Комментарии:
Users need open benchmarks for such claims.
Last time I benchmarked Timefold Solver - our open source solver in Java and Python - against Gurobi on the Vehicle Routing Problem (VRP), I saw far better results with Timefold than Gurobi, especially when scaling up.
I have benchmarked Gurobi, CBC and Bonmin. I have to say, the performance is not even close. I think for convex optimization or general NLPs open source solvers like IPOPT can be competitive, but as soon as integrality constraints are involved I would always recommend Gurobi. I think it would be extremely interesting to work there and always be on the forefront of optimization:)
ОтветитьIMHO, it would be fairer the comparison against more advanced MIP open-source solvers such as CBC, SCIP, MIPCL, or Google OR-Tools to mention a few. For example, Google OR-Tools has a competitive and comprehensive mathematical programming language.
Ответить