Originally posted by: oarslan
Hi!
When solving a model (call it cplex1 object) using callbacks by a branch and cut algorithm, I solve another cplex object (say cplex2) in the callback to generate cuts. cplex2 is solved in a loop for several times at each call to the callback.
My question is how to limit all processes in a single thread. In Java, I set IloCplex.Param.Threads equal to 1 for both cplex objects. But each time I start running the model (cplex1), the memory shows more than 100% (reaching up to 400%). This is related to the cplex2 object, because when I don't run it, the CPU never exceeds 100% (In such a case, I generate cuts by a heuristic, however I need exact separation by cplex2).
I am assuming that each time a cplex object is solved, a thread is reserved. However in my application, the two objects should not be running concurrently, because when generating the cuts, I think the cplex1 object simply 'pauses'.
I am trying to limit everything to a single thread because it is the policy of the university. They only allow a model to run in a cluster of computers if it uses only single thread.
Thank you!
Okan.
#CPLEXOptimizers#DecisionOptimization