Fix(ict,#20321): ICT-19 -- Gate ENJEU-1 re-exprime sur repair_gain (espace de champ) by jsboige · Pull Request #20334 · jsboige/CoursIA

jsboige · 2026-10-11T06:07:30Z

Fix(ict,#20321): ICT-19 -- Gate ENJEU-1 re-exprime sur repair_gain (espace de champ)

Grain: DEEP/notebook-python — lane myia-po-2025:CoursIA — prev: DEEP/research-code #20245

Fille case 4 de #11690 (point 9 de la seance du 10/10). L'increment etait nomine par ICT-19b section 13 : « la reconciliation vers l'instrument de champ pour cette ligne du gate principal est un incrément de consolidation ulterieur ».

What

Le Gate ENJEU-1 du notebook principal etait exprime (et echouait) sur l'instrument scalaire stake_index applique a S4 scalarise en structure(V) -- mauvais instrument pour un champ 2D, avec en prime une comparaison S4-vs-S5 non appariee en I_thermo (+5.62 vs 0). ICT-19b etablit la cause racine (instrument + regime immature) et mesure l'instrument de champ : ict.agency.repair_gain.

Cette PR porte la reconciliation dans ICT-19 lui-meme :

Mesure de champ fraiche integree au carnet (cellule verdict) : protocole ICT-19b section 8.1 -- regime de Pearson (F=0.0367, k=0.0649), grille 64x64, warmup 3000 pas, 5 ablations aleatoires, relaxation 1500 pas, temoin diffusion pure (meme operateur, terme reactif retire, apparie en I_thermo). Mesure : repair_gain = +0.8217 +/- 0.2722, recovery Gray-Scott +1.0013 vs diffusion +0.0018.
Enonce du gate re-exprime (principal) : repair_gain(S4) > 0 contre le temoin diffusion pure apparié en I_thermo -- verdict OK.
FAIL scalaire conserve comme temoin negatif : le verdict affiche les DEUX lignes ; le FAIL du banc naive (I_stake(S4) = -0.5083 < I_stake(S5) = -0.4651, delta I_thermo +5.62) reste visible, requalifie en discordance d'instrument (renvoi ICT-19b section 13) -- le temoin negatif n'est pas efface.
Lecture 3 et Lecture 6 reecrites : la Lecture 3 renvoie a la cause racine etablie par 19b ; la Lecture 6 lit le verdict dual (champ OK / scalaire FAIL explique). La cause du nul n'est plus attribuee a la fenetre de relaxation courte (ancienne hypothese du carnet, refutee par 19b).
Conclusion corrigee : la phrase « la paire (I_thermo, I_stake) separe enfin l'agent du pur dissipateur » promettait ce que le verdict scalaire refutait -- elle dit desormais que la separation tient en espace de champ (repair_gain apparie), le banc scalaire restant un contre-exemple documente.

Validation

Re-execution INTEGRALE via papermill (kernel python3) apres modification : 11 cellules code, 0 erreur, execution_counts continus.
Recoupement prose/sorties : les valeurs citees dans les Lectures 3/6 et la Conclusion proviennent des sorties fraiches committes.
Aucune erreur volontaire ajoutee (C.1 intact : grep 0) ; seules les cellules 16, 20, 23, 25, 32 (markdown) et 24 (code) ont change.

Closes #20321
See #11690, #20317

🤖 Generated with Claude Code

…space de champ) Mesure de champ integree au carnet (protocole ICT-19b section 8.1, regime Pearson, 5 ablations) : repair_gain(S4) = +0.8217 +/- 0.2722 contre temoin diffusion pure +0.0018 -> gate champ OK. Le FAIL scalaire du banc naive est conserve comme temoin negatif de discordance d'instrument (19b section 13). Lecture 3/6 et Conclusion re-ecrites sur le verdict dual ; l'hypothese 'fenetre de relaxation' est retiree (refutee par 19b). Re-execution integrale papermill : 11 cellules code, 0 erreur, exec continus. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

github-actions · 2026-10-11T06:09:24Z

Notebook outputs-required (H.4 schema): PASS (every code cell carries an `outputs: list`)

github-actions · 2026-10-11T06:11:05Z

prev: genre mot-clé fermant (#10093) — LEVÉ (2026-10-11T10:36:12Z).

aucun genre mots-clé fermant dans le body ni les commits ; prev: accepté(s) : #20245

Run vert du garde : ce commentaire bloquant est obsolète. Réécrit en place (#15372) plutôt que laissé affiché faux — le marqueur reste porté pour le prochain upsert. Historique : runs Always-on guards de la PR.

github-actions · 2026-10-11T06:13:18Z

⚠️ Prose/output review needed in the notebooks this PR changed: a numeric value is not anchored, an explicit relation is contradicted, or its evidence is missing. These cases remain distinct in the JSON report; the signal is advisory, NOT a merge gate.

Scope = notebooks CHANGED in this PR, not the whole corpus. Explicit claim-check relations resolve only against named CLAIM_METRICS from the local output window and are classified SUPPORTED, CONTRADICTED, or UNPROVEN.
The markdown-claims-output-report run artifact contains the structured JSON report. See python scripts/check_markdown_claims_output.py --help for re-running locally.
Detector rationale: c.290 / c.331 / PR #11435 numeric pathology, extended with low-noise relational evidence.

github-actions · 2026-10-11T06:13:22Z

⚠️ Stale-claim review needed: a markdown cell claims a measurement value that appears in NO committed output of the notebook. Advisory, NOT a merge gate — triage against the JSON artifact.

Scope = notebooks CHANGED in this PR, not the whole corpus. The stale-claim-report run artifact holds the structured JSON.
Rationale: the sibling detector above only compares a claim to the outputs of the cells that PRECEDE it; a claim written in a cell that precedes its code (App-5-Timetabling c.2/c.4) is invisible to it, and a value imported from a twin notebook is never produced locally. See python scripts/check_stale_claims.py --help.

github-actions · 2026-10-11T06:13:25Z

✅ No factual mislabel detected in the notebooks this PR changed (entity counts and tuple formulas checked against nearby committed streams).

Scope = notebooks CHANGED in this PR, not the whole corpus. The factual-mislabel-report run artifact holds the structured JSON.
Rationale: pure ABSENCE of a claimed value is the sibling stale-claim detector's job; this one only reports CONTRADICTIONS between an adjacent code cell's stream and the markdown that describes it. See python scripts/check_factual_mislabel.py --help.

github-actions · 2026-10-11T06:13:55Z

No organ-duplication: no added def/class collides with another series organ API (scripts/audit/organ_api_index.yaml).

Detector: python scripts/audit/detect_organ_duplication.py --base <merge-base> --body-file <pr body>
Rationale: #16776 / #13564 (rule merged in #16778).

github-actions · 2026-10-11T06:25:52Z

Golden-Set Execution (H.7 P3)

✅ 9/9 notebooks passed (certified reproducible)

Notebook	Status	Time
2.1-Workflow-ML.ipynb	✅ SUCCESS	4.6s
2.2-Descente-de-gradient.ipynb	✅ SUCCESS	3.2s
2.3-Regression-lineaire-logistique.ipynb	✅ SUCCESS	14.0s
2.4-Arbres-Forets-Ensembles.ipynb	✅ SUCCESS	10.1s
Search-01-StateSpace.ipynb	✅ SUCCESS	4.1s
SL-1-LogicalLearning.ipynb	✅ SUCCESS	1.7s
RL-04-Bandits-Manchots-Python.ipynb	✅ SUCCESS	14.4s
GameTheory-04c-NashExistence-Python.ipynb	✅ SUCCESS	2.7s
GameTheory-13d-Optimistic-CFR-Python.ipynb	✅ SUCCESS	8.5s

Pinned lockfile: scripts/notebook_tools/golden_set.lock.txt (H.7 P3, axe A #4208)

github-actions · 2026-10-11T06:32:08Z

Notebook PR Validation: PASS

Notebooks checked: 1
Code cells validated: 11
Result: All passed

Checks: H.1 (no errors), H.3 (execution_count), C.1 (no banned patterns)
Non-Python kernels (.NET/Lean): C.1 + errors only (execution_count advisory)
QuantConnect notebooks: C.1 + errors only (require QC Cloud for execution)

Le kernel drift guard (base vs PR) mesurait 3.12.13 -> 3.10.18 : la re-execution precedente avait tourne sous l'interpreteur par defaut. Re-execution integrale via kernel coursia-ml (Python 3.12.13, numpy 2.4.4) : 11/11 cellules, 0 erreur, execution_counts 1-11. Seule difference de sortie vs le run precedent : la version numpy affichee par la cellule de configuration (2.2.6 -> 2.4.4) ; toutes les valeurs de simulation citees en prose sont inchangees. Chemins papermill normalises au basename (tolerance metadata). Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

jsboige · 2026-10-12T07:27:04Z

[ADJOINT PREFLIGHT]
schema: 1
lane: myia-po-2025:CoursIA-2
pr: 20334
head: de6efde
complete: true
body: read
comments-reviewed: 8
reviews-reviewed: 0
threads-reviewed: 0
threads-unresolved: 0
surfaces-sha256: fb2ae253274f221aceb38c89fdb114e3fa065ac6d0d3807189f11d9a16f1824b
diff-files: 1
diff-additions: 469
diff-deletions: 95
checks: latest-wins-green
b0: clear
scope: pass
domain: not-applicable
verdict: READY
organ: check_adjoint_prevalidation.py
organ-command: python scripts/check_adjoint_prevalidation.py --derive-verdict 20334
organ-rc: 0
[/ADJOINT PREFLIGHT]

myia-ai-01

Disposition coordinateur à la tête de6efde : dossier tiers lu ; surfaces de discussion relues (body, commentaires, reviews, threads) sans réserve en attente ; portée conforme au titre ; preuve centrale vérifiée sur la PR.

— ai-01, coordination

jsboige mentioned this pull request Oct 11, 2026

[EPIC] Derive de degenerescence des notebooks a travers leurs revisions — forensic d'historique git #9768

Open

jsboige and others added 2 commits October 11, 2026 12:18

Merge branch 'main' into feature/ict-19-enjeu1-repair-gain

6204643

myia-ai-01 approved these changes Oct 12, 2026

View reviewed changes

myia-ai-01 merged commit b34d086 into main Oct 12, 2026
92 of 93 checks passed

Navigation Menu

Repository navigation

Fix(ict,#20321): ICT-19 -- Gate ENJEU-1 re-exprime sur repair_gain (espace de champ) - #20334

Fix(ict,#20321): ICT-19 -- Gate ENJEU-1 re-exprime sur repair_gain (espace de champ)#20334
myia-ai-01 merged 3 commits into
mainfrom
feature/ict-19-enjeu1-repair-gain

jsboige commented Oct 11, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Oct 11, 2026

Uh oh!

github-actions Bot commented Oct 11, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Oct 11, 2026

Uh oh!

github-actions Bot commented Oct 11, 2026

Uh oh!

github-actions Bot commented Oct 11, 2026

Uh oh!

github-actions Bot commented Oct 11, 2026

Uh oh!

github-actions Bot commented Oct 11, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Oct 11, 2026

Uh oh!

jsboige commented Oct 12, 2026

Uh oh!

myia-ai-01 left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

jsboige commented Oct 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What

Validation

Uh oh!

github-actions Bot commented Oct 11, 2026

Notebook outputs-required (H.4 schema): PASS (every code cell carries an outputs: list)

Uh oh!

github-actions Bot commented Oct 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Oct 11, 2026

Uh oh!

github-actions Bot commented Oct 11, 2026

Uh oh!

github-actions Bot commented Oct 11, 2026

Uh oh!

github-actions Bot commented Oct 11, 2026

Uh oh!

github-actions Bot commented Oct 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Golden-Set Execution (H.7 P3)

Uh oh!

github-actions Bot commented Oct 11, 2026

Notebook PR Validation: PASS

Uh oh!

jsboige commented Oct 12, 2026

Uh oh!

myia-ai-01 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

jsboige commented Oct 11, 2026 •

edited

Loading

Notebook outputs-required (H.4 schema): PASS (every code cell carries an `outputs: list`)

github-actions Bot commented Oct 11, 2026 •

edited

Loading

github-actions Bot commented Oct 11, 2026 •

edited

Loading