EPSO

Technology

EPSO (evolutionary particle swarm optimization) is an evolutionary method with self-adaptive recombination via a particle movement equation inspired in PSO. The success of this strategy delivers to the scientific community a tool, more powerful and robust than other existing meta-heuristics, to solve engineering problems of high complexity degree (non-linear, mixed integer, non-continuous domain).

This population-based method, developed at INESC TEC, was first published in 2002 and was successively improved in new versions afterwards, up to the present. It has been successfully benchmarked in international competitions, consistently winning a first place with implementations proposed by a diversity of researchers. It is one of the few cases reported where a meta-heuristic has been embedded in the software of a control centre (Attica, Greece, distribution network).

EPSO code is available in C++ and Python.

More details on EPSO features can be found here.

Advantages

• Easy to use – less parameters and simpler to tune;
• Robust – provides consistent answers in repeated runs;
• Efficient – provides solutions when other methods fail, and is usually faster than alternatives, for the same quality of results.

Benefits

• Reduces preparation work/cost, because it does not require many trial-and-error attempts to fine-tune parameters;
• Provides reliability to users, because the variance in response is much smaller than competing algorithms;
• Easy to implement in parallel processing arrangements.

License

EPSO source code can be downloaded from its dedicated website under the terms of version 3 of the General Public License (GPLv3) as published by the Free Software Foundation. To obtain a commercial license or one without the limitations of the open source agreement, please contact SAL (info.sal@inesctec.pt), the INESC TEC Technology Licensing Office (TLO).