The objective of this project is to investigate the performance of the MA-ES algorithm extended by the IPOP heuristics. The modified algorithm should be compared with the CMA-ES algorithm. For this purpose we use the BBOB benchmark from the coco framework.

The code was tested on Windows and Linux. We recommend using poetry for package versioning and installation. Coco must be built locally, refer to coco documentation for instructions.

1. Install dependencies with make install
2. Activate poetry shell with make shell
3. Run benchmarks with make run
4. Run post processing with make postprocessing
5. The results should be visible in your browser opened automatically.

• A. Auger, N. Hansen. “A restart CMA evolution strategy with increasing population size”. W: 2005 IEEE Congress on Evolutionary Computation. T. 2. 2005, 1769–1776 Vol. 2. doi: 10.1109/CEC.2005.1554902
• Hans-Georg Beyer, Bernhard Sendhoff. “Simplify Your Covariance Matrix Adaptation Evolution Strategy”. W: IEEE Transactions on Evolutionary Computation 21.5 (2017), p. 746–759. doi: 10.1109/TEVC.2017.2680320.