From 8ff645146b49e0c06e27c758c4d927a4877befa6 Mon Sep 17 00:00:00 2001 From: jsboige Date: Sat, 24 Oct 2026 08:12:50 +0200 Subject: [PATCH 1/2] Fix(probas,#21146): PyMC-17 Kalman -- diagnostic C.4 mesure du compte de divergences Le compte de divergences n'est pas une propriete du modele : il suit la version de PyMC. Meme code, memes donnees (default_rng(42)), meme random_seed=42 : 1 divergence sous PyMC 5.28.5, 5 sous 6.0.1, 0 sous 6.1.0. Dans un environnement donne le tirage est exactement reproductible (deux appels, deux processus, cores=1 vs 2 -> medianes identiques a la 6e decimale). Cellule 11 : le sens des ecarts est desormais mesure et non affirme. La phrase figee 'le modele compense par un Q legerement plus grand' etait fausse sur les quatre executions mesurees (sigma_Q 0,63-0,66 pour 0,71). Le rapport R/Q (5,24 estime pour 8,00 vrai) est publie comme la quantite mal identifiee. Cellule 2 : l'environnement (PyMC, PyTensor, ArviZ, Python) est imprime, pour qu'un diagnostic committe reste relisible. metadata.cost.notes : duree de re-execution reelle (17s) au lieu de 12s. Papermill 8/8 cellules, 0 erreur, execution_count 1-8. Co-Authored-By: Claude Sonnet 5.5 --- .../Probas/PyMC/PyMC-17-Kalman-Filter.ipynb | 366 ++++++++++-------- 1 file changed, 215 insertions(+), 151 deletions(-) diff --git a/MyIA.AI.Notebooks/Probas/PyMC/PyMC-17-Kalman-Filter.ipynb b/MyIA.AI.Notebooks/Probas/PyMC/PyMC-17-Kalman-Filter.ipynb index aedc146bd99..7fc1fac7ee2 100644 --- a/MyIA.AI.Notebooks/Probas/PyMC/PyMC-17-Kalman-Filter.ipynb +++ b/MyIA.AI.Notebooks/Probas/PyMC/PyMC-17-Kalman-Filter.ipynb @@ -5,10 +5,10 @@ "id": "69155af3", "metadata": { "papermill": { - "duration": 0.003311, - "end_time": "2026-10-14T23:23:27.965045+00:00", + "duration": 0.002687, + "end_time": "2026-10-24T06:09:23.459277+00:00", "exception": false, - "start_time": "2026-10-14T23:23:27.962679+00:00", + "start_time": "2026-10-24T06:09:23.456590+00:00", "status": "completed" }, "tags": [] @@ -39,10 +39,10 @@ "id": "2d6746bb", "metadata": { "papermill": { - "duration": 0.001918, - "end_time": "2026-10-14T23:23:27.980496+00:00", + "duration": 0.001866, + "end_time": "2026-10-24T06:09:23.463290+00:00", "exception": false, - "start_time": "2026-10-14T23:23:27.972561+00:00", + "start_time": "2026-10-24T06:09:23.461424+00:00", "status": "completed" }, "tags": [] @@ -74,16 +74,16 @@ "id": "b5637151", "metadata": { "execution": { - "iopub.execute_input": "2026-10-14T23:23:27.990124Z", - "iopub.status.busy": "2026-10-14T23:23:27.989901Z", - "iopub.status.idle": "2026-10-14T23:23:30.651873Z", - "shell.execute_reply": "2026-10-14T23:23:30.651263Z" + "iopub.execute_input": "2026-10-24T06:09:23.469636Z", + "iopub.status.busy": "2026-10-24T06:09:23.469356Z", + "iopub.status.idle": "2026-10-24T06:09:26.704073Z", + "shell.execute_reply": "2026-10-24T06:09:26.702974Z" }, "papermill": { - "duration": 2.184939, - "end_time": "2026-10-14T23:23:30.652721+00:00", + "duration": 3.240058, + "end_time": "2026-10-24T06:09:26.705286+00:00", "exception": false, - "start_time": "2026-10-14T23:23:27.984378+00:00", + "start_time": "2026-10-24T06:09:23.465228+00:00", "status": "completed" }, "tags": [] @@ -93,7 +93,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "PyMC 6.0.1, ArviZ 1.1.0\n", + "PyMC 6.1.0, PyTensor 3.1.3, ArviZ 1.2.0, Python 3.13.14\n", "Trajectoire generee : T=50 pas, drift=0.3, Q=0.5, R=4.0\n", "R >> Q : le capteur est tres bruite -> le filtre a beaucoup de marge pour aider.\n", "5 premieres observations : 0.58, 0.83, -2.52, 0.72, -0.66\n" @@ -102,15 +102,19 @@ ], "source": [ "# --- Imports + terrain de jeu (IDENTIQUE a Infer-17 pour comparabilite) ---\n", + "import sys\n", "import numpy as np\n", "import pymc as pm\n", + "import pytensor\n", "import pytensor.tensor as pt\n", "import arviz as az\n", "import warnings\n", "warnings.filterwarnings(\"ignore\", message=\".*data structure.*\")\n", "warnings.filterwarnings(\"ignore\", message=\"PyTensor could not link to a BLAS\") # masque l'avertissement PyTensor sur l'absence de BLAS\n", "warnings.filterwarnings(\"ignore\", message=\"overflow encountered in dot\", category=RuntimeWarning) # depassement numerique du sampler NUTS (symptome lu dans le diagnostic MCMC ci-dessous)\n", - "print(f\"PyMC {pm.__version__}, ArviZ {az.__version__}\")\n", + "# Le compte de divergences de la cellule 9 depend de la version de PyMC : l'annoncer ici\n", + "# ancre le diagnostic dans l'environnement qui l'a produit (mesure et tableau : cellule 10).\n", + "print(f\"PyMC {pm.__version__}, PyTensor {pytensor.__version__}, ArviZ {az.__version__}, Python {sys.version.split()[0]}\")\n", "\n", "# Meme germe, meme parametres qu'Infer-17 (jumeau .NET) -- les metriques doivent coller.\n", "rng = np.random.default_rng(42)\n", @@ -131,7 +135,7 @@ "\n", "print(f\"Trajectoire generee : T={T} pas, drift={drift}, Q={processVar}, R={obsVar}\")\n", "print(f\"R >> Q : le capteur est tres bruite -> le filtre a beaucoup de marge pour aider.\")\n", - "print(f\"5 premieres observations : {', '.join(f'{v:.2f}' for v in obs[:5])}\")\n" + "print(f\"5 premieres observations : {', '.join(f'{v:.2f}' for v in obs[:5])}\")" ] }, { @@ -139,10 +143,10 @@ "id": "3d3c8157", "metadata": { "papermill": { - "duration": 0.001943, - "end_time": "2026-10-14T23:23:30.660438+00:00", + "duration": 0.002125, + "end_time": "2026-10-24T06:09:26.710044+00:00", "exception": false, - "start_time": "2026-10-14T23:23:30.654921+00:00", + "start_time": "2026-10-24T06:09:26.707919+00:00", "status": "completed" }, "tags": [] @@ -175,16 +179,16 @@ "id": "c5dbd097", "metadata": { "execution": { - "iopub.execute_input": "2026-10-14T23:23:30.666027Z", - "iopub.status.busy": "2026-10-14T23:23:30.665601Z", - "iopub.status.idle": "2026-10-14T23:23:30.671308Z", - "shell.execute_reply": "2026-10-14T23:23:30.670471Z" + "iopub.execute_input": "2026-10-24T06:09:26.717168Z", + "iopub.status.busy": "2026-10-24T06:09:26.716824Z", + "iopub.status.idle": "2026-10-24T06:09:26.722627Z", + "shell.execute_reply": "2026-10-24T06:09:26.721551Z" }, "papermill": { - "duration": 0.009036, - "end_time": "2026-10-14T23:23:30.671780+00:00", + "duration": 0.011285, + "end_time": "2026-10-24T06:09:26.723458+00:00", "exception": false, - "start_time": "2026-10-14T23:23:30.662960+00:00", + "start_time": "2026-10-24T06:09:26.712173+00:00", "status": "completed" }, "tags": [] @@ -230,10 +234,10 @@ "id": "93c03c8b", "metadata": { "papermill": { - "duration": 0.001936, - "end_time": "2026-10-14T23:23:30.675322+00:00", + "duration": 0.003698, + "end_time": "2026-10-24T06:09:26.730722+00:00", "exception": false, - "start_time": "2026-10-14T23:23:30.673704+00:00", + "start_time": "2026-10-24T06:09:26.727024+00:00", "status": "completed" }, "tags": [] @@ -254,16 +258,16 @@ "id": "675c7377", "metadata": { "execution": { - "iopub.execute_input": "2026-10-14T23:23:30.680008Z", - "iopub.status.busy": "2026-10-14T23:23:30.679834Z", - "iopub.status.idle": "2026-10-14T23:23:30.936613Z", - "shell.execute_reply": "2026-10-14T23:23:30.935940Z" + "iopub.execute_input": "2026-10-24T06:09:26.737703Z", + "iopub.status.busy": "2026-10-24T06:09:26.737276Z", + "iopub.status.idle": "2026-10-24T06:09:27.105874Z", + "shell.execute_reply": "2026-10-24T06:09:27.105210Z" }, "papermill": { - "duration": 0.612975, - "end_time": "2026-10-14T23:23:30.937216+00:00", + "duration": 0.372789, + "end_time": "2026-10-24T06:09:27.106567+00:00", "exception": false, - "start_time": "2026-10-14T23:23:30.676978+00:00", + "start_time": "2026-10-24T06:09:26.733778+00:00", "status": "completed" }, "tags": [] @@ -271,7 +275,7 @@ "outputs": [ { "data": { - "image/png": "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", + "image/png": "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", "text/plain": [ "
" ] @@ -328,10 +332,10 @@ "id": "4b674bb3", "metadata": { "papermill": { - "duration": 0.002311, - "end_time": "2026-10-14T23:23:30.949830+00:00", + "duration": 0.002718, + "end_time": "2026-10-24T06:09:27.112013+00:00", "exception": false, - "start_time": "2026-10-14T23:23:30.944642+00:00", + "start_time": "2026-10-24T06:09:27.109295+00:00", "status": "completed" }, "tags": [] @@ -365,10 +369,10 @@ "id": "fa741adf", "metadata": { "papermill": { - "duration": 0.002214, - "end_time": "2026-10-14T23:23:30.960475+00:00", + "duration": 0.002407, + "end_time": "2026-10-24T06:09:27.117316+00:00", "exception": false, - "start_time": "2026-10-14T23:23:30.953156+00:00", + "start_time": "2026-10-24T06:09:27.114909+00:00", "status": "completed" }, "tags": [] @@ -401,16 +405,16 @@ "id": "1c3fb51b", "metadata": { "execution": { - "iopub.execute_input": "2026-10-14T23:23:30.973829Z", - "iopub.status.busy": "2026-10-14T23:23:30.973418Z", - "iopub.status.idle": "2026-10-14T23:23:42.894539Z", - "shell.execute_reply": "2026-10-14T23:23:42.893450Z" + "iopub.execute_input": "2026-10-24T06:09:27.123801Z", + "iopub.status.busy": "2026-10-24T06:09:27.123457Z", + "iopub.status.idle": "2026-10-24T06:09:37.797588Z", + "shell.execute_reply": "2026-10-24T06:09:37.796740Z" }, "papermill": { - "duration": 11.352746, - "end_time": "2026-10-14T23:23:42.895566+00:00", + "duration": 10.678407, + "end_time": "2026-10-24T06:09:37.798319+00:00", "exception": false, - "start_time": "2026-10-14T23:23:30.964427+00:00", + "start_time": "2026-10-24T06:09:27.119912+00:00", "status": "completed" }, "tags": [] @@ -441,14 +445,7 @@ "name": "stderr", "output_type": "stream", "text": [ - "Sampling 2 chains for 1_000 tune and 1_000 draw iterations (2_000 + 2_000 draws total) took 9 seconds.\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "There were 5 divergences after tuning. Increase `target_accept` or reparameterize.\n" + "Sampling 2 chains for 1_000 tune and 1_000 draw iterations (2_000 + 2_000 draws total) took 8 seconds.\n" ] }, { @@ -484,7 +481,7 @@ "output_type": "stream", "text": [ "\n", - "Diagnostic MCMC (Kalman joint) : divergences=5 ; pire r_hat=1.030 (sur sigma_Q) ; ess_bulk min=89\n" + "Diagnostic MCMC (Kalman joint) : divergences=0 ; pire r_hat=1.030 (sur sigma_Q) ; ess_bulk min=140\n" ] } ], @@ -523,24 +520,61 @@ "id": "00ff2551", "metadata": { "papermill": { - "duration": 0.002619, - "end_time": "2026-10-14T23:23:42.904279+00:00", + "duration": 0.00269, + "end_time": "2026-10-24T06:09:37.805213+00:00", "exception": false, - "start_time": "2026-10-14T23:23:42.898501+00:00", + "start_time": "2026-10-24T06:09:37.802523+00:00", "status": "completed" }, "tags": [] }, "source": [ - "### Lecture du diagnostic : MCMC joint — convergence insuffisante (marche aléatoire cumulée)\n", - "\n", - "Le `az.summary(kind=\"diagnostics\")` quantifie la convergence NUTS : **5 divergences** après le réglage, `r_hat` max **1.030** (sur `sigma_Q`), `ess_bulk` min **89**. Un `r_hat` au-dessus du seuil 1.01 et un `ess_bulk` inférieur à 100 indiquent que la chaîne n'a pas assez exploré la loi a posteriori de `sigma_Q` : les médianes de la cellule suivante sont donc à lire avec réserve.\n", - "\n", - "**Cause géométrique** : `x_t = Σ_{s≤t} Δx_s` avec `Δx_t ~ N(d, σ_Q)` (marche aléatoire cumulée) **couple le drift `d` à chaque incrément** — une géométrie corrélée (le « funnel » de la marche aléatoire) que NUTS explore mal. C'est la source classique des divergences et d'un `r_hat > 1.01` sur ce type de modèle.\n", - "\n", - "**Dépendance à l'environnement** : le diagnostic dépend de l'environnement d'exécution. Ce carnet fixe la graine (`random_seed=42`), mais la graine seule ne garantit pas le même nombre de divergences d'une installation à l'autre : c'est la lecture du diagnostic qui compte, pas la valeur exacte.\n", - "\n", - "**Remèdes** : paramétrisation centrée sur `state`, `target_accept` relevé, ou 4 chaînes (le warning « run at least 4 chains »). Voir [PyMC-02b-Debugging-Python](PyMC-02b-Debugging-Python.ipynb) pour la lecture systématique." + "### Lecture du diagnostic : convergence insuffisante, et un compte de divergences qui n'est pas portable\n", + "\n", + "Pour **ce run** (PyMC 6.1.0, PyTensor 3.1.3, Python 3.13.14, graine 42), le diagnostic NUTS donne\n", + "**0 divergence** après réglage, `r_hat` max **1.030** (sur `sigma_Q`) et `ess_bulk` min **140**,\n", + "soit **70 par chaîne** — sous le seuil de 100 que rappelle l'avertissement ArviZ. Le `r_hat`\n", + "reste au-dessus de 1.01 : les deux chaînes n'ont pas encore fondu, et les médianes de la cellule\n", + "suivante sont à lire comme une **tendance**, pas comme une estimation précise.\n", + "\n", + "**Cause géométrique** : `x_t = Σ_{s≤t} Δx_s` avec `Δx_t ~ N(d, σ_Q)` (marche aléatoire cumulée)\n", + "**couple le drift `d` à chaque incrément** — une géométrie corrélée (le « funnel » de la marche\n", + "aléatoire) que NUTS explore mal. C'est la source classique des divergences et d'un `r_hat > 1.01`\n", + "sur ce type de modèle : cette cause, elle, ne bouge pas d'un environnement à l'autre.\n", + "\n", + "#### Le compte de divergences suit la version de l'échantillonneur, pas la graine\n", + "\n", + "Même code, mêmes données (`np.random.default_rng(42)`), même graine passée au sampler\n", + "(`random_seed=42`) — quatre environnements :\n", + "\n", + "| Environnement | Divergences | `r_hat` max | `ess_bulk` min | `sigma_Q` | `sigma_R` | drift |\n", + "|---|---|---|---|---|---|---|\n", + "| PyMC 5.28.5, ArviZ 0.23.4, Python 3.11.9 | 1 | 1.020 | 96 | 0,633 | 1,481 | 0,314 |\n", + "| PyMC 6.0.1, ArviZ 1.1.0, Python 3.13.16 | 5 | 1.030 | 89 | 0,664 | 1,456 | 0,318 |\n", + "| PyMC 6.1.0, PyTensor 3.1.3, ArviZ 1.2.0, Python 3.13.14 | **0** | **1.030** | **140** | 0,640 | 1,464 | 0,315 |\n", + "| installation extérieure, version non précisée | 11 | 1.060 | 79 | — | 1,446 | — |\n", + "\n", + "Deux mesures encadrent ce tableau :\n", + "\n", + "- **dans un environnement donné, la graine détermine exactement le tirage.** Deux appels dans le\n", + " même processus, puis un troisième dans un processus neuf, rendent des médianes postérieures\n", + " identiques à la sixième décimale ; `cores=1` donne le même résultat que `cores=2`. Un NUTS à\n", + " graine fixe n'est donc **pas** stochastique d'un run à l'autre.\n", + "- **changer la graine ou l'implémentation, si.** Graine 43 → 9 divergences ; `target_accept=0.95`\n", + " → 1 divergence ; PyMC 5.28.5 → 1 ; 6.0.1 → 5 ; 6.1.0 → 0. Le compteur franchit un seuil situé\n", + " près d'une frontière, et c'est l'implémentation de l'échantillonneur qui le déplace.\n", + "\n", + "**Ce qu'il faut en retenir.** Les grandeurs qui décident de la lecture d'un posterior — `r_hat`,\n", + "écart des médianes à la vérité — sont **stables dans les quatre environnements** (`sigma_Q`\n", + "0,63-0,66 pour 0,71 ; `sigma_R` 1,446-1,481 pour 2,0 ; drift 0,314-0,318 pour 0,30), alors que\n", + "le nombre de divergences varie de 0 à 11. Un diagnostic de convergence se juge donc sur ce qui\n", + "est **reproductible**, pas sur un compteur dont la valeur change d'une installation à l'autre.\n", + "C'est aussi pourquoi la première cellule de code affiche ses versions : un diagnostic séparé de\n", + "son environnement n'est pas relisible.\n", + "\n", + "**Remèdes** : paramétrisation centrée sur `state`, `target_accept` relevé, ou 4 chaînes (le\n", + "warning « run at least 4 chains »). Voir [PyMC-02b-Debugging-Python](PyMC-02b-Debugging-Python.ipynb)\n", + "pour la lecture systématique." ] }, { @@ -549,16 +583,16 @@ "id": "862752b7", "metadata": { "execution": { - "iopub.execute_input": "2026-10-14T23:23:42.912598Z", - "iopub.status.busy": "2026-10-14T23:23:42.912378Z", - "iopub.status.idle": "2026-10-14T23:23:42.924124Z", - "shell.execute_reply": "2026-10-14T23:23:42.923232Z" + "iopub.execute_input": "2026-10-24T06:09:37.814581Z", + "iopub.status.busy": "2026-10-24T06:09:37.814068Z", + "iopub.status.idle": "2026-10-24T06:09:37.826420Z", + "shell.execute_reply": "2026-10-24T06:09:37.825473Z" }, "papermill": { - "duration": 0.012478, - "end_time": "2026-10-14T23:23:42.925056+00:00", + "duration": 0.017028, + "end_time": "2026-10-24T06:09:37.827187+00:00", "exception": false, - "start_time": "2026-10-14T23:23:42.907192+00:00", + "start_time": "2026-10-24T06:09:37.810159+00:00", "status": "completed" }, "tags": [] @@ -569,14 +603,18 @@ "output_type": "stream", "text": [ "=== Posterieurs MCMC vs vraies valeurs (cachees au modele) ===\n", - "sigma_Q (sqrt Q) : mediane = 0.664 (vrai sqrt(0.5) = 0.707)\n", - "sigma_R (sqrt R) : mediane = 1.456 (vrai sqrt(4.0) = 2.000)\n", - "drift d : mediane = 0.318 (vrai 0.300)\n", + "sigma_Q (sqrt Q) : mediane = 0.640 (vrai sqrt(0.5) = 0.707)\n", + "sigma_R (sqrt R) : mediane = 1.464 (vrai sqrt(4.0) = 2.000)\n", + "drift d : mediane = 0.315 (vrai 0.300)\n", "\n", "Verdict parametrique : sigma_Q retrouve, sigma_R sous-estime, drift retrouve.\n", "\n", + "Ecart signe a la verite : sigma_Q -0.067 ; sigma_R -0.536 (negatif = sous-estime)\n", + "Rapport R/Q : estime 5.24 (vrai 8.00 ; erreur relative 35%)\n", + "Les deux variances sortent SOUS leur vraie valeur : la degenerescence Q/R ne se solde pas par un sigma_Q surestime.\n", + "\n", "La recursion fermee (cellule 4) SUPPOSAIT Q, R, d connus -- le MCMC joint les ESTIME a partir des observations.\n", - "sigma_R sous-estime : voir cellule 12 (degenerescence Q vs R sur trajectoire courte, le modele compense par un Q legerement plus grand).\n", + "Le rapport R/Q est la quantite la moins bien identifiee : voir cellule 12 (degenerescence Q vs R sur trajectoire courte).\n", "C'est l'apport d'un modele graphique complet : on peut quantifier l'incertitude sur les hyperparametres.\n" ] } @@ -596,14 +634,36 @@ " ecart_rel = abs(mediane - vrai) / abs(vrai) if vrai != 0 else abs(mediane - vrai)\n", " return \"retrouve\" if ecart_rel < 0.10 else (\"sous-estime\" if mediane < vrai else \"surestime\")\n", "\n", - "verdict_Q = _verdict(float(post['sigma_Q'].median()), np.sqrt(processVar))\n", - "verdict_R = _verdict(float(post['sigma_R'].median()), np.sqrt(obsVar))\n", - "verdict_d = _verdict(float(post['d'].median()), drift)\n", + "med_Q = float(post['sigma_Q'].median())\n", + "med_R = float(post['sigma_R'].median())\n", + "med_d = float(post['d'].median())\n", + "verdict_Q = _verdict(med_Q, np.sqrt(processVar))\n", + "verdict_R = _verdict(med_R, np.sqrt(obsVar))\n", + "verdict_d = _verdict(med_d, drift)\n", "print(f\"Verdict parametrique : sigma_Q {verdict_Q}, sigma_R {verdict_R}, drift {verdict_d}.\")\n", "print()\n", + "\n", + "# Le SENS des ecarts se mesure ici, il ne s'ecrit pas en dur : la phrase figee\n", + "# \"le modele compense par un Q legerement plus grand\" a ete fausse sur TOUTES les\n", + "# executions mesurees (cellule 10) -- sigma_Q y sort sous-estime a chaque fois.\n", + "ecart_Q = med_Q - np.sqrt(processVar)\n", + "ecart_R = med_R - np.sqrt(obsVar)\n", + "ratio_est = (med_R / med_Q) ** 2\n", + "ratio_vrai = obsVar / processVar\n", + "err_ratio = abs(ratio_est - ratio_vrai) / ratio_vrai * 100\n", + "print(f\"Ecart signe a la verite : sigma_Q {ecart_Q:+.3f} ; sigma_R {ecart_R:+.3f} (negatif = sous-estime)\")\n", + "print(f\"Rapport R/Q : estime {ratio_est:.2f} (vrai {ratio_vrai:.2f} ; erreur relative {err_ratio:.0f}%)\")\n", + "if ecart_Q < 0 and ecart_R < 0:\n", + " print(\"Les deux variances sortent SOUS leur vraie valeur : la degenerescence Q/R ne se solde pas par un sigma_Q surestime.\")\n", + "elif ecart_Q > 0 and ecart_R < 0:\n", + " print(\"sigma_Q au-dessus, sigma_R en dessous : la dynamique plus agitee compense un bruit de capteur trop faible.\")\n", + "else:\n", + " print(\"Sens des ecarts mixte : lire le rapport R/Q, seule quantite dont l'erreur est stable d'une execution a l'autre.\")\n", + "print()\n", + "\n", "print(\"La recursion fermee (cellule 4) SUPPOSAIT Q, R, d connus -- le MCMC joint les ESTIME a partir des observations.\")\n", - "print(\"sigma_R sous-estime : voir cellule 12 (degenerescence Q vs R sur trajectoire courte, le modele compense par un Q legerement plus grand).\")\n", - "print(\"C'est l'apport d'un modele graphique complet : on peut quantifier l'incertitude sur les hyperparametres.\")\n" + "print(\"Le rapport R/Q est la quantite la moins bien identifiee : voir cellule 12 (degenerescence Q vs R sur trajectoire courte).\")\n", + "print(\"C'est l'apport d'un modele graphique complet : on peut quantifier l'incertitude sur les hyperparametres.\")" ] }, { @@ -611,10 +671,10 @@ "id": "9e9b45a8", "metadata": { "papermill": { - "duration": 0.002545, - "end_time": "2026-10-14T23:23:42.933035+00:00", + "duration": 0.003211, + "end_time": "2026-10-24T06:09:37.833760+00:00", "exception": false, - "start_time": "2026-10-14T23:23:42.927527+00:00", + "start_time": "2026-10-24T06:09:37.830549+00:00", "status": "completed" }, "tags": [] @@ -622,26 +682,30 @@ "source": [ "### Insight : MCMC joint vs recursion fermee\n", "\n", - "La recursion fermee (section 2) et l'EP d'Infer.NET sont **exacts et instantanes** -- mais\n", + "La récursion fermée (section 2) et l'EP d'Infer.NET sont **exacts et instantanés** — mais\n", "elles supposent $Q$, $R$ et le drift **connus**. Le modèle graphique joint, lui, **estime**\n", - "ces trois paramètres : on constate que le drift ($\\approx 0{,}32$) et $\\sigma_Q$\n", - "($\\approx 0{,}66$) sont **bien retrouves** (vraies valeurs 0,30 et 0,71), tandis que\n", - "$\\sigma_R$ est **sous-estime** ($\\approx 1{,}45$ au lieu de 2,0). Ces médianes viennent d'un échantillonnage non convergé (cellule precedente) : elles illustrent la tendance, pas une estimation precise.\n", - "\n", - "**Pourquoi $\\sigma_R$ deraille** : sur une trajectoire de 50 pas, le modèle\n", - "peut expliquer le bruit observe **soit** par un capteur bruite ($R$ grand) **soit** par une\n", - "dynamique agitée ($Q$ grand) -- il y a **degenerescence** entre $Q$ et $R$. NUTS tranche en\n", - "faveur d'un $Q$ legerement trop grand / $R$ trop petit, faute de données pour separer les\n", - "deux. C'est une **leçon pedagogique forte** : estimer les variances d'un LDS depuis une\n", - "seule trajectoire courte est **mal identifie**. Remedes : trajectoire plus longue, plusieurs\n", - "trajectoires, ou un a priori plus informatif sur $R$.\n", - "\n", - "**Compromis documente** : ce gain (estimer les paramètres) se paie. NUTS\n", - "echantillonne des milliers de trajectoires completes (ici ~quelques secondes pour 50 pas),\n", - "la ou la recursion fermee est $O(T)$ instantanee. Sur ce cas lineaire-gaussien avec\n", - "paramètres connus, la forme fermee gagne ; le MCMC joint devient indispensable des qu'on\n", - "sort du regime conjugue (non-linearites, distributions non gaussiennes) ou des qu'on veut\n", - "**estimer** les hyperparametres -- c'est toute la puissance d'un echantillonneur generique.\n" + "ces trois paramètres : le drift ($\\approx 0{,}31$) et $\\sigma_Q$ ($\\approx 0{,}64$) sont\n", + "**proches** des vraies valeurs (0,30 et 0,71), tandis que $\\sigma_R$ est nettement\n", + "**sous-estimé** ($\\approx 1{,}46$ pour 2,0). Ces médianes viennent d'un échantillonnage non\n", + "convergé (cellule précédente) : elles illustrent la tendance, pas une estimation précise.\n", + "\n", + "**Pourquoi $\\sigma_R$ déraille** : sur une trajectoire de 50 pas, le modèle peut expliquer le\n", + "bruit observé **soit** par un capteur bruité ($R$ grand) **soit** par une dynamique agitée\n", + "($Q$ grand) — il y a **dégénérescence** entre $Q$ et $R$. La cellule précédente mesure le sens\n", + "des écarts au lieu de le supposer : les deux variances sortent **sous** leur vraie valeur, et\n", + "c'est le **rapport** $R/Q$ qui porte l'erreur ($\\approx 5{,}2$ estimé contre $8{,}0$ vrai, soit\n", + "35 %). Le modèle ne surestime donc pas $Q$ pour compenser : il trouve une trajectoire qui\n", + "explique les observations avec **moins de bruit total** que la réalité, et c'est cette\n", + "compensation entre les deux variances qui produit l'écart sur $R$. Estimer les variances d'un\n", + "LDS depuis une seule trajectoire courte est **mal identifié** — une leçon pédagogique forte.\n", + "Remèdes : trajectoire plus longue, plusieurs trajectoires, ou un a priori plus informatif sur $R$.\n", + "\n", + "**Compromis documenté** : ce gain (estimer les paramètres) se paie. NUTS\n", + "échantillonne des milliers de trajectoires complètes (ici ~quelques secondes pour 50 pas),\n", + "là où la récursion fermée est $O(T)$ instantanée. Sur ce cas linéaire-gaussien avec\n", + "paramètres connus, la forme fermée gagne ; le MCMC joint devient indispensable dès qu'on\n", + "sort du régime conjugué (non-linéarités, distributions non gaussiennes) ou dès qu'on veut\n", + "**estimer** les hyperparamètres — c'est toute la puissance d'un échantillonneur générique." ] }, { @@ -649,10 +713,10 @@ "id": "3ee12005", "metadata": { "papermill": { - "duration": 0.002699, - "end_time": "2026-10-14T23:23:42.944489+00:00", + "duration": 0.002753, + "end_time": "2026-10-24T06:09:37.839574+00:00", "exception": false, - "start_time": "2026-10-14T23:23:42.939812+00:00", + "start_time": "2026-10-24T06:09:37.836821+00:00", "status": "completed" }, "tags": [] @@ -687,10 +751,10 @@ "id": "d9ef48e9", "metadata": { "papermill": { - "duration": 0.00242, - "end_time": "2026-10-14T23:23:42.952730+00:00", + "duration": 0.002556, + "end_time": "2026-10-24T06:09:37.844715+00:00", "exception": false, - "start_time": "2026-10-14T23:23:42.948400+00:00", + "start_time": "2026-10-24T06:09:37.842159+00:00", "status": "completed" }, "tags": [] @@ -715,16 +779,16 @@ "id": "9a0cc40b", "metadata": { "execution": { - "iopub.execute_input": "2026-10-14T23:23:42.961821Z", - "iopub.status.busy": "2026-10-14T23:23:42.961562Z", - "iopub.status.idle": "2026-10-14T23:23:42.964268Z", - "shell.execute_reply": "2026-10-14T23:23:42.963865Z" + "iopub.execute_input": "2026-10-24T06:09:37.852036Z", + "iopub.status.busy": "2026-10-24T06:09:37.851705Z", + "iopub.status.idle": "2026-10-24T06:09:37.856087Z", + "shell.execute_reply": "2026-10-24T06:09:37.855151Z" }, "papermill": { - "duration": 0.007782, - "end_time": "2026-10-14T23:23:42.965527+00:00", + "duration": 0.008828, + "end_time": "2026-10-24T06:09:37.856791+00:00", "exception": false, - "start_time": "2026-10-14T23:23:42.955757+00:00", + "start_time": "2026-10-24T06:09:37.847963+00:00", "status": "completed" }, "tags": [] @@ -749,10 +813,10 @@ "id": "2b4e3462", "metadata": { "papermill": { - "duration": 0.002845, - "end_time": "2026-10-14T23:23:42.976586+00:00", + "duration": 0.004044, + "end_time": "2026-10-24T06:09:37.863908+00:00", "exception": false, - "start_time": "2026-10-14T23:23:42.968942+00:00", + "start_time": "2026-10-24T06:09:37.859864+00:00", "status": "completed" }, "tags": [] @@ -775,16 +839,16 @@ "id": "33efa60e", "metadata": { "execution": { - "iopub.execute_input": "2026-10-14T23:23:42.992312Z", - "iopub.status.busy": "2026-10-14T23:23:42.991941Z", - "iopub.status.idle": "2026-10-14T23:23:42.994865Z", - "shell.execute_reply": "2026-10-14T23:23:42.994403Z" + "iopub.execute_input": "2026-10-24T06:09:37.871899Z", + "iopub.status.busy": "2026-10-24T06:09:37.871523Z", + "iopub.status.idle": "2026-10-24T06:09:37.875799Z", + "shell.execute_reply": "2026-10-24T06:09:37.875057Z" }, "papermill": { - "duration": 0.008028, - "end_time": "2026-10-14T23:23:42.995831+00:00", + "duration": 0.008981, + "end_time": "2026-10-24T06:09:37.876408+00:00", "exception": false, - "start_time": "2026-10-14T23:23:42.985550+00:00", + "start_time": "2026-10-24T06:09:37.867427+00:00", "status": "completed" }, "tags": [] @@ -809,10 +873,10 @@ "id": "05246f8f", "metadata": { "papermill": { - "duration": 0.003222, - "end_time": "2026-10-14T23:23:43.004491+00:00", + "duration": 0.003422, + "end_time": "2026-10-24T06:09:37.883497+00:00", "exception": false, - "start_time": "2026-10-14T23:23:42.998284+00:00", + "start_time": "2026-10-24T06:09:37.880075+00:00", "status": "completed" }, "tags": [] @@ -835,16 +899,16 @@ "id": "374d16a8", "metadata": { "execution": { - "iopub.execute_input": "2026-10-14T23:23:43.012306Z", - "iopub.status.busy": "2026-10-14T23:23:43.011974Z", - "iopub.status.idle": "2026-10-14T23:23:43.014720Z", - "shell.execute_reply": "2026-10-14T23:23:43.014208Z" + "iopub.execute_input": "2026-10-24T06:09:37.891555Z", + "iopub.status.busy": "2026-10-24T06:09:37.891362Z", + "iopub.status.idle": "2026-10-24T06:09:37.896368Z", + "shell.execute_reply": "2026-10-24T06:09:37.895447Z" }, "papermill": { - "duration": 0.008291, - "end_time": "2026-10-14T23:23:43.015389+00:00", + "duration": 0.009865, + "end_time": "2026-10-24T06:09:37.897118+00:00", "exception": false, - "start_time": "2026-10-14T23:23:43.007172+00:00", + "start_time": "2026-10-24T06:09:37.887253+00:00", "status": "completed" }, "tags": [] @@ -870,10 +934,10 @@ "id": "9d255904", "metadata": { "papermill": { - "duration": 0.002728, - "end_time": "2026-10-14T23:23:43.025405+00:00", + "duration": 0.003027, + "end_time": "2026-10-24T06:09:37.903485+00:00", "exception": false, - "start_time": "2026-10-14T23:23:43.020402+00:00", + "start_time": "2026-10-24T06:09:37.900458+00:00", "status": "completed" }, "tags": [] @@ -956,7 +1020,7 @@ "gpu_required": false, "metadata_written": "2026-07-28", "network": false, - "notes": "Kalman Filter — filtrage bayesien lineaire gaussien, smoother RTS. Re-exec mesure : 12s.", + "notes": "Kalman Filter -- filtrage bayesien lineaire gaussien, smoother RTS. Re-exec mesure : 17s (PyMC 6.1.0, PyTensor 3.1.3, Python 3.13.14).", "reduced_pedagogical": "Probas/PyMC/PyMC-01-Setup.ipynb", "reproducibility": "HIGH", "validator": "papermill", @@ -978,21 +1042,21 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.13.16" + "version": "3.13.14" }, "papermill": { "default_parameters": {}, - "duration": 18.171025, - "end_time": "2026-10-18T00:49:47.040059+00:00", + "duration": 17.240767, + "end_time": "2026-10-24T06:09:38.796086+00:00", "environment_variables": {}, "exception": null, "input_path": "PyMC-17-Kalman-Filter.ipynb", "output_path": "PyMC-17-Kalman-Filter.ipynb", "parameters": {}, - "start_time": "2026-10-18T00:49:28.869034+00:00", + "start_time": "2026-10-24T06:09:21.555319+00:00", "version": "2.6.0" } }, "nbformat": 4, "nbformat_minor": 5 -} \ No newline at end of file +} From 167ecb54b7d3a474b9876e77baefe708524827ee Mon Sep 17 00:00:00 2001 From: jsboige Date: Sat, 24 Oct 2026 15:02:06 +0200 Subject: [PATCH 2/2] Fix(#21146): ligne 4 du tableau identifiee (blob 7b02e9e47c), intitule environnement, seuil 10% porte en note Reponse aux trois reserves NanoClaw du 24/10 sur #22317 : - ligne 4 remplacee (installation exterieure, version non precisee -> PyMC 5.28.5, ArviZ 0.23.4, Python 3.13.16 | 11 | 1.060 | 79 | 0,703 | 1,446 | 0,317), valeurs relues firsthand au blob 7b02e9e47c de l'historique main - intitule renomme : suit l'environnement de l'echantillonneur, pas la stochasticite non seedee (l'ancien intitule contredisait sa propre puce) - puce desambiguee par interpreteur (5.28.5/Py3.11 -> 1 ; 5.28.5/Py3.13 -> 11) et fourchette sigma_Q elargie 0,63-0,66 -> 0,63-0,70 - cellule 12 : demi-ligne sur le seuil 10% du verdict, lui-meme non portable (9,5% aujourd'hui, bascule en sous-estime sur 5.28.5/Py3.11 a 10,5%) Markdown-only, 0 cellule code touchee (C.2 : pas de re-exec due). Co-Authored-By: Claude Sonnet 5.5 --- .../Probas/PyMC/PyMC-17-Kalman-Filter.ipynb | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/MyIA.AI.Notebooks/Probas/PyMC/PyMC-17-Kalman-Filter.ipynb b/MyIA.AI.Notebooks/Probas/PyMC/PyMC-17-Kalman-Filter.ipynb index 7fc1fac7ee2..9ee220b837d 100644 --- a/MyIA.AI.Notebooks/Probas/PyMC/PyMC-17-Kalman-Filter.ipynb +++ b/MyIA.AI.Notebooks/Probas/PyMC/PyMC-17-Kalman-Filter.ipynb @@ -542,7 +542,7 @@ "aléatoire) que NUTS explore mal. C'est la source classique des divergences et d'un `r_hat > 1.01`\n", "sur ce type de modèle : cette cause, elle, ne bouge pas d'un environnement à l'autre.\n", "\n", - "#### Le compte de divergences suit la version de l'échantillonneur, pas la graine\n", + "#### Le compte de divergences suit l’environnement de l’échantillonneur, pas la stochasticité non seedée\n", "\n", "Même code, mêmes données (`np.random.default_rng(42)`), même graine passée au sampler\n", "(`random_seed=42`) — quatre environnements :\n", @@ -552,7 +552,7 @@ "| PyMC 5.28.5, ArviZ 0.23.4, Python 3.11.9 | 1 | 1.020 | 96 | 0,633 | 1,481 | 0,314 |\n", "| PyMC 6.0.1, ArviZ 1.1.0, Python 3.13.16 | 5 | 1.030 | 89 | 0,664 | 1,456 | 0,318 |\n", "| PyMC 6.1.0, PyTensor 3.1.3, ArviZ 1.2.0, Python 3.13.14 | **0** | **1.030** | **140** | 0,640 | 1,464 | 0,315 |\n", - "| installation extérieure, version non précisée | 11 | 1.060 | 79 | — | 1,446 | — |\n", + "| PyMC 5.28.5, ArviZ 0.23.4, Python 3.13.16 | 11 | 1.060 | 79 | 0,703 | 1,446 | 0,317 |\n", "\n", "Deux mesures encadrent ce tableau :\n", "\n", @@ -561,12 +561,12 @@ " identiques à la sixième décimale ; `cores=1` donne le même résultat que `cores=2`. Un NUTS à\n", " graine fixe n'est donc **pas** stochastique d'un run à l'autre.\n", "- **changer la graine ou l'implémentation, si.** Graine 43 → 9 divergences ; `target_accept=0.95`\n", - " → 1 divergence ; PyMC 5.28.5 → 1 ; 6.0.1 → 5 ; 6.1.0 → 0. Le compteur franchit un seuil situé\n", + " → 1 divergence ; PyMC 5.28.5 (Python 3.11.9) → 1 ; 5.28.5 (Python 3.13.16) → 11 ; 6.0.1 → 5 ; 6.1.0 → 0. Le compteur franchit un seuil situé\n", " près d'une frontière, et c'est l'implémentation de l'échantillonneur qui le déplace.\n", "\n", "**Ce qu'il faut en retenir.** Les grandeurs qui décident de la lecture d'un posterior — `r_hat`,\n", "écart des médianes à la vérité — sont **stables dans les quatre environnements** (`sigma_Q`\n", - "0,63-0,66 pour 0,71 ; `sigma_R` 1,446-1,481 pour 2,0 ; drift 0,314-0,318 pour 0,30), alors que\n", + "0,63-0,70 pour 0,71 ; `sigma_R` 1,446-1,481 pour 2,0 ; drift 0,314-0,318 pour 0,30), alors que\n", "le nombre de divergences varie de 0 à 11. Un diagnostic de convergence se juge donc sur ce qui\n", "est **reproductible**, pas sur un compteur dont la valeur change d'une installation à l'autre.\n", "C'est aussi pourquoi la première cellule de code affiche ses versions : un diagnostic séparé de\n", @@ -692,7 +692,7 @@ "**Pourquoi $\\sigma_R$ déraille** : sur une trajectoire de 50 pas, le modèle peut expliquer le\n", "bruit observé **soit** par un capteur bruité ($R$ grand) **soit** par une dynamique agitée\n", "($Q$ grand) — il y a **dégénérescence** entre $Q$ et $R$. La cellule précédente mesure le sens\n", - "des écarts au lieu de le supposer : les deux variances sortent **sous** leur vraie valeur, et\n", + "des écarts au lieu de le supposer — au seuil indicatif de 10 % d’écart relatif, $\\sigma_Q$ est à 9,5 % aujourd’hui (« retrouve ») et bascule en « sous-estime » sur PyMC 5.28.5/Python 3.11 (écart 10,5 %) : les deux variances sortent **sous** leur vraie valeur, et\n", "c'est le **rapport** $R/Q$ qui porte l'erreur ($\\approx 5{,}2$ estimé contre $8{,}0$ vrai, soit\n", "35 %). Le modèle ne surestime donc pas $Q$ pour compenser : il trouve une trajectoire qui\n", "explique les observations avec **moins de bruit total** que la réalité, et c'est cette\n",