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
Fix equilibration detection #610
Conversation
I've tested these changes locally by checking that:
|
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.
This is great! Thanks!
@mikemhenry @ijpulidos : I just pushed changes to address Mike and John's feedback. Can you re-review and merge? |
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.
This looks good to me. We should try to come up with some sort of test that can help us validate the logic. I know some of these tests can be hard (it probably requires a trajectory if we want to test the MultiStateAnalyzer
, for example) but maybe some others don't need a full trajectory. We had a conversation with @zhang-ivy on the possibility of sharing some of her notebooks and then, hopefully I can come up with some sort of test based on them.
I think we have some tests that run the multistate sampler that run on the GPU. Perhaps we can just add an analysis step that checks to make sure that |
I just pushed a commit that adds an analysis step checking to make sure n_equilibration_iterations is > 1 in |
I just pushed changes that also check:
|
@mikemhenry @ijpulidos : Can you review again and merge this? |
Looking at it now! |
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.
This looks good to me, thanks for going through this. Just a minor non-blocking suggestion.
Description
Changes:
statistical_inefficiency
withoutn_equilibration_iterations
n_equilibration_iterations
is specified andstatistical_inefficiency
is not, make sure that the finaln_equilibration_iterations
includes the t0 selected from the time origin analysis (akai_t[i_max]
here)_get_equilibration_data_per_sample()
max_subset
when creating the aMultiStateSamplerAnalyzer
For more context on why these changes are necessary, see #609 (comment)
Fixes #609
Todos
Status