Log overall global hypervolume of whole archive for MOME - Githubissues

schrum2 / MM-NEAT

Modular Multiobjective (Hyper) Neuro-Evolution of Augmenting Topologies + MAP-Elites: Java code for evolving intelligent agents in Ms. Pac-Man, Tetris, and more, as well as code for Procedural Content Generation in Mario, Zelda, Minecraft, and more!

http://people.southwestern.edu/~schrum2/re/mm-neat.php

Other

50 stars 20 forks source link

Log overall global hypervolume of whole archive for MOME #928

Closed schrum2 closed 1 year ago

schrum2 commented 1 year ago

Description is pretty clear

JoannaBlatt commented 1 year ago

Logs total hypervolume in archive log with other hypervolume logging

JoannaBlatt commented 1 year ago

I think we should move the total hypervolume to it's own log, since it is scaled in such a way that it obscures the other hypervolume information.

JoannaBlatt commented 1 year ago

Synchronization issue between logging and add method was found. Currently fixed by having the whole archive locked for both of them.

JoannaBlatt commented 1 year ago

I think the mean hypervolume is still having issues because I assume it includes a large number of empty bins. Maybe this needs to be filtered?

schrum2 commented 1 year ago

I think we need to have all of the following on one plot:

global hypervolume (should be largest)
max bin hypervolume (second largest or tied with global)
average occupied bin hypervolume (only divide by num occupied bins)
min occupied bin hypervolume (only min over occupied bins)
average overall hypervolume (includes 0 entries for empty bins)

JoannaBlatt commented 1 year ago

finished logging all hypervolume data that we have thought about all logging seems to function and looks appropriate so far