Which open source toolkits are available for solving multi-objective optimization problems? Which supports all multi-objective algorithms (NSGA, NSGA-II, SPEA, PEAS)? Which one is best to use to solve these problems?
There are many, many different MOEAs to choose from. However, it is encouraged to pick a second generation one (like NSGA-II and the like) due to their properties. For instance, here you can find a MOEA for multiobjective clustering http://grsi.salle.url.edu/alvarog/software/ that makes use of PESA-II. In the reference http://www.sciencedirect.com/science/article/pii/S0951832005002012 you can find a nice tutorial with the state-of-the-art (it's a bit old IMHO, but worth the time) of MOEAs.
I used NSGA II in Matlab from http://www.mathworks.com/matlabcentral/fileexchange/31166-ngpm-a-nsga-ii-program-in-matlab-v1-4. It also have graphical representation.
HyFlex (http://www.hyflex.org, free) is used to generate solutions for difficult problems. It was developed for Hyper-heuristic but I believe can be used in Meta-heuristics too.
I have used in my PhD ECJ (http://cs.gmu.edu/~eclab/projects/ecj/, free), which it is a research Evolutionary Computation (EC) system written in Java. It is limited to the type of Evolutionary algorithms but powerful.
One code which is free on the web is WWW-NIMBUS, written by Kaisa Miettinen. We will actually utilize it in a few weeks in a PhD course that she will give at our department. After that I can provide an assessment of it. :-)
There is a nice software tool for multi-criteria optimization that uses exhaustive iterative search, ideal for validating meta-heuristic optimization algorithms:
Please consider TridentOpt for MultiObjective Optimization. It uses a Hybrid Genetic Algorithm Multi-Agent System to solve problems using fuzzy programming methods. It can be found at
Many thanks for the question and replies, which are all useful. We have the following MS Excel based programs for MOO free of cost. These were developed by my project students as part of our extensive research on MOO and its applications to Chemical Engineering; browse the link: http://cheed.nus.edu.sg/stf/chegpr. These MS Excel programs have been used by many others. Interested readers can contact me through email to [email protected].
EMOO based on NSGA-II, includes both binary and real coding, and two progress-based termination criteria besides maximum number of generations
IMODE based on adaptive differential evolution includes two progress-based termination criteria besides maximum number of generations
NDS for sorting and finding non-dominated solutions from the list of given solutions
Ranking/Selection Methods includes 10 different techniques (such as TOPSIS, SAW and GRA) for choosing one of the Pareto-optimal solutions
The combination of optimization algorithms included in the Microsoft Excel Solver add-in with Derringer's Desirability function is an alternative to solve multi-response optimization problems without programming. The determination of Hansen's solubility parameters is a good example of this. https://www.hansen-solubility.com/HSP-science/basics.php