Dear Experts

I have modeled a timetabling problem as MIP in GAMS and have solved it by Cplex.

I have a problem with time of solving the model. In each run of model

I get different time and objective function. some of this is as:

# processing time (second) objective

1 720 112146

2 779 112146

3 14460 116340

4 4160 116340

5 3668 116692

6 520 108615

The environment of processing in all of them is the same.

I have confused about Cplex solver. If it use an exact method, the

time of processing and objective function must be a given value, or at

most 10% for windows bugs.

can you explain this problem?

which time is the real solving time for my model?

Is it related to my model?

or is related to Cplex?

how I can fix it?

Thanks

Afshin

\

Processing time may vary because of the background processes on your PC !

I think it’s normal…

On the other hand, objective values should be the same, regardless of processing time, unless you had set a max processing period…

In that case, the solver stops early… The time is not enough to obtain the optimal solution.

On Fri, Feb 3, 2012 at 22:28, afshin oroojlooy wrote:

Dear Experts

I have modeled a timetabling problem as MIP in GAMS and have solved it by Cplex.

I have a problem with time of solving the model. In each run of model

I get different time and objective function. some of this is as:

# processing time (second) objective

1 720 112146

2 779 112146

3 14460 116340

4 4160 116340

5 3668 116692

6 520 108615

The environment of processing in all of them is the same.

I have confused about Cplex solver. If it use an exact method, the

time of processing and objective function must be a given value, or at

most 10% for windows bugs.

can you explain this problem?

which time is the real solving time for my model?

Is it related to my model?

or is related to Cplex?

how I can fix it?

Thanks

Afshin

Aut Viam Inveniam Aut Faciam

–

To post to this group, send email to gamsworld@googlegroups.com.

To unsubscribe from this group, send email to gamsworld+unsubscribe@googlegroups.com.

For more options, visit this group at http://groups.google.com/group/gamsworld?hl=en.

Afshin,

Try if the following helps.

OPTION OPTCR = 0.0;

It is completely normal to get different processing times for the same model as this highly depends on the number of other programs running in your computer while running your GAMS code. Plus the route, the solver takes in search of the global optimal, may be different in which longer routes take longer time and shorter routes do smaller time.

It seems weird (yet sometimes possible) to get a different objective value for each run with the same solver. I assume the solutions that you get are also different although there may be some in common. In MIP, it is also possible to have multiple optimal solutions. This happens when you have flat objective function in the neighborhood of the global optimum.

Bests,

Destin

On Fri, Feb 3, 2012 at 9:28 PM, afshin oroojlooy wrote:

Dear Experts

I have modeled a timetabling problem as MIP in GAMS and have solved it by Cplex.

I have a problem with time of solving the model. In each run of model

I get different time and objective function. some of this is as:

# processing time (second) objective

1 720 112146

2 779 112146

3 14460 116340

4 4160 116340

5 3668 116692

6 520 108615

The environment of processing in all of them is the same.

I have confused about Cplex solver. If it use an exact method, the

time of processing and objective function must be a given value, or at

most 10% for windows bugs.

can you explain this problem?

which time is the real solving time for my model?

Is it related to my model?

or is related to Cplex?

how I can fix it?

Thanks

Afshin

To post to this group, send email to gamsworld@googlegroups.com.

To unsubscribe from this group, send email to gamsworld+unsubscribe@googlegroups.com.

For more options, visit this group at http://groups.google.com/group/gamsworld?hl=en.