Does Gurobi support mathematical integers or is it limited to machine precision for Integer programs?

Answered

Ryan Brummet

• July 22, 2019 16:58

Suppose that I have the very simple helloworld style integer program below:

 

Max X s.t.

X * W <= A
X <= 1
0 <= X


Where X is the integer variable (can be 0 or 1), and W and A are constant integers.

Suppose further, that W is very large, for example, 2251799813685248.

With such a large size, using standard primitives (like a 32 bit int) will cause problems.

Therefore, my question is does Gurobi support mathematical integers (i.e. no bounds and uses something like GMP under the hood) or is it limited to machine precision?

0

• Official comment

  

  Greg Glockner

  Gurobi Staff

  • July 22, 2019 20:06
  • Edited

  Gurobi uses standard mixed-integer programming (MIP) techniques: it solves MIP models via a series of continuous (linear programming) relaxations. All variables are therefore continuous (double) values; a value is considered an integer if it is within IntFeasTol of an integer value.

  Additionally, a large coefficient like 2251799813685248 will cause numerical problems for the underlying LP relaxation.

  (I moved this to the modeling topic).
• Official comment

  

  Simranjit Kaur

  Gurobi Staff

  • December 06, 2024 00:55

  This post is more than three years old. Some information may not be up to date. For current information, please check the Gurobi Documentation or Knowledge Base. If you need more help, please create a new post in the community forum. Or why not try our AI Gurobot?.

Post is closed for comments.