Skip to content

[Completed] N580 delete-one recovery and full covariance; no further replay #575

Description

@LightChainr

Closed as completed — 2026-09-12

The original deterministic replay was delivered in #577, including retained delete-one shards. Merged #703 now reads those EXISTING 100 aligned shard rows and recovers the previously omitted cov(r2,r4) without simulation. The current complete covariance is stored in results/research-control-20260912/n580-complete-covariance-summary.json; the executed recovery and interpretation are in notes/n580-complete-covariance-20260912.md.

Recovered cov(r2,r4)=5.5621485097801135e-9, corr=0.11001108422637242. The retrospective three-rung bare-aspect pure-line score is D=10.8644599/2, nominal p=.00437333: not rejected at the declared nominal 3-sigma threshold. The original frozen two-rung underpowered verdict remains unchanged. #704 adds a separate, same-block rung-specific bounded-leakage analysis.

No more N580 replay is needed to recover this entry. Do not relaunch the historical three-machine command for this purpose. All original runs, seed/offset provenance, shards and frozen results remain preserved; completing this recovery adds no independent evidence and changes no production freeze.

#703 final head 909d4b3 passed full CI: 1626 tests with 3 existing skips, plus compile/C++ checks, before merge. This issue closes the recovery job, not the physical amplitude-law question. The remaining design channel is #589; within-model shape analysis is #622.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    priority:P1Bounded parallel analysis or a concrete reserve direction; not all run at once.

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions