This is a python implementation of asynchronous evolutionary algorithm. The main algorithm is implemented in asynch_ea.py It is heavily based on DEAP and scoop.
To test it there are two scripts: test_asynch_ea.py and test_nested_asynch_ea.py
At the moment one experiment is implemented using the algorithm: modular_2d_walker.py This experiment evolve 2d creatures to solve a walking task. The algorithm is a nested optimisation process with the asynch_ea to evolve the shape of the creatures and a simple genetic algorithm to optimise their controllers.
To run it you have to clone gym_rem2D repository. And copy or create a symbolic link of the folder ModularER_2D in the task folder. To create a symbolic link:
cd tasks
ln -s path/to/gym_rem2D/ModularER_2D