0-1 Integer Optimization Benchmarks

This page contains a collection of real and realistic benchmarks for general 0-1 integer optimization. Each benchmark contains a mix of inequality constraints over 0-1 variables and varies in size or constrainedness. The benchmarks are available in a number of different formats:

.mps, the standard format of many IP solvers
.opb, a format suitable for lp_solve and OPBDP
.oip, the format of `over-constrained integer programs,' suitable for WSAT(OIP)

Unless otherwise noted, problem size is reported "number of variables" x "number of constraints".

Benchmarks:

Scheduling of a basketball season (the ACC 97/98 problem).
Problem instances have an increasing constraint tightness, and have a size up to 1773x3479.
- Problems in .mps format: acc-benchmarks-mps.tar.gz [tightness level 0-6] (495K).
- Problems in .opb/.oip format: acc-benchmarks-oip.tar.gz [tightness level 0-9] (310K).
Radar surveillance covering:
12 test instances, randomly generated as described in Solving Linear Pseudo-Boolean Constraint Problems with Local Search and in a forthcoming SICS technical report. Problem sizes are 10x10, 10x20, 30x30, and 30x70 cells; the problems also vary in the percentage of insignificant cells (0% and 2%) and spread (even and uneven).
- Problems in .mps format: radar-mps.tar.gz (1.45M)
- Problems in .opb format: radar-opb.tar.gz (930K)
- Problems in .epb format: radar-epb.tar.gz (494K)
The progressive party problem instances
- Problems in .opb format: 1.15M compressed tar file containing six `.opb' files
- Problems in .oip format: 968K compressed tar file containing six `.oip' files
Assignment: Course allocation for the law school of the Universität des Saarlandes
5 test problems varying in size from (906x640 to 8404x11350)
- Problems in .mps format: course-ass-mps.tar.gz (395K)
- Problems in .opb format: course-ass-opb.tar.gz (262K)
- Problems in .oip format: course-ass-oip.tar.gz (174K)

This page maintained by J.P. Walser
Last Updated: October 6, 1998