- Cplex. Optimal solution \(z=60\) found in about 8 seconds. Using 4 threads on an old laptop. Log is below.
- CBC. Hit timelimit of 2 hours. Objective = 62 (non-optimal). Also using 4 threads on the same machine. The strange thing is that with CBC the best possible bound is not changing at all. Not by a millimeter. See the highlighted numbers in the log below. CBC is a very good solver, but sometimes I see things like this.
- Gurobi and Xpress solve this problem very fast. Ran on NEOS via an MPS file.
- SCIP has also problems with this model. Ran on NEOS via an MPS file.
All runs were with default settings. It looks like the commercial solvers do a much better job than the open source and academic codes. Sometimes you get what you pay for.