Supplementary Codes for "Parallelized Bayesian Optimization for Expensive Robot Controller Evolution"
The experiments are based on the linux task manager SLURM. The files should be run in the order:
- 00RunExperiments.R - In order to run the experiments. After submitting them to slurm wait for task completion. You might need to adapt 'runSingleCore.slurm' to your local cluster configuration.
- 01DataAnalysis.R - In order to generate the plots as given in the papers' supplementary material.
- 02cleanAll.R - In order to remove all output files, for a fresh restart of the experiments.
To run the robot experiments first pull the docker image using the command docker pull cigroup/revolve:master and modify it with the given Dockerfile