Re: [Gurobi] Decreasing root relaxation time

[Greg Glockner]

Have you checked the memory use on your computer? Are you running 64-bit Gurobi?

[Greg Glockner]

Please post more of the log, along with the output of the Model.printStats() function from the interactive (Python) shell.

[Greg Glockner]

> I am using YALMIP in MATLAB for calling Gurobi, I dont know how to get the Model.printStats().

Export the model in MPS format, then open the interactive shell and do the following commands:

m = read('mymodel.mps')
m.printStats()

changing 'mymodel.mps' to the name of your model file.

[Christopher Maes]

> I was trying to find a way to export my model in YALMIP-MATLAB to MPS
> format, but I didn't find a way. I would highly appreciate it if you can
> help me with this issue.

Johan Löfberg, the author of YALMIP, would be the person to ask.
YALMIP might have a supported way of extracting a Gurobi model.

But from looking at the solvers/callgurobi.m function in YALMIP
R20130405, it looks like if you put a breakpoint in that function
after the line:
model = yalmip2gurobi(interfacedata)

you will have the model struct that is used to call Gurobi. If you
then issue the following commands in the MATLAB debugger:
params = model.params;
params.ResultFile = 'mymodel.mps';
params.TimeLimit = 1;
gurobi(model, params);

you should be able to write your model out as an MPS file.

Chris

[Zengjian Hu]

I 'd suggest the following from the high level.

1. if your problem is scalable, Try to solve some smaller instances and get some stats for the runtime and see how it grows along with the problem size.

2. try to identify a set of constraints that contribute the most to the run time, hopefully you 'll gain some insight and be able to replace/approximate them.

Jian

On Tue, Jun 11, 2013 at 9:17 AM, Mostafa Nick <mosta...@gmail.com> wrote:

Sorry for posting the the question again and thank you very much for your previous responses.

As you asked, I posted the log and printStats of my model.

I would highly appreciate it if you can tell me wether there is any way to decrease the root relaxation time or it is normal.

Thank you in advance

--
 
---
You received this message because you are subscribed to the Google Groups "Gurobi Optimization" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gurobi+un...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Ed Rothberg]

Those are good suggestions.

Another thing to try is to set the MIQCPMethod parameter to 0.  Outer approximation can sometimes be quite expensive.

Ed