Lambda Days – Daniel Krzywicki – The continuous evolution of asynchronous agents


By Erlang Central | Published: April 29, 2014



Evolutionary algorithms are a class of meta-heuristics used to solve hard optimization problems. The idea of evolutionary multi-agent systems (EMAS) introduces algorithms more similar to biological evolution than classical evolutionary methods. However, the possibilities of the existing implementations have been limited by the underlying technology, which is usually Java.

In this talk, I am going to show how actors and functional patterns can be used to design fully autonomous and asynchronous agents. We will see that such multi-agent systems simulate biological processes even more closely, as changes in the population are continuous. On the occasion, we will also compare the performance of similar Scala and Erlang implementations.