-
Notifications
You must be signed in to change notification settings - Fork 528
[MRG] Free support Sinkhorn barycenters #387
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
Codecov Report
@@ Coverage Diff @@
## master #387 +/- ##
==========================================
- Coverage 93.80% 93.76% -0.04%
==========================================
Files 23 23
Lines 5936 5967 +31
==========================================
+ Hits 5568 5595 +27
- Misses 368 372 +4 |
you should use autopep8 to check and correct pep8 mistakes |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hello @eddardd and thank you for the changes. There remain a few things that should be done before merging but it is looking good.
The barycenter of the duck is weird because the regularization is too strong and that one of the problem of entropic barycenter when the two supports are too far, we should keep it that way. I like your example on the 4 distributions so also keep it.
Hello @eddardd this is great we are ready for merge IMHO, do you want to add your name website ad contribution to the contributors.md file ?I can wait for you to commit that. |
Hi @rflamary , thanks for your help in adding this new feature to POT. I have been using the library for quite some time, and it feels good to contribute to it! I plan to contributing with a few features (mainly in DA) in the days to come. I added my info on CONTRIBUTORS.md :) |
Types of changes
This PR includes
ot.bregman
, calledfree_support_sinkhorn_barycenter
, that calculates Wasserstein barycenters using the Sinkhorn solver./examples/barycenters
calledplot_free_support_sinkhorn_barycenter.py
, illustrating a possible use case of Free Support Sinkhorn BarycentersMotivation and context / Related issue
The Free Support Wasserstein barycenter already exists in the library (
ot.lp.free_support_barycenter
). In some applications it may be useful to solve the barycenter problem with entropic regularization.How has this been tested (if it applies)
A small script has been added to
./examples/barycenters
, calledplot_free_support_sinkhorn_barycenter.py
, illustrating the calculation of the Free Support Sinkhorn Barycenter in a small scale use case.PR checklist