v0.2.1

Fix a few bugs with inbuilt functions that make the SOM appear to not function properly

What's Changed

• improve plot_mU_matrix function by @LuisSanchez25 in #30
• Change c som order by @LuisSanchez25 in #31
• Bump version to 0.2.0 by @LuisSanchez25 in #32
• remove strax/straxen dependancies by @LuisSanchez25 in #33
• add optional hist, needs to be incorporated in kSOM as well by @LuisSanchez25 in #34
• added if histories true to save in nf by @LuisSanchez25 in #35
• fix track bug by @LuisSanchez25 in #36
• fix bug by @LuisSanchez25 in #37

Full Changelog: v0.2.0...v0.2.1

v0.2.0

What's Changed

• bump version to v0.1.0 by @LuisSanchez25 in #18
• make readthedocs with the tutorial notebook by @LuisSanchez25 in #19
• add nbsphinx dependency by @LuisSanchez25 in #20
• Fix typo in index.rst file by @LuisSanchez25 in #21
• add dependencies by @LuisSanchez25 in #22
• fix typo by @LuisSanchez25 in #23
• add numba dependency by @LuisSanchez25 in #24
• fix order of notebook by @LuisSanchez25 in #25
• include more diagnosis tools + cSOM example by @LuisSanchez25 in #26
• Make example for cSOM usage by @LuisSanchez25 in #28
• update example in notebook by @LuisSanchez25 in #29

Full Changelog: v0.1.0...v0.2.0

fixed SOM asymmetrical convergence!

While the SOM was converging before, the distribution of neurons in the data space always resulted in a distribution that did not match the density distribution of the data, leaving some cluster with only 1 or 0 overlapping neurons. This was found to be a bug in how we computed the neighborhoods which is now fixed!

v0.0.0-beta

This will be the first stable version of this repository, maybe functions are still not fully working but it should be enough for people to try training a kohonen SOM, even this is not fully working as expected, but, it seems to be able to construct clusters so I will check back on this. Things that we will need to have working properly before getting out of beta:

• The conscious-SOM which will be the first implementation of this in python! (as far as I am aware)
• More thorough testing: Need to build up more tests for all functions
• Many depreciated functions: Many functions in this code are designed to work with an old system and need to be made more general
• Better documentation: there is some basic documentation with https://scisom.readthedocs.io/en/latest/ but there are still some issues to polish.

With all this in mind, users are welcome to try this package and report on any issues!

What's Changed

• restructure some files by @LuisSanchez25 in #1
• Improve docstrings by @LuisSanchez25 in #2
• Test kSOM learning function by @LuisSanchez25 in #3
• reorganize files and switch to toml by @LuisSanchez25 in #4
• functioning kSOM by @LuisSanchez25 in #5
• Add tests for some basics by @LuisSanchez25 in #6
• Tests v2 by @LuisSanchez25 in #7
• insert more debugging tools into the SOM training by @LuisSanchez25 in #8
• update notebook by @LuisSanchez25 in #9
• Start documentation page by @LuisSanchez25 in #10
• rebuild docs after changing conf file by @LuisSanchez25 in #11
• redo build by @LuisSanchez25 in #12
• make a functioning html site by @LuisSanchez25 in #13
• remove _build files so readthedocs plays nice by @LuisSanchez25 in #16

New Contributors

• @LuisSanchez25 made their first contribution in #1

Full Changelog: https://github.com/RiceAstroparticleLab/SOM_package/commits/v0.0.0-beta