MIPLIB 3.0

Abstract:

In response to the needs of researchers for access to real-world mixed integer programs a group of researchers at Rice University created MIPLIB, an electronically available library of both pure and mixed integer programs.

Since its introduction, MIPLIB has become a standard test set used to compare the performance of mixed integer optimizers. Its availability has provided an important stimulus for researchers in this very active area.

Version 3.0:

Due to improvements in state-of-the-art optimizers, as well as improvements in computing machinery, several instances have become too easy; simultaneously, a need has emerged for more difficult instances. Finally, it has been observed that certain types of problems were overrepresented in former versions of MIPLIB and others underrepresented. In addition, we have included, for each problem, certain auxillary information describing the structure of the constraint matrix. The purpose of this information is to identify constraint classes that may be useful in the various phases of problem solving, such as preprocessing, constraint generation, and branching. These considerations have prompted the present update, MIPLIB 3.0.

Developers:

Robert E. Bixby, Rice University

Sebastian Ceria, Columbia University

Cassandra M. McZeal, Rice University

Martin W.P. Savelsbergh, Georgia Institute of Technology

Files:

Catalog of models contained in MIPLIB 3.0.

Introduction to the MPS expression of linear problems.

The MIPLIB 3.0 Problem Set

MIPLIB 3.0 Technical Report

Previous Version of MIPLIB

Submissions:

Since mixed integer programming is such an active research area, and the performance of optimizers keeps improving, we anticipate that this update will not be the last. Subsequent updates are planned on a yearly basis. We encourage both researchers and practitioners in integer programming to submit real-world instances for consideration and possible inclusion in MIPLIB.