A question about Julia Jump with Python cvxpy package


#1

Author: DANQING ZHANG

Hi all,

I am a newbie to Julia optimization, and I was using cvxopt and Gurobi ,SCS solvers in cvxpy. I am wondering if it will be faster in Julia if I use the same solver as in Python cvxpy, for example Gurobi for mixed integar programs.

I am now having some running time issues with my Python program, and wonder if it will be faster in Julia to solve this. Thanks!


#2

The answer to this question depends on where the code is facing the bottleneck.

If you are solving mixed integer programs then once the matrices describing the problem are handed off to the solver the speed will be nearly exactly the same across languages. The solver will not get faster by switching languages (they are all handing it off to the same code on the backend) – However, it may be faster to do the things before and after the solver in different languages.