This code is meant to isolate a weird behavior by CPLEX (12.7.1 and 12.8).
When adding the optimal solution for the markshare2
problem, provided in
MIPLIB2017,
as a MIP start, CPLEX claims that no feasible solution was found from it. This does not
happen when using Gurobi.
Edit: I had a dumb bug in my code. This code reads a .SOL solution (as provided by MIPLIB) and adds it as a MIP start. There is a CPLEX and a Gurobi version.
- Boost (just for the split function for strings).
- Update paths in makefile
- For CPLEX, use
make cpx
. For Gurobi,make grb
.
./example{_grb} data/markshare.mps.gz data/markshare.sol