I've been running a very tough pure binary LP problem with zero objective function in Gurobi. I'm happy to wait several months if that's what it takes to get a solution. I have experimented with various parameters including tuning with smaller models. The question I have concerns the IntInf parameter in the Current Node column of the branch-and-cut tree search log. In another post someone made a comment that I interpreted to mean that the IntInf parameter ('number of integer infeasabilities' ?) should be going to zero. I attach a graph I plotted of IntInf against time (in seconds). The graph does suggest to me that IntInf is going to zero, and if things progress at roughly the same rate I estimate that IntInf will reach zero in about 6 - 10 weeks. So is my interpretation correct here? Does the log data here suggest I'm heading towards an optimal solution? Thank you, Marcus.
Please sign in to leave a comment.