Problem with number of solutions
Answered
I am using gurobi 10.0.0 as well as gurobi 9.1.1 to solve a MILP model which I know it has thousands of solutions. However, whenever the model is optimized, version 10.0.0 reports only 5 solutions and version 9.1.1 reports only 7 solutions. It doesn't matter if I set the "poolSolution" parameter or not since the number of solution is less than the default value for this parameter which is 10.
It shoul be noted that I am sure about the number of solutions which is beyond these reported ones because when I remove the previous solutions from the feasible ones by adding some new constraints to the model, I receive new solutions.
-
With the default value of the PoolSearchMode parameter (0), Gurobi tries to find an optimal solution to the model. The number of solutions Gurobi finds while searching for an optimal solution is no indication of how many feasible solutions exist. Additionally, different versions of Gurobi may take different paths to solve the problem, which can result in the solver finding different numbers of feasible solutions, or even different optimal solutions if multiple exist.
Are you interested in finding more feasible solutions? If so, set the PoolSearchMode parameter to 1 to direct Gurobi to try finding \(n\) feasible solutions, or to 2 to direct Gurobi to try finding the best \( n \) feasible solutions (where \( n \) is equal to the value of the PoolSolutions parameter).
0
Please sign in to leave a comment.
Comments
1 comment