jMetal project documentationΒΆ

Author: Antonio J. Nebro <ajnebro@uma.es>

jMetal is an open source Java-based framework for multi-objective optimization with metaheuristics. It includes a wide set of resources, including state-of-the-art multi-objective algorithms, solution encodings, benchmark problems, quality indicators, and utilities for performing experimental studies.

The current stable version is 6.0 (https://github.com/jMetal/jMetal). The working version in GitHub is 6.1-SNAPSHOT.

jMetal is described in the following papers:

Summary of features:

  • Multi-objective algorithms: NSGA-II, SPEA2, PAES, PESA-II, OMOPSO, MOCell, AbYSS, MOEA/D, GDE3, IBEA, SMPSO, SMPSOhv, SMS-EMOA, MOEA/D-STM, MOEA/D-DE, MOCHC, MOMBI, MOMBI-II, NSGA-III, WASF-GA, GWASF-GA, R-NSGA-II, CDG-MOEA, ESPEA, SMSPO/RP, AGEMOEA, CDG, FAME, MicroFAME, MOSA.
  • Single-objective algorithms: genetic algorithm (variants: generational, steady-state), evolution strategy (variants: elitist or mu+lambda, non-elitist or mu, lambda), DE, CMA-ES, PSO (Stantard 2007, Standard 2011), Coral reef optimization.
  • Parallel models: Synchronous (multi-threaded, Apache Spark), asynchronous
  • Variable representations (encodings): binary, real, integer, permutation, mixed
  • Problems:
    • Problem families: ZDT, DTLZ, WFG, RE, CRE, FDA, CEC2009, LZ09, GLT, MOP, LIRCMOP, MOP, UF
    • Classical problems: Kursawe, Fonseca, Schaffer, Viennet2, Viennet3
    • Constrained problems: Srinivas, Tanaka, Osyczka2, Constr_Ex, Golinski, Water, Viennet4
    • Combinatorial problems: multi-objective TSP
    • Academic problems: OneMax, OneZeroMax
  • Quality indicators: hypervolume, normalized hypervolume, spread, generational distance, inverted generational distance, inverted generational distance plus, additive epsilon.
  • Support for experimental studies
  • Support for automatic algorithm configuration and design