Originally posted by: A6UM_Can_Nguyen

Hi Alex,

Could you help, how to model work faster?

My the model is MIP, with objective function is that minimize costs.  However, the model can not give results when it runs over 10h and more time.

The computer is 64 bit, core i7, RAM 32G

Thank you,

Bests,

Nguyen

Hi

1)

How many decision variables ? Have you tried any decomposition yet ?

2)

You could have a look at https://www-01.ibm.com/support/docview.wss?uid=swg21400023

regards

Originally posted by: A6UM_Can_Nguyen

Hi Alex,

For (1), decision variables are below:

For (2), I used the Emphasis parameter of 3.

execute
{
cplex.covers=3;
}

Could you tell me how to get quick results?

Thank you,

Bests,

Nguyen

Hi,

if you want to set mip emphasis, try mipemphasis instead of covers.

Do not hesitate to try to spend some time on improving your model.

regards

Originally posted by: A6UM_Can_Nguyen

Hi Alex,

I will try, thanks.

That means mipemphasis=3. Right?

execute
{
cplex.mipemphasis=3;
}

Bests,

Nguyen

The technote suggest 1:

If you don't need an optimal solution, set the MIP Emphasis parameter to 1 so that CPLEX finds more feasible solutions.

3 means

Originally posted by: A6UM_Can_Nguyen

Hi Alex,

I need an optimal solution, but the model has not given results.

Thank you,

Nguyen

Hi,

have you thought about decomposition ?

can you try to aggregate some decision variables in order to make your model smaller ?

In CPLEX > User's Manual for CPLEX > Programming considerations > Developing CPLEX applications

Prototype the model

regards

Originally posted by: A6UM_Can_Nguyen

Hi Alex,

I can not use the decomposition, because it is too difficult for me.

At the present, my model is working about 3 days, and then I get an error as the figure below. Could you help me how to solve it?

Thank you,

Bests,

Nguyen

Originally posted by: A6UM_Can_Nguyen

Hi Alex,

My model is working about over 50 hours, but I don't get the results, please attach the log file, could you tell me how to run faster?

Thank you,

Bests,

Nguyen

Originally posted by: A6UM_Can_Nguyen

Hi Alex,

How to set up a startalgorithm parameter set to 4 to use the barrier method at the node LPs?

Thank you,

Bests and regards,

Nguyen

Hi,

have you tried

cplex.lpmethod=4;

?

You could also use a .ops file in your run configuration

regards

https://www.linkedin.com/feed/update/urn:li:activity:6428601933681950720

Originally posted by: A6UM_Can_Nguyen

Hi Alex,

The time limit that can not be a good way, because I need an optimal solution.

I am trying cplex.nodealg = 4, and the model is running that spent over 2.5 hours. 

Thank you,

Bests,

Nguyen

Originally posted by: A6UM_Can_Nguyen

Hi Alex,

The model uses cplex.lpmethod=4, and the model is running that spent over 4 hours (see figure below).

Could you give me an advice, how to solve it?

 

 

Thank you,

Nguyen

Originally posted by: A6UM_Can_Nguyen

Hi Alex,

Please try to tell you about the objective function is to minimize costs

In the model, I see that when I use the constraint below then model run with a very long time (3 days) but I don't get the results. Could you have any advice to help me?. or do you have any way to write the constraint to model run faster? 

sum (<i> in tuple_flow) F[<i>] >= 50000;

Thank you,

Bests,

Nguyen

Hi,

I see you get a solution.

You could set a time limit and then after the time limit your solve will stop

execute { cplex.tilim=60; }

regards