Repository navigation
Spec 017: the standing time-late control, and is the shifted-prior gap identifiability? (in progress) - #34
Merged
Merged
Conversation
…classical control stands; is the shifted-prior gap identifiability?) Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019erPajyAwQ6mBQ2Xz4xNHs
… the detectable ones, the search under the true prior, IMM and EKF beside; plan, research, tasks; draft plan and tuning run file Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019erPajyAwQ6mBQ2Xz4xNHs
…aders (the weights' digests study L's record lacks); export check on the three trained checkpoints (agrees to 8e-5) Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019erPajyAwQ6mBQ2Xz4xNHs
…cord guard Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019erPajyAwQ6mBQ2Xz4xNHs
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019erPajyAwQ6mBQ2Xz4xNHs
…vered from its shard headers (all 241 shards, the three training checkpoints) Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019erPajyAwQ6mBQ2Xz4xNHs
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019erPajyAwQ6mBQ2Xz4xNHs
…s estimands and forecasts without a decision rule; SRD section 9 lines (proposed) Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019erPajyAwQ6mBQ2Xz4xNHs
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019erPajyAwQ6mBQ2Xz4xNHs
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019erPajyAwQ6mBQ2Xz4xNHs
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.
Summary
Owner's decision 2026-11-14 (option A after study L): the classical control becomes the standing time-late system, and one classical study asks whether the gap left under a shifted manoeuvre prior can be closed by anyone. Nothing trained; CPU only. In progress: the tuning run is on CI; registration, validation and the test follow.
So far:
specs/017-shifted-gap/(spec, plan, research, tasks) and the kickoff post.timelate/detect.py: a true manoeuvre's detectability (the bearings' cost its omission adds, times held) and references told only the detectable manoeuvres.timelate/run5.py(plan keyrunner: run5): study L's control unchanged, those references, the search under the split's own prior, the tuned IMM and EKF read as tracks, references on edited bearings.timelate/study5.py: the gap split into an attainable share and identifiability, with the rule.experiments/20261114-tlm-runs/export-check.json), a workflow that recovers a finished run's shard headers (headers/**), tests for study L's split readings and the controls-record guard.Still to come in this PR: tuning record, SRD §9 line (the owner merges an SRD change), decision record, validation (seed 343), test (seed 344, once), panel, write-up and slides.
Test plan
uv run ruff check .,ruff format --check .,uv run mypypytest tests/unit/test_timelate5.py tests/unit/test_timelate4.pyGenerated by Claude Code