if the tolerance is greater 0 (I haven't checked what comes out there now). Btw, is there a function to get the bound from the lp relaxation after a solve with the advanced B&B algorithm and MIPGAP>0 ? Best, Stefan -- Stefan Vigerske Humboldt University Berlin, Numerical Mathematics http://www.math.hu-berlin.de/~stefan _______________________________________________ Bug-glpk mailing list [email protected] http://lists.gnu.org/mailman/listinfo/bug-glpk
