Normalization of the archive

[naceee]

Added the MOArchiveBase class with normalization implemented as discussed in issue #16.

[naceee]

some comments:

• I am not sure about the version, I set it to 1.1.0, maybe it could also be 1.0.1?
• I renamed folder readme_files to README_files, as the png plots are otherwise not displayed in the README. It is possible to edit the README to use plots from readme_files folder but it has to be done manually every time after creating readme from the moarchiving notebook. Maybe there is also some other solution?
• Unittests for new part of the code are in test_normalization.py
• We didn't talk about this in the issue, but with @ttusar we decided to also normalize the constraints in the case of constrained moarchive (user can provide max value for each constraint, then they are normalized and summed together for the total_cv which is used in hypervolume_plus_constr indicator)

[nikohansen]

Thanks! Could you streamline/squash your commits to avoid commits like started work on MOArchiveBase class, the tests pass i guess, more tests for normalization, small fix after rebasing, etc? Work in progress (when it has never been published) doesn't need to be in the (final) commit history.

[nikohansen]

Re. the pngs: I don't think this works for https://pypi.org/project/moarchiving. For this reason, I was executing a script which truncates the README.md accordingly as documented at https://github.com/CMA-ES/moarchiving/blob/master/howto/release.md, namely

python -c "open('README.md', 'w').write(open('../README.md', 'r').read().split('### 11')[0])"

I forgot to commit the truncated readme file in the current repository.

[naceee]

Could you streamline/squash your commits

Makes sense. It looks like I managed to fix it.

Re. the pngs: I don't think this works for https://pypi.org/project/moarchiving. For this reason, I was executing a script which truncates the README.md

Ah I see, okay, then I just renamed it back to what it was before.

[nikohansen]

Did you commit the .idea folder accidentally?

[naceee]

Yes i think that was by accident

[nikohansen]

Sorry, I'd like to not have the .idea folder first added and then removed.

[nikohansen]

Do we really want to overwrite self.hypervolume_final_float_type with a default when it was already set? This could happen depending on where and when super is called.

[naceee]

Sorry, I'd like to not have the .idea folder first added and then removed.

Makes sense yeah. Btw, wouldn't squashing commits when accepting PR fix this?

Also I would like to encourage you to fix the code yourself (or maybe @ttusar has someone to do it?). I am not really working in this field anymore and am a bit busy with other stuff in my life at the moment.

[nikohansen]

Makes sense yeah. Btw, wouldn't squashing commits when accepting PR fix this?

Sure. The minimal change would be to move the folder-remove-commit right behind the commit where the folder was added and then squash/fixup the mistaken commit with the moved commit. Moving commits and fixup can be done with interactive rebase.

[nikohansen]

Also I would like to encourage you to fix the code yourself (or maybe @ttusar has someone to do it?). I am not really working in this field anymore and am a bit busy with other stuff in my life at the moment.

Thanks for letting us know. FYI, we also found some bugs in the current code as described in the new issues.