fix(estimation): replay estimation options in data-changing refits - #1524
Draft
s3alfisc wants to merge 16 commits into
Draft
fix(estimation): replay estimation options in data-changing refits#1524s3alfisc wants to merge 16 commits into
s3alfisc wants to merge 16 commits into
Conversation
…mation-refits Resolves the architecture conflict by placing this layer's refit sentences in the post-estimation paragraph under the implemented domains.
…port Comment-only: states why a prebuilt LSMR factorization is rebuilt for each changed row set and why the refit imports the estimation package lazily.
This was referenced Sep 5, 2026
…mation-refits # Conflicts: # docs/changelog.qmd # docs/developer/architecture.md
…mation-refits Adapt the refit layer to the hook-free base: the architecture note keeps only this layer's refit-replay paragraph, because the base review removed the `fixef()` and `IV_Diag()` sentence it was appended to. No code adaptation was needed; the replayed options keep their attribute names.
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Poisson CRV3 jackknife refits and slow randomization-inference refits now replay the original estimation options: weights, small-sample corrections, singleton handling, solver, demeaner, offset, and IRLS tolerances. A prebuilt LSMR preconditioner is the one deliberate exception: its factorization belongs to one fixed-effect design, so a refit keeps the variant but rebuilds it for the changed row set.
Review question: Does a refit replay the original specification?
Layer 8 of 9 replacing #1500/#1501; depends on #1523 (
fix/retained-estimation-storage). The original PRs stay open until the replacement stack is validated.Verification
tests/test_ses.pyandtests/test_ritest.py -k "replay or refit or estimation_contract or preconditioner"3 passed;tests/test_estimator_state_lifecycle.py55 passed.ruff-check/ruff-formathooks passed at this tip.Exact targeted command:
pixi run -e py312-r pytest tests/test_ses.py tests/test_ritest.py -q --no-cov -k "replay or refit or estimation_contract or preconditioner".The repository's Python/R/docs workflow only runs for PRs targeting
master; this stacked PR receives pre-commit CI only. Broad cumulative checks ran locally on the final replacement head (layer 9). Human review is required; no automatic merge.