Multi-Objective Optimization Repository (MOrepo)

This repository is a response to the needs of researchers from the MCDM society to access multi-objective (MO) optimization instances. The repository contains instances, results, generators etc. for different MO problems and is continuously updated. The repository can be used as a test set for testing new algorithms, validating existing results and for reproducibility. All researchers within MO optimization are welcome to contribute.

The repository consists of a main repository MOrepo at GitHub and a set of sub-repositories, one for each contribution. Sub-repositories are named MOrepo-<name> where name normally is the surname of the first author and year of the study. All repositories are located within the MCDMSociety organization at GitHub.

The main repository contains documentation about how to use and contribute to MOrepo. Moreover, a set of tools are given in the R package MOrepoTools which can be used to retrieve info about test instance groups, results and problem classes.

Maintainers of MOrepo are Lars Relund Nielsen [email protected] and Sune Gadegaard [email protected].

Current maintainers of sub-repositories are Sune Lauth Gadegaard [email protected], Lars Relund [email protected], Thomas Stidsen [email protected] and Nathan Adelgren [email protected].

Current contributors to the repository are S.L. Gadegaard, A. Klose, L.R. Nielsen, C.R. Pedersen, K.A. Andersen, D. Tuyttens, J. Teghem, Ph. Fortemps, K. Van Nieuwenhuyze, M.P. Hansen, N. Adelgren and A. Gupte.

Usage

Instances can be downloaded in different ways depending on usage:

• If you want a whole sub-repository, download it as a zip file or clone it on GitHub.
• Browse to a single instance and download it using the raw format at GitHub.
• Use the R package MOrepoTools to download instances.

We recommend the last option and illustrate how it works. You don’t need much knowledge about R to use the package. But of course it is preferable. You need R and preferable RStudio installed on your computer. First you have to install the MOrepoTools package. From the R command line write:

library(devtools)  # if the package is missing see ?install.package 
install_github("MCDMSociety/MOrepo/misc/R/MOrepoTools")

To get an overview over the current problem classes run:

library(MOrepoTools)
getProblemClasses()  # current problem classes in MOrepo
## [1] "Facility Location"  "Assignment"         "Traveling Salesman"
## [4] "MILP"
getInstanceInfo(class = "Assignment")  # info about instances for the assignment problem
## 
## #### Contribution Pedersen08
## 
## Source:
## Warning: partial match of 'len' to 'lengths'
## Pedersen, C, L. Nielsen, and K. Andersen (2008). "The Bicriterion Multi
## Modal Assignment Problem: Introduction, Analysis, and Experimental
## Results". In: _Informs Journal on Computing_ 20.3, pp. 400-411. DOI:
## 10.1287/ijoc.1070.0253.
## 
## Test problem classes: Assignment  
## Subfolders: AP and MMAP  
## Formats: xml  
## 
## #### Contribution Tuyttens00
## 
## Source:
## Warning: partial match of 'len' to 'lengths'
## Tuyttens, D, J. Teghem, P. Fortemps, et al. (2000). "Performance of the
## MOSA Method for the Bicriteria Assignment Problem". In: _Journal of
## Heuristics_ 6.3, pp. 295-310. DOI: 10.1023/A:1009670112978.
## 
## Test problem classes: Assignment  
## Formats: raw and xml

Now download the Tuyttens00 contribution as a zip file using

getContributionAsZip("Tuyttens00")
## Download MOrepo-Tuyttens00.zip ... finished.

Or download selected instances

getInstance(name="Tuyttens.*n10")
## Download Tuyttens00_AP_n10.raw ...finished
## Download Tuyttens00_AP_n100.raw ...finished
## [1] "Tuyttens00_AP_n10.raw"  "Tuyttens00_AP_n100.raw"

How to contribute

All researchers are welcome to contribute to MOrepo. The repository mainly contains MO test instances and results from various sources. However, also generators, format converters, algorithms etc. related to MO optimization are welcome. Have a look at the documentation file contribute.md which describes different ways to do it.

Test instances @ MOrepo

Currently MOrepo contains instances for problem classes Facility Location, Assignment, Traveling Salesman and MILP. The contributions listed after class are:

Problem class - Facility Location

Contribution - Gadegaard16

Source:

## Warning: partial match of 'len' to 'lengths'

Gadegaard, S, A. Klose, and L. Nielsen (2016). “A bi-objective approach to discrete cost-bottleneck location problems”. In: Annals of Operations Research, pp. 1-23. DOI: 10.1007/s10479-016-2360-8.

Test problem classes: Facility Location
Subfolders: CFLP_UFLP and SSCFLP
Formats: raw

Problem class - Assignment

Contribution - Pedersen08

Source:

## Warning: partial match of 'len' to 'lengths'

Pedersen, C, L. Nielsen, and K. Andersen (2008). “The Bicriterion Multi Modal Assignment Problem: Introduction, Analysis, and Experimental Results”. In: Informs Journal on Computing 20.3, pp. 400-411. DOI: 10.1287/ijoc.1070.0253.

Test problem classes: Assignment
Subfolders: AP and MMAP
Formats: xml

Contribution - Tuyttens00

Source:

## Warning: partial match of 'len' to 'lengths'

Tuyttens, D, J. Teghem, P. Fortemps, et al. (2000). “Performance of the MOSA Method for the Bicriteria Assignment Problem”. In: Journal of Heuristics 6.3, pp. 295-310. DOI: 10.1023/A:1009670112978.

Test problem classes: Assignment
Formats: raw and xml

Problem class - Traveling Salesman

Contribution - Hansen00

Source:

## Warning: partial match of 'len' to 'lengths'

Hansen, M. (2000). “Use of Substitute Scalarizing Functions to Guide a Local Search Based Heuristic: The Case of moTSP”. In: Journal of Heuristics 6.3, pp. 419-431. DOI: 10.1023/A:1009690717521.

Test problem classes: Traveling Salesman
Formats: raw

Problem class - MILP

Contribution - Adelgren16

Source:

## Warning: partial match of 'len' to 'lengths'

Adelgren, N. and A. Gupte (2016). Branch-and-bound for biobjective mixed-integer programming. Optimization Online. Research rep. <URL: http://www.optimization-online.org/DB_HTML/2016/10/5676.html>.

Test problem classes: MILP
Subfolders: LP_1, LP_2, LP_3, LP_4, LP_5 and LP_6
Formats: lp

Results @ MOrepo

Currently MOrepo contains results for instances in problem classes Assignment. The contributions listed after class are:

Problem class - Assignment

Contribution - Pedersen08

Source:

## Warning: partial match of 'len' to 'lengths'

Pedersen, C, L. Nielsen, and K. Andersen (2008). “The Bicriterion Multi Modal Assignment Problem: Introduction, Analysis, and Experimental Results”. In: Informs Journal on Computing 20.3, pp. 400-411. DOI: 10.1287/ijoc.1070.0253.

Results given for contributions: Pedersen08 and Tuyttens00

How to cite

To cite use

@Electronic{MOrepo,
  Title                    = {Multi-Objective Optimization Repository (MOrepo)},
  Author                   = {L. R. Nielsen},
  Url                      = {https://github.com/MCDMSociety/MOrepo},
  Year                     = {2017}
}