I have a MILP Problem with 2 millions of variables and 8000 of constraints. I set time limit for cplex as 3 hours. I am getting around 7% MIP gap. But I want to solve this problem to optimality.

Problem description: constraints matrix corresponding to some variables are sparse and for some variables it is dense. I am using barrier with dual crossover at root node and dual simplex at other node.

Could you help me how can I achieve optimal solution with atleast 1% MIP gap ?

asked 22 Jul '16, 04:17

Amit%20Sarkar's gravatar image

Amit Sarkar
accept rate: 0%

Provide a link to the mps-file and we can take a look at it.

(22 Jul '16, 16:29) opti100

First comment: You may not be able to reduce the gap to 1% in three hours.

Second comment: CPLEX has a huge number of configurable parameters, and there's no way of knowing a priori which combinations might help. There's a tuning tool included with CPLEX. You might want to run it on a substantially scaled down version of the problem and see what combination of parameter changes it recommends.

Third comment: Any ideas that might result in faster closure of the gap will likely need to exploit specific elements of the problem structure (more than just sometimes it's sparse, sometimes it isn't). You might profit from looking for papers that solve similar types of problems and see how they go about it.


answered 22 Jul '16, 16:09

Paul%20Rubin's gravatar image

Paul Rubin ♦♦
accept rate: 19%

Your answer
toggle preview

Follow this question

By Email:

Once you sign in you will be able to subscribe for any updates here



Answers and Comments

Markdown Basics

  • *italic* or _italic_
  • **bold** or __bold__
  • link:[text](http://url.com/ "Title")
  • image?![alt text](/path/img.jpg "Title")
  • numbered list: 1. Foo 2. Bar
  • to add a line break simply add two spaces to where you would like the new line to be.
  • basic HTML tags are also supported



Asked: 22 Jul '16, 04:17

Seen: 804 times

Last updated: 22 Jul '16, 16:29

OR-Exchange! Your site for questions, answers, and announcements about operations research.