Search code examples
pythoncplex

Making CPLEX use a greedy solution


How do you make cplex use a greedy optimization solution as opposed to the optimal solution? Are there parameters you can set or is this not possible?


Solution

  • What you can do is to compute the greedy solution yourself and then submit this as a warmstart/MIP start.

    There are no parameters to force CPLEX to use a greedy algorithm to construct a solution since it is not clear how weights for a greedy algorithm should be chosen for an arbitrary MIP. So for a general MIP, it is not clear what a greedy algorithm should do exactly and whether it would have a chance to succeed or not.

    If a model is suitable for a greedy approach then it frequently is easy to implement the greedy approach in an external tool and just warm start CPLEX with that solution.