Gurobi NEOS issue

Answered

A.Omidi

• March 24, 2020 22:05

Dear support team,

 

I'm trying to solve a large scale optimization problem using Gurobi solver (V.9) on my local machine. As I have not accessed to an academic licence for Gurobi, I will need to use NEOS to lunch my problem. I wrote my code in the Java but I can translate it to Python. What I'm looking for is, I have used some advanced features of the solver such as callback function, etc which can be used directly on the local machine through a programming language. 

I was wondering if, is there any way to deal with the NEOS/Gurobi solver using a programming language like Python? (I mean using the Gurobi solver directly and without using a framework like Pyomo.).

 

 

Thanks in advance

A. Omidi

0

• 

  Eli Towle

  Gurobi Staff

  • April 04, 2020 00:00

  Hi,

  A list of NEOS interfaces can be found here. There isn't a way to directly submit a job to NEOS using only the Gurobi Python interface. You could always build the model in Python, save it as an MPS file, then submit the MPS file to NEOS to solve.

  NEOS passes a model file, as well as any parameter, solution, etc. files provided by the user, to Gurobi to solve. Thus, you aren't able to use callbacks with the NEOS server. The closest you could get to a callback is to a priori enumerate a set of lazy constraints to use, set the Lazy attributes to a nonzero value, then submit the job to NEOS using the parameter LazyConstraints=1. Even this is limited in its practicality, since in many cases it is impractical to enumerate all possible lazy constraints.

  To use all of the advanced features of Gurobi, the best solution is to obtain an academic license for your local machine. If you are an academic but unable to obtain an academic license for any reason, please submit a request through the online support portal and we will help you obtain one. Thanks!

  Eli

  0

Please sign in to leave a comment.