intelligent-control-lab / CMA-ES

CMA-ES

A CMA-ES implementation.

To use CMA-ES, define an evaluator in evaluator.py following test_evaluator, and create a learning config file following test_config.yaml. The following command runs CMA-ES to minimize a quadratic equation.

python cma-es.py test_config.yaml

We also give another example: tuning parameters of a PD controller by CMA-ES.

python cma-es.py pd_config.yaml

Tried parameters and rewards are stored in logs.