Skip to content

Conversation

pierreablin
Copy link
Collaborator

This PR adds a 'll_reject' parameter. During the iterations, each sample with a negative log likelihood > ll_reject is going to be rejected. This helps finding artifacts.

@codecov-io
Copy link

Codecov Report

Merging #15 into master will decrease coverage by 1.16%.
The diff coverage is 85.45%.

@@            Coverage Diff             @@
##           master      #15      +/-   ##
==========================================
- Coverage   97.44%   96.27%   -1.17%     
==========================================
  Files           6        6              
  Lines         509      564      +55     
==========================================
+ Hits          496      543      +47     
- Misses         13       21       +8

return W


def fuse_mask(mask1, mask2):
Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is much much faster in numba. This is not really a bottleneck of the algorithm though.

for n in range(max_iter):
# Rejection
if rejection and (n + 1) % reject_every == 0:
ll = density.log_lik(Y)
Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm wondering how to make it work for ortho=True and/or extended=True

max_iter=1000, tol=1e-7, lambda_min=0.01, ls_tries=10,
verbose=False, covariance=None):
verbose=False, covariance=None, ll_reject=None,
reject_every=None):
Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Not sure how to set a default value for ll_reject if the user wants to use rejection

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants