vOptLib: Library of numerical instances for MultiObjective Linear Optimization problems

About

vOptLib (short for vector optimization library) is a collection problem instances for benchmarking multi-objective solvers. It covers a variety of Multiobjective linear optimization problems (multiobjective combinatorial problems, multiobjective integer linear programs, multiobjective mixed integer programs).

Available online since 1998, and redesigned in 2017, the repository is currently maintained by Xavier.Gandibleux@univ-nantes.fr If you have produced your own instances and you would like to become contributor to vOptLib with your collection, please do not hesitate to contact me.

News

03-Sep-2022: instances of 0/1 uncapacited facility location problem uploaded
12-Jul-2021: instances of set partitioning problem uploaded
02-Sep-2017: moved from gitlab to here (repository on gitlab will be closed soon)
05-May-2017: New repository launched.

Problem Classes

vOptLib is organized in structured and non-structured problems. It distinguishes the following problem classes:

• Multi-objective structured problems

  Several linear objectives; feasible region is a polyhedron with a structure; all variables are restricted to be binary.

  • Linear assignment problem (LAP)

  • 0/1 unidimensional knapsack problem (UKP)

  • 0/1 bidimensional knapsack problem (BKP)

  • Set covering problem (SCP)

  • Set partitioning problem (SPA)

  • Set packing problem (SPP)

  • 0/1 uncapacited facility location problem (UFLP)

  • Single source capacited facility location problem (SSCFLP)

  Several linear objectives; feasible region is a polyhedron with a structure; some but not all of the variables are restricted to be integer.

  • Uncapacited mixed facility location problem (UMFLP)

  • Capacited facility location problem (CFLP)

• Multi-objective non-structured problems

  Several linear objectives; feasible region is a polyhedron; all variables are restricted to be integer.

  • Integer program (MOIP)

  Several linear objectives; feasible region is a polyhedron; some but not all of the variables are restricted to be integer.

  • Mixed integer program (MOMIP)

Download

The following instances can be downloaded and used for free. If you use them for your research, we would appreciate a reference to vOptLib in your publication.

History

Created in 1998 and available online since, the library -originaly named MCDMlib and later the MOCOlib- was dedicated to MultiObjective Combinatorial Optimization problems. It is referenced from:

• the International Society on Multiple Criteria Decision Making, section Digital Library
• the OR-Library, item Multiobjective optimisation

Amongst the instances available, the library hosts instances used by us during the following research projects:

• ANR project GUEPARD: Guaranteed efficiency for pareto optimal solutions determination in multiobjective combinatorial optimization problems
• ANR-DFG project vOpt: Exact efficient solution of mixed integer programming problems with multiple objective functions

The obsolete (but active) URLs are:

• MCDMlib (since 1998)
• MOCOlib, the MOCO section of the MCDMlib (since 2007)

Others collections

• Zitzler-Laumanns’s instances (since 2001)
• Joshua Knowles’s instances (since 2002)
• Luís Paquete’s instances (since 2004)
• José Luis Esteves dos Santos’s instances (since 2005)
• SOA problem instances (since 2008)
• GUEPARDlib (since 2010)
• MoCObench (since 2012)
• Gokhan Kirlik’s instances (since 2014)
• MOPLIB (since 2015)
• Fritz Bökler’s instances (since 2015)
• Boland-Charkhgard-Savelsbergh’s instances (since 2015)
• MOREPO (since 2017)
• Thibaut Lust’s instances (since 2017)
• Aritra Pal’s instances (since 2017)

If you are aware of the existence of a collection available online that is not listed here, please do not hesitate to contact us for adding a link to it.