Repository navigation
Directed information per edge, and directed-information baselines - #3
Merged
Merged
Conversation
…e) and di = te + iie; model_di in exact flows Co-authored-by: Cursor <cursoragent@cursor.com>
…CSE) and benchmark; validation table Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Autoplectic
added a commit
that referenced
this pull request
Oct 14, 2026
Directed information per edge and DI baselines (re-land #3 on main); lock sofic with multi-trial CSSR
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Stacked on #2 (
machines-layer).Summary
Directed information per edge (
infer_multiplex(directed_information=True))iie: the instantaneous information exchange of each pair, I[X_t : Y_t | both selected pasts and lagged parents], with a permutation p-valueiie_pvalue. It is computed by the newinfoflow.contemporaneous.instantaneous_exchange, which the lag-0 layer now reuses.di = weight[te] + iie: the bivariate decomposition of directed information into transfer entropy plus instantaneous exchange (Massey 1990; Amblard & Michel 2013). The docs note that with side information it is only approximate.model_di: the composed model's exact one-step directed information, added to the machine layer's exact flows.Baselines and benchmark (
validation/di_baselines.py,validation/di_benchmark.py,validation/run_di.sh)docs/validation.rst. Highlights:Tests
tests/test_directed_information.py:iie> 0 withdi = te + iie;diis aboutte;model_diequalsmodel_tefor a lagged channel.tyand ruff are clean, and the docs build without warnings.Made with Cursor