Originally posted by: PetBe

Hi guys,

I have been using GAMS with CPLEX as the solver. For a specific model I get the message "Resource limit exceeded" and the gaps are bigger than the desired gaps. My question now is, whether there is a way to tackle this problem. And what exactly is meant by Resource? The memory?

Thanks for you comments in advance

Originally posted by: EdKlotz

This is a GAMS message rather than a CPLEX message, so you would probably be better off posting this on a GAMS forum.   However, googling for ' GAMS "resource limited exceeded"' eventually led to https://gams.com/24.8/docs/userguides/mccarl/reslim.htm:

This option specifies the maximum time in seconds that the computer can run during execution of a solver, before the solver terminates the run.  This option is used by setting

Option Reslim=realnumber;

The default value is 1000.

So try increasing this value in your GAMS specifications before the solve.   If you think CPLEX ought be able to solve the model faster, then post the CPLEX node log here and maybe someone will have some suggestions for improving performance.   Also, have a look at the CPLEX user manual regarding improving MIP performance, and take a look at the technote at http://www-01.ibm.com/support/docview.wss?uid=swg21400023.

Originally posted by: PetBe

Hi,

Thanks EDKlotz for your answer. I increased the number to 3600 seconds (=1 hour) but it was still not sufficient. I attached the log so maybe you'll notice something which is striking. Because I think that this model should be solvable in shorter time. If I use a slightly modified version of the model where one Parameter COP which is used in the constraints is constant and not time depending (COP (t)) then this model is solved within 30 seconds.

Thanks in advance

EDIT:

Something has to be wrong either with GAMS or the Solver. Because I tried to solve a model that I previously solved within 3 or 4 minutes. I did not change anything on that mode. Nonetheless now the solver needs more than 3600 seconds to solve it. This is really strange. I also uploaded the log for that (log2).

About the second log: if you kept the model and its data exactly the same, the question is: what else changed? Did you change computer, did you change the CPLEX or GAMS version? Did you change any parameters?

For the first log: I would try and see if setting CPX_PARAM_MIPEMPHASIS to different values (I don't know how to do that in GAMS) can help.

Originally posted by: PetBe

Does anybody have an idea? Because for me this kind of behaviour is super strange. I mean a model that was solved within 30 seconds with a gap of 0 now can just be solved with a huge gap. And I rellay have not changed anything :-(