IMP-MARL dataset

<p>IMP-MARL dataset</p> <p>Authors: Pascal Leroy (pleroy@uliege.be), Pablo G. Morato (pgmdo@dtu.dk), Jonathan Pisane, Athanasios Kolios, Damien Ernst</p> <p>GitHub repository: <a href="https://github.com/moratodpg/imp_marl">https://github.com/moratodpg/imp_marl</a></p> <p>IMP-MARL environments:<br> &nbsp; &nbsp;struct_uc: k-out-of-n system<br> &nbsp; &nbsp; &nbsp;&nbsp;2-out-of-3, 4-out-of-5, 9-out-of-10, 48-out-of-50, 95-out-of-100<br> &nbsp; &nbsp; &nbsp;&nbsp;(with and without campaign cost)<br> &nbsp; &nbsp;struct_c: correlated k-out-of-n system<br> &nbsp; &nbsp; &nbsp;&nbsp;2-out-of-3, 4-out-of-5, 9-out-of-10, 48-out-of-50, 95-out-of-100<br> &nbsp; &nbsp; &nbsp;&nbsp;(with and without campaign cost)<br> &nbsp; &nbsp;owf: offshore wind farm<br> &nbsp; &nbsp; &nbsp;&nbsp;1 wind turbine, 2 wind turbines, 5 wind turbines, 25 wind turbines, 50 wind turbines<br> &nbsp; &nbsp; &nbsp;&nbsp;(with and without campaign cost)</p> <p>MARL methods investigated: QMIX, QVMIX, QPLEX, COMA, FACMAC, IQL, DQN.</p> <p>This dataset contains:<br> &nbsp; &nbsp;MARL logs.zip =&gt; The configuration, execution, and result files of all investigated MARL methods.<br> &nbsp; &nbsp; &nbsp;&nbsp;Each zip file contains the logs from an IMP-MARL environment.<br> &nbsp; &nbsp; &nbsp;&nbsp;There, test and train files are stored for each method:<br> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;cout.txt contains stored console ouputs.<br> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;config.json contains the list of all parameter values for the experiments.<br> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;info.json is the log file, listing reward values, for example.<br> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;run.json contains simply logged information about the host, typically hardware.<br> &nbsp; &nbsp;best_agent_networks.zip =&gt;The controller networks&rsquo; weights of the best MARL-based policies. &nbsp;<br> &nbsp; &nbsp; &nbsp;&nbsp;Controller files are stored as a pytorch file for each method/environment investigated (agent.th).<br> &nbsp; &nbsp;heur_logs.zip =&gt; The configuration, execution, and result files of the computed expert-based heuristic policies. &nbsp;<br> &nbsp; &nbsp; &nbsp;&nbsp;/heur_search =&gt; logs resulting from all evaluated heuristic policies, stored as numpy files.<br> &nbsp; &nbsp; &nbsp;&nbsp;/heur_search =&gt; results from the best performing heuristics over 10,000 policy realisations, stored as pickle files.</p>