Originally posted by: suji248

Hello CP Experts,

I am beginner  and started using CP optimizer just recently . I have a scheduling problem having all integer variables. When I solve using CPLEX optimizer it takes a long time to give a feasible solution.  I want to try if CP optimizer will be able to help me  get a feasible solution faster . 

1) My first question is it possible to directly import the MPS file using C# API and solve using CP engine.

2) Will the performance improve if I model using the decision variables and constraints of  CP optimizer? 

Regards

Sujay

Originally posted by: PhilippeLaborie

Hello,

For question 1: It is not possible in .NET but you could do a write a small C++  code that does this translation, see https://www.ibm.com/developerworks/community/forums/html/topic?id=d2416a5b-5729-47a5-a2e7-65fb65d19a77&ps=25.

This being said, and this will answer your second question, if you are solving a scheduling problem (where important decision variables of the problem are the start/end time of some activities) , you should try to leverage the concepts of CP Optimizer dedicated to scheduling problems, in particular the notion of interval variable and the related expressions/constraints. Solving with CP Optimizer the very same CPLEX model, especially for a scheduling problem, does not really make sense. If you are familiar with CPLEX, reformulating the problem in CP Optimizer should not be a very big conceptual change but you will have to learn the additional types of decision variables and constraints available in CP Optimizer.  There are a number of classical scheduling problems (like jobshop, RCPSP) modeled with CP Optimizer available in the distribution (CPLEX_Studio/cpoptimizer/examples/src/csharp). For a very introductory description you could look at: https://www.slideshare.net/PhilippeLaborie/an-introduction-to-cp-optimizer.

You can also shortly describe your problem in this forum if needed.

Philippe

Originally posted by: suji248

Am working on an assignment problem having purely set partitioning constraints. The number of variables are around 4 millions. But I observe that CPLEX is taking lots of time to find a feasible solution. 

Hence I was trying to explore will using CP optimizer help me in finding a good  feasible solution quickly. 

Regards

Sujay