Skip to content

Fix(gametheory): contre-audit GT-16j Screening-Loyer — « convexe » → « concave » (prose vs mesure) - #24204

Closed
jsboige wants to merge 1 commit into
mainfrom
fix/gt16j-counteraudit
Closed

jsboige wants to merge 1 commit into
mainfrom
fix/gt16j-counteraudit

Conversation

@jsboige

@jsboige jsboige commented Nov 2, 2026

Copy link
Copy Markdown
Owner

Grain: DEEP/notebook-python — lane myia-po-2024:CoursIA-2 — prev: DEEP/notebook-python #24202

Contre-audit de fond — GT-16j Screening & loyer d'information (Python)

Arc contre-audit GameTheory, carnet 16j. Méthode 4 passes ; claims
quantitatives, attributions et références croisées vérifiées FIRSTHAND
contre origin/main.

Défaut (pass 1 — prose en contradiction avec la mesure)

Les cellules 15 et 22 disaient que le coût de l'asymétrie d'information
« croît de façon convexe avec l'écart de type ». Mais le tableau de
sensibilité (cellule 14) mesure une croissance à rythme décroissant :
les pentes successives de la colonne « coût esp. » entre les lignes sont
0.084, 0.040, 0.022, 0.012 (pour θ_H = 1.25 → 1.5 → 2.0 → 2.5 → 3.0).
Une fonction croissante à pente décroissante est concave, pas convexe.
L'adjectif était donc inversé par rapport à la mesure que la phrase
elle-même cite (≈ 0.04 / 0.083 / ≈ 0.10). Corrigé en « concave »
(2 occurrences, markdown seul).

Passes 1–3 — clean sinon (vérifiées à la main)

  • Quantitatif vs sortie : recomputé à la main et concordant — premier
    optimum (a* = θ² : a_L=1, a_H=4, surplus 0.5 / 4.0) ; second optimum
    (screening, non-distorsion au sommet : a_H=4, a_L = 0.5/0.75 = 0.667,
    loyer U_H = 0.111, t_H = 4.111) ; décomposition du coût — fraction exacte
    1/12 = 1/18 + 1/36 (coût espéré 0.083 = loyer 0.056 + distorsion
    0.028, soit 67 % / 33 %) ; et le tableau de sensibilité complet (ligne
    θ_H=3.00 re-tracée à la main : a_L=0.600, loyer 0.120, distorsion
    0.080, coût 0.100).
  • Attributions : Maskin–Riley (screening / sélection adverse) et Myerson
    (conception optimale) — réels et appropriés ; « théorème de non-distorsion
    au sommet » = nom descriptif du résultat classique (type haut non déformé),
    pas une fausse attribution.
  • Provenance « mesuré » : les résultats proviennent de l'exécution réelle
    numpy (cellules 2, 5, 8, 11, 14) ; les cellules d'exercice (17, 19, 21) sont
    des stubs C.1 conformes (= None) — aucune claim « mesuré » ne repose sur un
    stub.

Pass 4 — clean

Les cibles de nav de la cellule 0 existent sur origin/main
(git ls-tree) : GameTheory-16e-LLM-Players-Othman-Sandholm-Python.ipynb,
README.md, GameTheory-21-Capstone-Choix-Social-Python.ipynb.

Fix

  • cellules 15 et 22 (markdown) : « croît de façon convexe » / « croît
    convexe » → « concave » — markdown seul, pas de re-exécution.

Diff : 1 fichier, 2 lignes.


🤖 Generated with Claude Code

…concave"

Cellules 15 et 22 : "le cout de l'asymetrie croit de facon convexe avec
l'ecart de type". Mais le tableau de sensibilite (cellule 14) mesure une
croissance a rythme DECREISSANT : les pentes successives de la colonne "cout
esp." sont 0.084, 0.040, 0.022, 0.012 (theta_H = 1.25 -> 3.0). Une fonction
croissante a pente decreissante est concave, pas convexe. L'adjectif etait
inverse par rapport a la mesure que la phrase elle-meme cite. Corrige en
"concave" (2 occ, markdown seul).

Diff : 1 fichier, 2 lignes.

Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
@github-actions

github-actions Bot commented Nov 2, 2026

Copy link
Copy Markdown
Contributor

Golden-Set Execution (H.7 P3)

✅ 9/9 notebooks passed (certified reproducible)

Notebook Status Time
2.1-Workflow-ML.ipynb ✅ SUCCESS 4.7s
2.2-Descente-de-gradient.ipynb ✅ SUCCESS 4.8s
2.3-Regression-lineaire-logistique.ipynb ✅ SUCCESS 5.4s
2.4-Arbres-Forets-Ensembles.ipynb ✅ SUCCESS 4.9s
Search-01-StateSpace.ipynb ✅ SUCCESS 4.2s
SL-1-LogicalLearning.ipynb ✅ SUCCESS 2.6s
RL-04-Bandits-Manchots-Python.ipynb ✅ SUCCESS 26.3s
GameTheory-04c-NashExistence-Python.ipynb ✅ SUCCESS 3.4s
GameTheory-13d-Optimistic-CFR-Python.ipynb ✅ SUCCESS 17.4s

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

@github-actions github-actions Bot added the variation-adjacency-deep-med Adjacence DEEP/MED hors LIGHT : §2 l'autorise si substance distincte (coordinateur) label Nov 2, 2026
@github-actions

github-actions Bot commented Nov 2, 2026

Copy link
Copy Markdown
Contributor

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

@github-actions

github-actions Bot commented Nov 2, 2026

Copy link
Copy Markdown
Contributor

Notebook PR Validation: PASS

  • Notebooks checked: 1
  • Code cells validated: 8
  • 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)

@github-actions

github-actions Bot commented Nov 2, 2026

Copy link
Copy Markdown
Contributor

✅ No prose/output mismatch detected in the notebooks this PR changed.

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

github-actions Bot commented Nov 2, 2026

Copy link
Copy Markdown
Contributor

⚠️ 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

github-actions Bot commented Nov 2, 2026

Copy link
Copy Markdown
Contributor

✅ 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

github-actions Bot commented Nov 2, 2026

Copy link
Copy Markdown
Contributor

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

github-actions Bot commented Nov 2, 2026

Copy link
Copy Markdown
Contributor

Path-collision (organ #13359/#13615)

Cette PR #24204 (Fix(gametheory): contre-audit GT-16j Screening-Loyer — « convexe » → « concave » (prose vs mesure)) touche au moins un chemin de fichier aussi modifie par d'autres PRs ouvertes. Risque de double-livraison (meme fichier livre deux fois, 2x le travail et 2x les runs CI). Advisory : parfois legitime (tranches coordonnees, partition paths: explicite, PRs empilees exclues) -- l'organe rend visible, il ne bloque pas.

@jsboige

jsboige commented Nov 3, 2026

Copy link
Copy Markdown
Owner Author

Contenu fondu en fournee dans #24354 (contre-audit GameTheory, concern #24280) — le commit est preserve par cherry-pick, rien n'est reecrit. Cette PR se ferme vers la fournee.

@jsboige

jsboige commented Nov 3, 2026

Copy link
Copy Markdown
Owner Author

[REPLIÉE] Repliée en fournée dans la PR #24361 (classe contre-audit GameTheory, mandat user 02/11 « thin-prs-go-out-in-class-batches »). Le(s) commit(s) d'origine sont cherry-pickés tels quels dans la fournée — rien n'est réécrit. Voir #24361.

myia-ai-01 added a commit that referenced this pull request Nov 4, 2026
…en une PR (#24361)

* Fix(gametheory): contre-audit fournée — 8 carnets GameTheory repliés en une PR

Repli des 8 PRs de la classe contre-audit GameTheory en une seule PR
(mandat user 02/11 #24280 "thin-prs-go-out-in-class-batches"). Cherry-pick
des commits existants ; 7 carnets byte-identiques a la tete de leur PR
d'origine. GT-06 reconcilie sa barre de nav avec l'ordre courant de main
(conflit additif : main a reordonne le lien 06i entre-temps) en y appliquant
le fix 5->5c de la PR d'origine.

- GT-06   #23995 (0d74329, 4da630b, 5e0c807)
- GT-16b  #24176 (40cca9d)
- GT-16g  #24190 (88d579e)
- GT-16h  #24196 (3c82573)
- GT-16i  #24202 (6523d4d)
- GT-16j  #24204 (7f9945f)
- GT-18b  #24226 (e199bbf)
- GT-18c  #24241 (f5770bd)

Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>

* Fix(gametheory): re-execution GT-16h + GT-16b bout en bout (ratchet C.2)

Re-execution Papermill (kernel python3) des deux carnets de la fournée, en
réponse à la réserve CHANGES_REQUESTED du coordinateur à la tête 5d3a719 :
la sortie de la cellule 6 de GT-16h (print 2R > T+S) avait été retouchée à la
main pour suivre la source ((8 > 5) -> (6 > 5)) sans ré-exécution, et GT-16b
avait une cellule de code modifiée (commentaire) sans ré-exécution. Les sorties
committées sont maintenant celles du kernel.

Les carnets sont déterministes (seeds gelées [0,1,7,42,99]) : les sorties
stochastiques se reproduisent byte-identiques, seule la cellule 15 de GT-16h
gagne un saut de ligne cosmétique. La prose reste valide (nombres identiques).

See #24280

Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>

* Fix(gametheory,#24361): re-attest the EvolutionTrust twin pair (markdown-only nav retarget)

The fournee retargets the two navigation pointers of
GameTheory-06-EvolutionTrust-Python.ipynb (head navigation + footer
"Notebook precedent") from GameTheory-05-ZeroSum-Minimax to
GameTheory-05c-KKT-Dualite-Lagrangienne, which is the notebook the 06
actually follows.

Markdown-only: 0 code cell touched, execution_count and outputs
byte-identical to the base (C.2 markdown-only exception). The C# twin
blob (166ef01) and its content hash (ddc0a1e35b) are unchanged, so
the parity level "semantic" is preserved.

The registry stored the base blob (c5dabb8), which this PR makes
stale -> DRIFT_INTRODUCED on the Twin parity audit (#8057). Re-attested
at ordinal 0055 with a known_differences line describing the edit
(cf #8957: the attestation comes last).

Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Haiku 5.5 <noreply@anthropic.com>
Co-authored-by: myia-ai-01 <myia.ai.01.myia@gmail.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

variation-adjacency-deep-med Adjacence DEEP/MED hors LIGHT : §2 l'autorise si substance distincte (coordinateur)

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant