add determinsitic sites for extracting filtering from predictives.#72
Merged
10 commits merged intomainfrom Feb 7, 2026
Merged
Conversation
…y want to be careful about memory during MCMC though...
DanWaxman
requested changes
Feb 6, 2026
Collaborator
DanWaxman
left a comment
DanWaxman
left a comment
There was a problem hiding this comment.
As you mention, saving these things can be a big issue in some cases (long MCMC and/or really large state-space). Before, we had commented-out options to save/not save -- I think we should reinstate that (maybe also option to save only diagonal covar, etc.).
Comment on lines
+209
to
+212
| "tnj,tnk,tn->tjk", particles, particles, jnp.exp(log_weights) | ||
| ) | ||
| filtered_covariances = second_mom - jnp.einsum( | ||
| "tj,tk->tjk", filtered_means, filtered_means |
Collaborator
There was a problem hiding this comment.
... for when #60 lands? (it will eventually...)
Collaborator
Author
There was a problem hiding this comment.
also not sure how to know where to put ... in anticipation of batching
Collaborator
|
Two remaining things: documentation of what's happening, and also some tests. |
DanWaxman
reviewed
Feb 6, 2026
Comment on lines
+246
to
+251
| second_mom = jnp.einsum( | ||
| "tnj,tnk,tn->tjk", particles, particles, jnp.exp(log_weights) | ||
| ) | ||
| filtered_covariances = second_mom - jnp.einsum( | ||
| "tj,tk->tjk", filtered_means, filtered_means | ||
| ) |
Collaborator
There was a problem hiding this comment.
I think this should work
Suggested change
| second_mom = jnp.einsum( | |
| "tnj,tnk,tn->tjk", particles, particles, jnp.exp(log_weights) | |
| ) | |
| filtered_covariances = second_mom - jnp.einsum( | |
| "tj,tk->tjk", filtered_means, filtered_means | |
| ) | |
| second_mom = jnp.einsum( | |
| "...tnj,...tnk,...tn->...tjk", particles, particles, jnp.exp(log_weights) | |
| ) | |
| filtered_covariances = second_mom - jnp.einsum( | |
| "...tj,...tk->...tjk", filtered_means, filtered_means | |
| ) |
Collaborator
Author
There was a problem hiding this comment.
perfect, done, thanks
mattlevine22
commented
Feb 7, 2026
Collaborator
Author
mattlevine22
left a comment
mattlevine22
left a comment
There was a problem hiding this comment.
should be ready
DanWaxman
requested changes
Feb 7, 2026
Comment on lines
+273
to
+284
| add_mean = record_kwargs.get( | ||
| "record_filtered_states_mean", False | ||
| ) and _should_add_site(states.mean.shape, max_elems) | ||
| add_chol_cov = record_kwargs.get( | ||
| "record_filtered_states_chol_cov", False | ||
| ) and _should_add_site(states.chol_cov.shape, max_elems) | ||
| add_filtered_states_cov = record_kwargs.get( | ||
| "record_filtered_states_cov", False | ||
| ) and _should_add_site((T1, state_dim, state_dim), max_elems) | ||
| add_filtered_states_cov_diag = record_kwargs.get( | ||
| "record_filtered_states_cov_diag", False | ||
| ) and _should_add_site((T1, state_dim), max_elems) |
Collaborator
There was a problem hiding this comment.
Why and instead of or? Seems weird I can't override our (helpful, yet arbitrary) _should_add_site suggestion.
Collaborator
Author
There was a problem hiding this comment.
I see, yeah. I guess then I should default these things to True and switch to or?
Merged
Collaborator
|
I don’t think defaulting to true does anything, right? Still can’t override
a _should_add = False. I actually think the nicest solution is to set
_should_add to be the default, then no need for conjunction.
…On Fri, Feb 6, 2026 at 8:50 PM Matt Levine ***@***.***> wrote:
***@***.**** commented on this pull request.
------------------------------
In dynestyx/inference/cuthbert/discrete_time_filters.py
<#72 (comment)>
:
> + add_mean = record_kwargs.get(
+ "record_filtered_states_mean", False
+ ) and _should_add_site(states.mean.shape, max_elems)
+ add_chol_cov = record_kwargs.get(
+ "record_filtered_states_chol_cov", False
+ ) and _should_add_site(states.chol_cov.shape, max_elems)
+ add_filtered_states_cov = record_kwargs.get(
+ "record_filtered_states_cov", False
+ ) and _should_add_site((T1, state_dim, state_dim), max_elems)
+ add_filtered_states_cov_diag = record_kwargs.get(
+ "record_filtered_states_cov_diag", False
+ ) and _should_add_site((T1, state_dim), max_elems)
I see, yeah. I guess then I should default these things to True and switch
to or?
—
Reply to this email directly, view it on GitHub
<#72 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AC2DPONLRJNNE2S6QDMRQRD4KVAETAVCNFSM6AAAAACUFBT4ACVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZTONRVHA4DANJTGM>
.
You are receiving this because your review was requested.Message ID:
***@***.***>
|
Collaborator
Author
|
Yeah I realized that after I merged the stuff, sorry. Fixing it now. |
mattlevine22
added a commit
that referenced
this pull request
Feb 7, 2026
Collaborator
Author
|
Addressing in #81 |
DanWaxman
pushed a commit
that referenced
this pull request
Feb 7, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
{name}_filtered_states_mean,{name}_filtered_states_cov, and{name}_marginal_loglik