Skip to content

graphical-ep: reconcile EP campaign state 2026-10-23 - #38

Merged
Jammy2211 merged 5 commits into
mainfrom
ep-campaign-reconcile-2026-10-23
Oct 23, 2026
Merged

Jammy2211 merged 5 commits into
mainfrom
ep-campaign-reconcile-2026-10-23

Conversation

@Jammy2211

Copy link
Copy Markdown
Contributor

Summary

Reconciles the graphical/EP campaign against what has shipped and the current Cortex ledgers (PyAutoCortex origin/main 5fa4eb98).

  • Shipped since the 2026-09-10 table (all released in PyAutoFit 2026.10.7.1): #1632/#1634 (EMFILE), #1643/#1644/#1645 (factor-search overhead 2.45 s → ~0.95 s), #1655 (ProjectionException recovery, the IC50 N=50 cure), #1656 (LaplaceOptimiser(projection="moments"), the hierarchical-scatter cure).
  • Cortex Now states for analytic_gaussian, ep_toy_gaussian, slope_hierarchy_scale and ic50_workspace are recorded per phase with each ledger's next step. The shared RAL mirror 404b3e5f7 stays frozen for euclid_dr1 (it lacks #1655/#1656), so the reruns need a private PyAutoFit checkout.
  • New dev phase PyAutoFit#1711 (zero-iteration Laplace convergence counted as stale), plus a "Wave-1 findings follow-ups (2026-10-23)" table covering the three ruling findings.
  • campaigns.yaml: graphical-ep → active. The four science rows get updated next, recent_progress, blockers and evidence (pinned to Cortex 5fa4eb98). Their statuses are unchanged and job_status is still unknown because the live scheduler was not queried. The ep_campaign task row points at the addendum.

Note: why there is an addendum file instead of an edit to tasks/ep_campaign.md

tasks/ep_campaign.md is a byte-preserved migration original. migration.yaml pins its SHA256, and pyauto-insight check fails if the file changes. The 2026-10-23 state therefore lives in tasks/ep_campaign_state_2026-10-23.md, which adds to the original State cells without replacing them. If you would rather edit the original in place, the migration hash has to be re-pinned. That is a provenance decision for you to make.

Validation

  • ruff check . and ruff format --check . pass; pytest tests -q: 204 passed.
  • bin/pyauto-insight check --offline: campaigns, registry, fit, dashboard and state are ok. lens FAIL comes from unresolved prepared-* / stage problem records in the lens snapshot. The same failure happens on main without this change.
  • Dashboard re-rendered with board --offline.

🤖 Generated with Claude Code

Jammy2211 and others added 2 commits October 23, 2026 15:00
Record the PyAutoFit fixes shipped since the 2026-09-10 phase table
(#1632/#1634, #1643-#1645, #1655, #1656; all in 2026.10.7.1), the four
Cortex "Now" states (origin/main 5fa4eb98), the new dev phase
PyAutoFit#1711 and the wave-1 findings follow-ups.

tasks/ep_campaign.md is a byte-preserved migration original (SHA256
pinned in migration.yaml, enforced by pyauto-insight check), so the
dated addendum lives in tasks/ep_campaign_state_2026-10-23.md rather
than in the original State column.

campaigns.yaml: graphical-ep -> active; analytic-gaussian,
ep-toy-gaussian, slope-hierarchy-scale and ic50-scale get updated
next/recent_progress/blockers and evidence pinned to Cortex 5fa4eb98
(statuses unchanged, job_status left unknown); ep_campaign task row
points at the addendum. Dashboard re-rendered offline.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…ckout

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Jammy2211 and others added 3 commits October 23, 2026 15:37
…shboard

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…shboard

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
@Jammy2211
Jammy2211 merged commit 728799e into main Oct 23, 2026
2 checks passed
@Jammy2211
Jammy2211 deleted the ep-campaign-reconcile-2026-10-23 branch October 23, 2026 14:44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant