Originally posted by: SébastienM

Hi All,

As many other before, i'm a new user of glpk and cplex. I'm currently facing a strange problem and i hope that some one may understand it or tell me how to fix it.

I'm using glpsol to convert a mathprog model (model.mod) with data (model.data) into a cplex model (model.lp). So far every thing is fine! Then i tried to solve that model using cplex interactive and oplide (ibm ilog cplex Optimization Studio). The problem is that the time to solve the problem with this two ways are totaly different. I'm going from 52 minuts with oplide to more than two hours with cplex interactive.

So i'm wondering if :

- The way glpsol write model.lp isn't that good and may impact a lot my performances (by the way when i use oplide, i write a new file containing the model and new data so that it can be read) ?
- Some options that i don't know yet might be used by oplide by default and help it to solve the problem way faster than cplex?

If some one has faced that problem and can tell me what is the problem, i would be very glad.

Many Thanks,

Regards,

SébastienM

#DecisionOptimization
#MathematicalProgramming-General

Hi,

in the OPL IDE have you used OPL or do you read the lp file and solve it ?

regards

#DecisionOptimization
#MathematicalProgramming-General

Originally posted by: SébastienM

Hi,

First of all thank you for answering.

With OPL IDE i created 2 new files : one for the model and one for the data, using the proper IDE langage. I didn't read the lp file, i didn't knew that it could be done with OPL IDE to be honest.
#DecisionOptimization
#MathematicalProgramming-General

Hi,

in OPL you can read and solve an lp file:

main

{

cplex.importModel("file.lp");

cplex.solve();

}

I guess that using OPL you wrote a better model and that is why you get better performance.

Regards

#DecisionOptimization
#MathematicalProgramming-General

Originally posted by: SébastienM

Hi,

Again thank you for answering. The two model are the same and after trying your idea, i found that it wasn't my problem. The problem was that oplide was working locally on my computer unlike cplex wich worked on the university server (i didn't think about that)... so the communication bursted my solving time.

After trying Cplex locally on my computer, the solving time was much much better and very close to the OPL IDE time.

Thank you for your help and have a good day :)

Regards,

Sébastien M.

#DecisionOptimization
#MathematicalProgramming-General