diff --git a/Chapitre 9 - Produits scalaires et espaces euclidens/9.10-9.11 La meilleure approximation quadratique.ipynb b/Chapitre 9 - Produits scalaires et espaces euclidens/9.10-9.11 La meilleure approximation quadratique.ipynb
index 006c6bc..ebe2504 100644
--- a/Chapitre 9 - Produits scalaires et espaces euclidens/9.10-9.11 La meilleure approximation quadratique.ipynb
+++ b/Chapitre 9 - Produits scalaires et espaces euclidens/9.10-9.11 La meilleure approximation quadratique.ipynb
@@ -1,296 +1,301 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# **Concept(s)-clé(s) et théorie**\n",
"\n",
"## Proposition 1\n",
"Soient $V$ un espace euclidien et $W \\subset V$ un sous-espace vectoriel de $V$. Alors pour tout $x \\in V$ et tout $y \\in W$, on a\n",
"\n",
"\\begin{equation}\n",
"||x - proj_W x|| \\leq ||x-y||\n",
"\\end{equation}\n",
"\n",
"## Définition 1\n",
"Soient $V$ un espace euclidien, $W \\subset V$ un sous-espace vectoriel de $V$ et $x \\in V$; considérez aussi le produit scalaire usuel. Alors le vecteur $proj_Wx$ est appelé la **meilleure approximation quadratique** (ou la **meilleure approximation au sens des moindres carrées**) **de $\\boldsymbol{x}$ par un vecteur dans $\\boldsymbol{W}$**."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Exercises et Examples"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import Librairie.AL_Fct as al\n",
"import Corrections.corrections as corrections\n",
"import numpy as np\n",
"import sympy as sp"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Exercice 1\n",
"\n",
"Soit $V = \\mathbb{R}^n$. Considérez les paires suivantes, faites par un ensemble de vecteurs $\\mathcal{S}$ générant un sous-espace vectoriel $W$ de $V$ et par un élément $v$ de $V$. Calculez la meilleure approximation au sens des moindres carrés de $v$ par un vecteur dans $W$.\n",
"\n",
"1. $V = \\mathbb{R}^2 \\qquad \\mathcal{S} = \\left\\{ \\begin{pmatrix}1 \\\\ -2\\end{pmatrix} \\right\\} \\qquad \\qquad \\quad \\ v = \\begin{pmatrix} -2 \\\\ 1 \\end{pmatrix}$\n",
"2. $V = \\mathbb{R}^3 \\qquad \\mathcal{S} = \\left\\{ \\begin{pmatrix}0 \\\\ 1 \\\\ 0\\end{pmatrix}, \\begin{pmatrix} 1 \\\\ -1 \\\\ 0 \\end{pmatrix} \\right\\} \\qquad \\qquad v = \\begin{pmatrix} -3 \\\\ 2 \\\\ 1 \\end{pmatrix}$\n",
"3. $V = \\mathbb{R}^4 \\qquad \\mathcal{S} = \\left\\{ \\begin{pmatrix}1 \\\\ 2 \\\\ -1 \\\\-2 \\end{pmatrix}, \\begin{pmatrix}0 \\\\ 1 \\\\ 0 \\\\-1 \\end{pmatrix} \\right\\} \\qquad \\quad \\ \\ \\ v = \\begin{pmatrix} 0 \\\\ -1 \\\\ 1 \\\\ -1\\end{pmatrix}$"
]
},
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "### Aide\n",
+ "\n",
+ "Pour calculer la meiileure approximation quadratique de $v$ par un vecteur dans $W$,, il peut être utile de dériver une base ortogonale (ou orthonormée) pour ce dernier. Vous pouvez utiliser la cellule suivante pour exécuter l'algorithme interactif de Gram-Schmidt.\n",
+ "\n",
+ "#### Instructions\n",
+ "\n",
+ "Pour utiliser la méthode interactive de Gram-Schmidt, procédez comme suit:\n",
+ "\n",
+ "1. Insérez le numéro du cas souhaité dans la cellule suivante. Exécutez le cellules appelées \"SÉLECTION DU CAS\" et \"INITIALISATION DES VARIABLES\"\n",
+ "2. Exécutez la cellule appelée \"SÉLECTION DES PARAMÈTRES\" pour sélectionner le type d'opération et les coefficients nécessaires\n",
+ "3. Exécutez la cellule appelée \"EXÉCUTER L'ÉTAPE DE L'ALGORITHME GRAM-SCHMIDT\" pour exécuter l'étape de l'algorithme de Gram-Schmidt avec les paramètres précédemment sélectionnés\n",
+ "4. Répétez les étapes 2 et 3 jusqu'à ce que l'algorithme soit terminée\n",
+ "\n",
+ "En outre:\n",
+ "\n",
+ "1. Vous pouvez annuler une opération en sélectionnant le bouton \"Revert\".\n",
+ "\n",
+ "2. Si les coefficients insérés sont incorrects, vous pouvez essayer avec de nouvelles valeurs sans effectuer une opération \"Revert\".\n",
+ "\n",
+ "3. Les coefficients qui ne sont pas liés à l'opération sélectionnée peuvent être définis sur n'importe quelle valeur, car ils ne sont pas utilisés dans le code."
+ ]
+ },
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
- "case_number = 1"
+ "# SÉLECTION DU CAS\n",
+ "case_number = 1 # CHOISISSEZ LE NUMÉRO DE CAS ICI ET EXECUTEZ LA CELLULE SUIVANTE!"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
+ "# INITIALISATION DES VARIABLES\n",
"if case_number == 1:\n",
" S = [[1,-2]]\n",
" v = [-2,1]\n",
" dim=1\n",
"elif case_number == 2:\n",
" S = [[0,1,0], [1,-1,0]]\n",
" v = [-3,2,1]\n",
" dim=2\n",
"elif case_number == 3:\n",
" S = [[1,2,-1,-2], [0,1,0,-1]]\n",
" v = [0,-1,1,-1]\n",
" dim=2\n",
"else:\n",
" print(f\"{case_number} n'est pas un numéro de cas valide!\" \n",
" f\"Numéros de cas disponibles: [1,2,3]\")\n",
"\n",
"step = 0\n",
"VectorsList = [S]"
]
},
- {
- "cell_type": "markdown",
- "metadata": {},
- "source": [
- "### Aide\n",
- "\n",
- "Pour calculer la meiileure approximation quadratique de $v$ par un vecteur dans $W$,, il peut être utile de dériver une base ortogonale (ou orthonormée) pour ce dernier. Vous pouvez utiliser la cellule suivante pour exécuter l'algorithme interactif de Gram-Schmidt.\n",
- "\n",
- "#### Instructions\n",
- "\n",
- "Pour utiliser la méthode interactive de Gram-Schmidt, procédez comme suit:\n",
- "\n",
- "1. Insérez le numéro de dossier souhaité dans la cellule suivante\n",
- "2. Exécutez la cellule appelée \"SÉLECTION DES PARAMÈTRES\" pour sélectionner le type d'opération et les coefficients nécessaires\n",
- "3. Exécutez la cellule appelée \"EXÉCUTER L'ÉTAPE DE L'ALGORITHME GRAM-SCHMIDT\" pour exécuter l'étape de l'algorithme de Gram-Schmidt avec les paramètres précédemment sélectionnés\n",
- "\n",
- "En outre:\n",
- "\n",
- "1. Vous pouvez annuler une opération en sélectionnant le bouton \"Revert\".\n",
- "\n",
- "2. Si les coefficients insérés sont incorrects, vous pouvez essayer avec de nouvelles valeurs sans effectuer une opération \"Revert\".\n",
- "\n",
- "3. Les coefficients qui ne sont pas liés à l'opération sélectionnée peuvent être définis sur n'importe quelle valeur, car ils ne sont pas utilisés dans le code."
- ]
- },
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# SÉLECTION DES PARAMÈTRES\n",
"norm_coeff, proj_coeffs, operation, step_number = al.manual_GS(dim=dim)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# EXÉCUTER L'ÉTAPE DE L'ALGORITHME GRAM-SCHMIDT\n",
"S = al.interactive_gram_schmidt(norm_coeff, proj_coeffs,\n",
" operation, step_number, \n",
" S.copy(), VectorsList)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# INSÉREZ ICI LE VALEUR DE LA MEILLEURE APPROXIMATION DE v AU SENS DES MOINDRES CARRÉES DANS W\n",
"best_appr = [0, 0] "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"corrections.Ex1Chapitre9_10_11(best_appr, \n",
" case_nb=case_number)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Exercice 2\n",
"\n",
"Soit $V = \\mathcal{C}\\left(I, \\mathbb{R}\\right)$, ou $I$ est un interval dans $\\mathbb{R}$. Considérez les paires suivantes, faites par un ensemble de fonctions $\\mathcal{S}$ générant un sous-espace vectoriel $W$ de $V$ et par un élément $v$ de $V$. Calculez la meilleure approximation au ses des moindres carrés de $v$ par un vecteur dans $W$.\n",
"\n",
"1. $\\quad \\mathcal{S} = \\left\\{ 1, x \\right\\} = \\mathbb{P}^1(\\mathbb{R}) \\qquad \\qquad \\quad v = |x| \\qquad \\qquad \\ I = [-1,1]$\n",
"2. $\\quad \\mathcal{S} = \\left\\{ 1, x, x^2 \\right\\} = \\mathbb{P}^2(\\mathbb{R}) \\qquad \\quad \\ \\ v = |x| \\qquad \\qquad \\ I = [-1,1]$\n",
"3. $\\quad \\mathcal{S} = \\left\\{ 1, x, x^2 \\right\\} = \\mathbb{P}^2(\\mathbb{R}) \\qquad \\quad \\ \\ v = sin(x) \\qquad \\quad I = [-\\pi,\\pi]$\n",
"4. $\\quad \\mathcal{S} = \\left\\{ 1, x, x^2, x^3 \\right\\} = \\mathbb{P}^3(\\mathbb{R}) \\qquad \\ v = e^x \\qquad \\qquad \\ I=[0,1]$"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
- "case_number=1"
+ "# SÉLECTION DU CAS\n",
+ "case_number = 1 # CHOISISSEZ LE NUMÉRO DE CAS ICI ET EXECUTEZ LA CELLULE SUIVANTE!"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
+ "# INITIALISATION DES VARIABLES\n",
"x = sp.Symbol('x')\n",
"if case_number == 1:\n",
" S = [1+0*x, x]\n",
" v = sp.Abs(x)\n",
" int_limits = [-1,1]\n",
" dim=2\n",
"elif case_number == 2:\n",
" S = [1+0*x, x, x**2]\n",
" v = sp.Abs(x)\n",
" int_limits = [-1,1]\n",
" dim=3\n",
"elif case_number == 3:\n",
" S = [1+0*x, x, x**2]\n",
" v = sp.sin(x)\n",
" int_limits = [-np.pi,np.pi]\n",
" dim=3\n",
"elif case_number == 4:\n",
" S = [1, x, x**2, x**3]\n",
" v = sp.exp(x)\n",
" int_limits = [0,1]\n",
" dim=4\n",
"else:\n",
" print(f\"{case_number} n'est pas un numéro de cas valide!\" \n",
" f\"Numéros de cas disponibles: [1,2,3,4]\")\n",
"\n",
"step = 0\n",
"VectorsList = [S]"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Aide\n",
"\n",
"Pour calculer la meiileure approximation quadratique de $v$ par un vecteur dans $W$, il peut\n",
"aider à dériver une base orthogonale (ou orthonormée) pour ce dernier. Vous pouvez utiliser la cellule suivante pour exécuter l'algorithme interactif de Gram-Schmidt pour fonctions."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# SÉLECTION DES PARAMÈTRES\n",
"norm_coeff, proj_coeffs, operation, step_number = al.manual_GS(dim=dim)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# EXÉCUTER L'ÉTAPE DE L'ALGORITHME GRAM-SCHMIDT\n",
"S = al.interactive_gram_schmidt_func(norm_coeff, proj_coeffs,\n",
" operation, step_number, \n",
" S.copy(), VectorsList,\n",
" int_limits=int_limits,\n",
" weight_function=None)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# INSÉREZ ICI LE VALEUR DE LA MEILLEURE APPROXIMATION DE v AU SENS DES MOINDRES CARRÉES DANS W\n",
"best_appr = 1 + 0*x"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"corrections.Ex2Chapitre9_10_11(best_appr, \n",
" int_limits=int_limits, \n",
" case_nb=case_number)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"[Passez au notebook du chapitre 9.12: Solution au sens du moindres carrées](./9.12%20Solution%20au%20sens%20du%20moindres%20carrées.ipynb)"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.7.4"
}
},
"nbformat": 4,
"nbformat_minor": 4
}
diff --git "a/Chapitre 9 - Produits scalaires et espaces euclidens/9.12 Solution au sens du moindres carr\303\251es.ipynb" "b/Chapitre 9 - Produits scalaires et espaces euclidens/9.12 Solution au sens du moindres carr\303\251es.ipynb"
index 061ed98..d8ca310 100644
--- "a/Chapitre 9 - Produits scalaires et espaces euclidens/9.12 Solution au sens du moindres carr\303\251es.ipynb"
+++ "b/Chapitre 9 - Produits scalaires et espaces euclidens/9.12 Solution au sens du moindres carr\303\251es.ipynb"
@@ -1,407 +1,412 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# **Concept(s)-clé(s) et théorie**\n",
"\n",
"## Définition 1\n",
"Soient $A \\in \\mathcal{M}_{m \\times n}(\\mathbb{R})$, $b \\in \\mathcal{M}_{m \\times 1}(\\mathbb{R})$ et $X = \\left(x_1, \\dots, x_n\\right)^T$. Aussi, désignons par $\\phi: \\mathbb{R}^n \\rightarrow \\mathbb{R}^m$ l'application linéaire associée à $A$. Une **solution du système $\\boldsymbol{AX=b}$ au sens du moindres carrées** est une solution du systeme\n",
"\n",
"\\begin{equation}\n",
"AX = proj_{Im(\\phi)}b\n",
"\\end{equation}\n",
"\n",
"## Théorème 1\n",
"Soient $A \\in \\mathcal{M}_{m \\times n}(\\mathbb{R})$, $b \\in \\mathcal{M}_{m \\times 1}(\\mathbb{R})$ et $X = \\left(x_1, \\dots, x_n\\right)^T$. Alors une solution du système $AX=b$ au sens du moindres carrées est une solution du système $A^TAX = A^Tb$."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Exercises et Examples"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import Librairie.AL_Fct as al\n",
"import Corrections.corrections as corrections\n",
"import numpy as np\n",
"import sympy as sp"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Exercice 1\n",
"\n",
"Considérez les systèmes linéaires suivants, avec forme $Ax=b$. Calculez leur solution au sens des moindres carrés en projetant $b$ sur l'espace image de $A$ et en résolvant $Ax = proj_{Im(A)}b$.\n",
"\n",
"1. $\\quad A = \\begin{pmatrix}1 & 0 \\\\ 1 & 0\\end{pmatrix} \\qquad \\qquad b = \\begin{pmatrix}1 \\\\ 3\\end{pmatrix}$\n",
"2. $\\quad A = \\begin{pmatrix}1 & 1 \\\\ 1 & -1 \\\\ 2 & 0\\end{pmatrix} \\qquad \\qquad b = \\begin{pmatrix}1 \\\\ 2 \\\\ -2\\end{pmatrix}$\n",
"3. $\\quad A = \\begin{pmatrix}1 & 0 & 0\\\\ 0 & 1 & 1 \\\\ 1 & 0 & 0\\end{pmatrix} \\qquad \\quad \\ \\ b = \\begin{pmatrix}-1 \\\\ 2 \\\\ 1\\end{pmatrix}$\n",
"4. $\\quad A = \\begin{pmatrix}1 & 0 & 1\\\\ -1 & 1 & -1 \\\\ 0 & 1 & 1 \\\\ 1 & 1 & 0 \\\\ -1 & 0 & 1\\end{pmatrix} \\qquad \\ \\ b = \\begin{pmatrix}0 \\\\ 2 \\\\ 0 \\\\ 1 \\\\ 4\\end{pmatrix}$"
]
},
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "### Aide\n",
+ "\n",
+ "Pour calculer la projection orthogonal de $b$ sur $Im(A)$, il peut être utile de dériver une base ortogonale (ou orthonormée) pour ce dernier. Vous pouvez utiliser la cellule suivante pour exécuter l'algorithme interactif de Gram-Schmidt.\n",
+ "\n",
+ "#### Instructions\n",
+ "\n",
+ "Pour utiliser la méthode interactive de Gram-Schmidt, procédez comme suit:\n",
+ "\n",
+ "1. Insérez le numéro du cas souhaité dans la cellule suivante. Exécutez le cellules appelées \"SÉLECTION DU CAS\" et \"INITIALISATION DES VARIABLES\"\n",
+ "2. Exécutez la cellule appelée \"SÉLECTION DES PARAMÈTRES\" pour sélectionner le type d'opération et les coefficients nécessaires\n",
+ "3. Exécutez la cellule appelée \"EXÉCUTER L'ÉTAPE DE L'ALGORITHME GRAM-SCHMIDT\" pour exécuter l'étape de l'algorithme de Gram-Schmidt avec les paramètres précédemment sélectionnés\n",
+ "4. Répétez les étapes 2 et 3 jusqu'à ce que l'algorithme soit terminée\n",
+ "\n",
+ "En outre:\n",
+ "\n",
+ "1. Vous pouvez annuler une opération en sélectionnant le bouton \"Revert\".\n",
+ "\n",
+ "2. Si les coefficients insérés sont incorrects, vous pouvez essayer avec de nouvelles valeurs sans effectuer une opération \"Revert\".\n",
+ "\n",
+ "3. Les coefficients qui ne sont pas liés à l'opération sélectionnée peuvent être définis sur n'importe quelle valeur, car ils ne sont pas utilisés dans le code."
+ ]
+ },
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
+ "# SÉLECTION DU CAS\n",
"case_number = 1 # CHOISISSEZ LE NUMÉRO DE CAS ICI ET EXECUTEZ LA CELLULE SUIVANTE!"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
+ "# INITIALISATION DES VARIABLES\n",
"if case_number == 1:\n",
" A_cols = [[1,1], [0,0]]\n",
" A = np.array(A_cols).T\n",
" b = [1,3]\n",
" dim=2\n",
"elif case_number == 2:\n",
" A_cols = [[1,1,2], [1,-1,0]]\n",
" A = np.array(A_cols).T\n",
" b = [1,2,-2]\n",
" dim=2\n",
"elif case_number == 3:\n",
" A_cols = [[1,0,1], [0,1,0], [0,1,0]]\n",
" A = np.array(A_cols).T\n",
" b = [-1,2,1]\n",
" dim=3\n",
"elif case_number == 4:\n",
" A_cols = [[1,-1,0,1,-1], [0,1,1,1,0], [1,-1,1,0,1]]\n",
" A = np.array(A_cols).T\n",
" b = [0,2,0,1,4]\n",
" dim=3\n",
"else:\n",
" print(f\"{case_number} n'est pas un numéro de cas valide!\" \n",
" f\"Numéros de cas disponibles: [1,2,3,4]\")\n",
"\n",
"step = 0\n",
"VectorsList = [A_cols]"
]
},
- {
- "cell_type": "markdown",
- "metadata": {},
- "source": [
- "### Aide\n",
- "\n",
- "Pour calculer la projection orthogonal de $b$ sur $Im(A)$, il peut être utile de dériver une base ortogonale (ou orthonormée) pour ce dernier. Vous pouvez utiliser la cellule suivante pour exécuter l'algorithme interactif de Gram-Schmidt.\n",
- "\n",
- "#### Instructions\n",
- "\n",
- "Pour utiliser la méthode interactive de Gram-Schmidt, procédez comme suit:\n",
- "\n",
- "1. Insérez le numéro de dossier souhaité dans la cellule suivante\n",
- "2. Exécutez la cellule appelée \"SÉLECTION DES PARAMÈTRES\" pour sélectionner le type d'opération et les coefficients nécessaires\n",
- "3. Exécutez la cellule appelée \"EXÉCUTER L'ÉTAPE DE L'ALGORITHME GRAM-SCHMIDT\" pour exécuter l'étape de l'algorithme de Gram-Schmidt avec les paramètres précédemment sélectionnés\n",
- "\n",
- "En outre:\n",
- "\n",
- "1. Vous pouvez annuler une opération en sélectionnant le bouton \"Revert\".\n",
- "\n",
- "2. Si les coefficients insérés sont incorrects, vous pouvez essayer avec de nouvelles valeurs sans effectuer une opération \"Revert\".\n",
- "\n",
- "3. Les coefficients qui ne sont pas liés à l'opération sélectionnée peuvent être définis sur n'importe quelle valeur, car ils ne sont pas utilisés dans le code."
- ]
- },
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# SÉLECTION DES PARAMÈTRES\n",
"print(f\"Current vectors: {A_cols}\")\n",
"norm_coeff, proj_coeffs, operation, step_number = al.manual_GS(dim=dim)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# EXÉCUTER L'ÉTAPE DE L'ALGORITHME GRAM-SCHMIDT\n",
"A_cols = al.interactive_gram_schmidt(norm_coeff, proj_coeffs,\n",
" operation, step_number, \n",
" A_cols.copy(), VectorsList)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Aide\n",
"\n",
"Pour résoudre le système linéaire, vous pouvez tirer parti des cellules interactives suivantes qui permettent d'appliquer la méthode d'élimitation de Gauss. Notez que vous devez **entrer la valeur trouvée pour la projection de $\\boldsymbol{b}$ sur l'espace image de $\\boldsymbol{A}$ dans la première ligne!**"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"proj_b = [[0], [2], [0]] # INSEREZ ICI LA VALEUR TROUVÉE POUR LA PROJECTION DE b SUR Im (A)\n",
"al.printA(A, proj_b)\n",
"[i,j,r,alpha]= al.manualEch(A,proj_b)\n",
"m=np.concatenate((A,proj_b), axis=1)\n",
"MatriceList=[A]\n",
"RhSList=[proj_b]"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"m=al.echelonnage(i,j,r,alpha,A,m,MatriceList,RhSList)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# INSEREZ ICI LA SOLUTION\n",
"x,y,z = sp.symbols('x, y, z')\n",
"sol = sp.sets.FiniteSet((0,y,2-y))"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# ÉVALUATION DE LA SOLUTION\n",
"corrections.Ex1_Chapitre9_12(sol, case_nb=case_number)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Exercice 2\n",
"\n",
"Considérez les systèmes linéaires suivants, avec forme $Ax = b$. Calculez leur solution au sens des moindres carrés en résolvant le système linéaire $A^TAx=A^Tb$.\n",
"\n",
"1. $\\quad A = \\begin{pmatrix}0 & 1 \\\\ 0 & -1\\end{pmatrix} \\qquad \\qquad b = \\begin{pmatrix}0 \\\\ 2\\end{pmatrix}$\n",
"2. $\\quad A = \\begin{pmatrix}0 & 1 \\\\ 1 & 1 \\\\ 1 & -2\\end{pmatrix} \\qquad \\qquad \\ \\ b = \\begin{pmatrix}1 \\\\ 1 \\\\ 0\\end{pmatrix}$\n",
"3. $\\quad A = \\begin{pmatrix}0 & 1 & 0\\\\ 1 & 0 & 1 \\\\ -2 & 0 & 0\\end{pmatrix} \\qquad \\quad b = \\begin{pmatrix}0 \\\\ 2 \\\\ -1\\end{pmatrix}$\n",
"4. $\\quad A = \\begin{pmatrix}1 & 0 & 0\\\\ 1 & -1 & -1 \\\\ 0 & 1 & 0 \\\\ 1 & 0 & 0 \\\\ 0 & 1 & -1\\end{pmatrix} \\qquad \\ \\ b = \\begin{pmatrix}2 \\\\ 0 \\\\ 0 \\\\ -1 \\\\ -1\\end{pmatrix}$"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
+ "# SÉLECTION DU CAS\n",
"case_number = 1 # CHOISISSEZ LE NUMÉRO DE CAS ICI ET EXECUTEZ LA CELLULE SUIVANTE!"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
+ "# INITIALISATION DES VARIABLES\n",
"if case_number == 1:\n",
" A_cols = [[0,0], [1,-1]]\n",
" A = np.array(A_cols).T\n",
" b = [[0], [2]]\n",
"elif case_number == 2:\n",
" A_cols = [[0,1,1], [1,1,-2]]\n",
" A = np.array(A_cols).T\n",
" b = [[1], [1], [0]]\n",
"elif case_number == 3:\n",
" A_cols = [[0,1,-2], [1,0,0], [0,1,0]]\n",
" A = np.array(A_cols).T\n",
" b = [[0], [2], [-1]]\n",
"elif case_number == 4:\n",
" A_cols = [[1,1,0,1,0], [0,-1,1,0,1], [0,-1,0,0,-1]]\n",
" A = np.array(A_cols).T\n",
" b = [2,0,0,-1,-1]\n",
"else:\n",
" print(f\"{case_number} n'est pas un numéro de cas valide!\" \n",
" f\"Numéros de cas disponibles: [1,2,3,4]\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Aide\n",
"\n",
"Pour résoudre le système linéaire, vous pouvez tirer parti des cellules interactives suivantes qui permettent d'appliquer la méthode d'élimitation de Gauss."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"al.printA(A.T@A, A.T@np.array(b))\n",
"[i,j,r,alpha]= al.manualEch(A.T@A, A.T@np.array(b))\n",
"m=np.concatenate((A.T@A,A.T@np.array(b)), axis=1)\n",
"MatriceList=[A.T@A]\n",
"RhSList=[A.T@np.array(b)]"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"m=al.echelonnage(i,j,r,alpha,A.T@A,m,MatriceList,RhSList)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# INSEREZ ICI LA SOLUTION\n",
"x,y,z = sp.symbols('x, y, z')\n",
"sol = sp.sets.FiniteSet((x,-1)) "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# ÉVALUATION DE LA SOLUTION\n",
"corrections.Ex2_Chapitre9_12(sol, case_nb=case_number)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Exercice 3\n",
"\n",
"Un scénario important dans lequel le calcul des solutions de systèmes linéaires au sens des moindres carrés est utilisé est celui de la **régression linéaire**. Supposons que l'on donne $N$ mesures de certaines quantités $\\left\\{\\left\\{x_j^i\\right\\}_{j=1}^K, y^i\\right\\}_{i=1}^N$; les variables $x$ sont appelées variables explicatives et $K$ est leur nombre. Par exemple, si $K=1$, ils peuvent être le poids ($x_1$) et la taille ($y$) de $N$ personnes différentes; si $K=2$ ils peuvent être le poids ($x_1$), la longueur des pieds ($x_2$) et la taille ($y$) de $N$ personnes différentes.\n",
"\n",
"L'objectif est de déterminer la meilleure relation linéaire possible entre ces grandeurs, c'est-à-dire de déterminer les meilleures valeurs possibles pour les coefficients $\\left\\{c_j\\right\\}_{j=0}^K$. En termes mathématiques, cela revient à résoudre le problème de minimisation suivant $$ \\left\\{c_j\\right\\}_{j=0}^K = \\underset{\\tilde{c}_0, \\dots, \\tilde{c}_k}{argmin} \\sum\\limits_{i=1}^N \\left(y^i - \\tilde{c}_0 - \\sum\\limits_{j=1}^K \\tilde{c}_j x_j^i \\right)^2$$ \n",
"\n",
"S'il existe une relation linéaire entre les données, alors $$y^i = c_0 + \\sum\\limits_{j=1}^K c_j x_j^i \\quad \\forall \\ i \\in \\{1, \\dots, N\\}$$ Cela implique que les coefficients $\\left\\{c_j\\right\\}_{j=0}^K$ sont des solutions au système linéaire suivant:\n",
"\n",
"\\begin{equation}\n",
"\\begin{pmatrix}\n",
"1 & x_1^1 & x_2^1 & \\dots & x_K^1 \\\\\n",
"1 & x_1^2 & x_2^2 & \\dots & x_K^2 \\\\ \n",
"\\vdots & \\vdots & \\vdots & \\ddots & \\vdots \\\\\n",
"1 & x_1^N & x_2^N & \\dots & x_K^N\n",
"\\end{pmatrix}\n",
"\\begin{pmatrix}\n",
"c_0 \\\\ c_1 \\\\ c_2 \\\\ \\vdots \\\\ c_K\n",
"\\end{pmatrix} = \n",
"\\begin{pmatrix}\n",
"y^1 \\\\ y^2 \\\\ \\vdots \\\\ y^N\n",
"\\end{pmatrix}\n",
"\\end{equation}\n",
"\n",
"Quoi qu'il en soit, comme $N$ dans les applications du monde réel est beaucoup plus grand que $K$, il est très probable que ce système n'admette aucune solution. Ainsi, c'est une approche commun de recourir au calcul d'une solution au sens des moindres carrés; on peut en fait prouver que la solution au sens des moindres carrés est égal à la solution au problème de minimisation quadratique susmentionné, dont dérive le nom de \"moindres carrés\".\n",
"\n",
"### Instructions\n",
"e but de l'exercice est de montrer un scénario de cas réel où des solutions des moindres carrés aux systèmes linéaires sont employées; ainsi aucun calcul numérique n'est requis et peu de quantités doivent être correctement insérées.\n",
"\n",
"1. **GÉNÉRATION DE DONNÉES**: la cellule appelée \"GÉNÉRATION DE DONNÉES\" est responsable de la génération des données. En particulier, la méthode \"Ex3_Chapitre9_12_generate_data\" génère les données en superposant du bruit gaussien blanc à des données linéairement dépendantes. Les deux premiers arguments d'entrée régulent l'intensité du bruit. L'argument d'entrée appelée \"K\" définit le nombre de variables explicatives; les seules valeurs disponibles sont K = 1 et K = 2, de sorte que les données peuvent être visualisées via des nuages de points. Essayez les deux! Finalement, les variables X et Y stockent la matrice de gauche et le vecteur de droite du système linéaire précédemment introduit.\n",
"\n",
"2. **INSERTION DE SOLUTION**: la cellule appelée \"INSERTION DE SOLUTION\" vous permet de saisir les valeurs de la matrice de gauche M et du vecteur de droite f, définissant le système linéaire à résoudre afin de calculer la solution souhaitée. Dans ce but, nous rappelons que, étant donné deux matrices A, B:\n",
" * A.T $\\rightarrow$ calcule la transposée de A\n",
" * A @ B $\\rightarrow$ calcule le produit matriciel entre A et B\n",
" * A * B $\\rightarrow$ calcule le produit élément par élément entre A et B\n",
"\n",
"\n",
"3. **VISUALISATION DE LA SOLUTION**: la cellule appelée \"VISUALISATION DE LA SOLUTION\" vous permet de visualiser la solution au problème donnée."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# GÉNÉRATION DE DONNÉES\n",
"data, fig = corrections.Ex3_Chapitre9_12_generate_data(0.15, 0.075, K=2)\n",
"X = data[0]\n",
"Y = data[1]"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# INSERTION DE SOLUTION\n",
"M = X.T @ X # !! inserez ici la matrice !!\n",
"f = X.T @ Y # !! inserez ici le vecteur de droit !!\n",
"sys = M, f"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"scrolled": false
},
"outputs": [],
"source": [
"# VISUALISATION DE LA SOLUTION\n",
"corrections.Ex3_Chapitre9_12(sys, data, fig)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"[Passez au notebook du chapitre 9.13-9.14: La factorisation QR: application à la résolution d'un système au sens du moindres carrées](./9.13-9.14%20La%20factorisation%20QR%20-%20application%20à%20la%20résolution%20d'un%20système%20au%20sens%20des%20moindres%20carrées.ipynb)"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.7.4"
}
},
"nbformat": 4,
"nbformat_minor": 4
}
diff --git "a/Chapitre 9 - Produits scalaires et espaces euclidens/9.2 - Produit scalaires, d\303\251finitions, exemples.ipynb" "b/Chapitre 9 - Produits scalaires et espaces euclidens/9.2 - Produit scalaires, d\303\251finitions, exemples.ipynb"
index a5faaea..4dfbaa5 100644
--- "a/Chapitre 9 - Produits scalaires et espaces euclidens/9.2 - Produit scalaires, d\303\251finitions, exemples.ipynb"
+++ "b/Chapitre 9 - Produits scalaires et espaces euclidens/9.2 - Produit scalaires, d\303\251finitions, exemples.ipynb"
@@ -1,253 +1,193 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# **Concept(s)-clé(s) et théorie**\n",
"\n",
"## Définition 1 - Produit Scalaire\n",
"Soit $V$ un $\\mathbb{R}$-espace vectoriel. Un **produit scalaire** sur $V$ est une application qui fait correspondre à chaque paire ordonnée $(u,v) \\in V \\times V$ un nombre réel, noté $\\langle u, v \\rangle \\in \\mathbb{R}$, telle que les conditions suivantes soient vérifiées, pour tous $u,v,w \\in V, \\alpha \\in \\mathbb{R}$:\n",
"\n",
"1. *Symmétrie*: $\\langle u,v \\rangle = \\langle v, u \\rangle$\n",
"2. *Additivité*: $\\langle u+v, w \\rangle = \\langle u,w \\rangle + \\langle v,w \\rangle$\n",
"3. *Bilinearité (combinè avec 2)*: $\\langle \\alpha u, v \\rangle = \\alpha \\langle u,v \\rangle = \\langle u, \\alpha v \\rangle$\n",
"4. *Definié Positivité*: $\\langle u,u \\rangle \\geq 0 \\ \\forall u \\in V$ et si $\\langle u,u \\rangle = 0$ alors $u=0$.\n",
"\n",
"## Définition 2 - Espace Euclidien\n",
"Un $\\mathbb{R}$-espace vectoriel *de dimension finie* muni d'un produit scalaire s'appelle un **espace euclidien**.\n",
"\n",
"## Définition 3 - Orthogonalité\n",
"Soiet $V$ un $\\mathbb{R}$-espace vectoriel muni d'un produit scalaire $\\langle \\cdot,\\cdot \\rangle$ et $u,v \\in V$. On dit que $u$ et $v$ sont **orthogonaux** si $\\langle u,v \\rangle = 0$. \n",
"\n",
"## Example 1 \n",
"Un example de produit scalaire dans $V = \\mathcal{M}_{n \\times n}(\\mathbb{R})$ est: $$ \\langle A,B \\rangle = Trace(A^TB)$$ ou la trace d'un matrice carée de dimension $n$ est definie comme suit: $$Trace(A) = \\sum\\limits_{i=1}^n a_{ii}$$\n",
"\n",
"## Example 2\n",
"Un example de produit scalaire dans $V = \\mathcal{C}^0([a;b], \\mathbb{R}) =: \\{f: [a;b] \\rightarrow \\mathbb{R} : f \\ fonction \\ continue\\}$ (avec $[a;b]$ un intervalle de $\\mathbb{R}$) est: $$ \\langle f,g \\rangle = \\int_a^b f(x)g(x) \\ dx$$ où $\\int_a^b f(x) \\ dx$ désigne l'intégrale de Riemann de $f$ dans l'intervalle $[a;b]$."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Exercises et Exemples"
]
},
{
"cell_type": "code",
- "execution_count": 1,
+ "execution_count": null,
"metadata": {},
- "outputs": [
- {
- "data": {
- "text/html": [
- " \n",
- " "
- ]
- },
- "metadata": {},
- "output_type": "display_data"
- }
- ],
+ "outputs": [],
"source": [
"import Librairie.AL_Fct as al\n",
"import Corrections.corrections as corrections\n",
"import numpy as np\n",
"import matplotlib.pyplot as plt"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Exercise 1\n",
"Considérez les couples de $\\mathbb{R}$-espaces vectoriels et d'opérateurs suivants et marquez ceux des déclarations suivantes qui sont corrects.\n",
"\n",
"1. $\\big(\\mathbf{V}, \\langle a, b \\rangle\\big) = \\big(\\mathbb{R}^3, \\ a_1b_1 - a_1b_2 - a_2b_1 + a_2b_2 - a_2b_3 - a_3b_2 + 2a_3b_3\\big)$\n",
"2. $\\big(\\mathbf{V}, \\langle a, b \\rangle\\big) = \\big(\\mathbb{R}^3, \\ a^TMb\\big) \\qquad$ with $M = \\begin{pmatrix} 2 & 0 & 1\\\\ 0 & 2 & -1 \\\\ -1 & -1 & 2 \\end{pmatrix}$\n",
"3. $\\big(\\mathbf{V}, \\langle a, b\\rangle\\big) = \\big(\\mathbb{P}^2(\\mathbb{R}), \\ 2c^a_0c^b_0 - c^a_0c^b_2 + c^a_1c^b_1 - c^a_2c^b_0 + c^a_2c^b_2\\big) \\qquad$ with $a(x) =: c^a_2 x^2 + c^a_1 x + c^a_0 \\ $ and $ \\ b(x) =: c^b_2x^2 + c^b_1 x + c^b_0$\n",
"4. $\\big(\\mathbf{V}, \\langle a, b\\rangle\\big) = \\big(\\mathbb{P}^3(\\mathbb{R}), \\ 2c^a_0c^b_0 - c^a_0c^b_2 + c^a_1c^b_1 - c^a_2c^b_0 + c^a_2c^b_2\\big) \\qquad$ with $a(x) =: c^a_3 x^3 + c^a_2 x^2 + c^a_1 x + c^a_0 \\ $ and $ \\ b(x) =: c^a_3 x^3 + c^b_2x^2 + c^b_1 x + c^b_0$\n",
"5. $\\big(\\mathbf{V}, \\langle a, b\\rangle\\big) = \\big(\\mathcal{C}^1([x_0, x_1]; \\mathbb{R}), \\ a(x_0)b(x_0) + \\int_{x_0}^{x_1} a'(x)b'(x) \\ dx \\big)$\n",
"6. $\\big(\\mathbf{V}, \\langle a, b\\rangle\\big) = \\big(\\mathcal{C}^2([x_0, x_1]; \\mathbb{R}), \\ a(x_0)b(x_0) + \\int_{x_0}^{x_1} a''(x)b''(x) \\ dx \\big)$\n",
"\n",
"### Remarques\n",
"- $\\mathbb{P}^n(\\mathbb{R})$ désigne l'ensemble des polynômes a valeurs réelles de degré au plus $n$, qui peuvent alors être exprimés de manière unique en termes de $n+1$ coefficients scalaires\n",
"- $\\mathcal{C}^n([x_0, x_1]; \\mathbb{R})$ désigne l'ensemble des fonctions à valeurs réelles qui sont $n$-fois différenciables, avec toutes les dérivées continues, sur l'intervalle $[x_0; x_1]$"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"corrections.Ex1Chapitre9_2()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Exercise 2\n",
"Étant donné les couples de matrices suivants, déterminez celles qui sont orthogonales par rapport au produit scalaire défini via l'opérateur de trace\n",
"\n",
"1. $ \\qquad A = \\begin{pmatrix} 1 & 2 & 0 \\\\ 0 & -1 & -1 \\\\ 1 & 3 & 1 \\end{pmatrix} \\qquad \\quad \\ \\ \n",
" B = \\begin{pmatrix} 1 & -1 & 4 \\\\ 3 & 1 & -2 \\\\ 1 & 0 & -1 \\end{pmatrix}$ \n",
"2. $ \\qquad A = \\begin{pmatrix} 0 & 2 \\\\ -1 & 3 \\end{pmatrix} \\qquad \\qquad \\quad\n",
" B = \\begin{pmatrix} 3 & 1 \\\\ 1 & -1 \\end{pmatrix}$\n",
"3. $ \\qquad A = \\begin{pmatrix} 0 & 1 & 3 & 0 \\\\ 1 & 0 & 1 & 0 \\\\ -1 & -2 & 2 & 1 \\\\ 3 & 4 & 1 & 2 \\end{pmatrix} \\qquad\n",
" B = \\begin{pmatrix} 3 & 1 & -1 & 2 \\\\ 2 & 2 & 0 & 1 \\\\ -1 & 1 & -1 & 3 \\\\ -1 & 1 & 1 & -1 \\end{pmatrix}$\n",
"4. $ \\qquad A = \\begin{pmatrix} 1 & -3 \\\\ 2 & 1 \\end{pmatrix} \\qquad \\qquad \\quad\n",
" B = \\begin{pmatrix} 1 & -2 \\\\ -2 & 1 \\end{pmatrix}$"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"corrections.Ex2Chapitre9_2()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Exercise 3\n",
"Étant donné les couples de fonctions suivants, déterminez s'ils sont orthogonaux par rapport au produit scalaire donné\n",
"\n",
"1. $\\qquad f(x) = 1 - x^2; \\quad g(x) = -(x-1)^2 \\qquad \\quad with: \\quad \\langle f, g \\rangle = c^f_0c^g_0 + c^f_1c^g_1 + c^f_2c^g_2$\n",
"2. $\\qquad f(x) = 1 - x^2; \\quad g(x) = -(x-1)^2 \\qquad \\quad with: \\quad \\langle f, g \\rangle = \\int_{-1}^{1} f(x)g(x) \\ dx$\n",
"3. $\\qquad f(x) = x - \\dfrac{1}{2}; \\quad g(x) = x^2 - x - \\dfrac{1}{6} \\qquad \\ \\ with: \\quad \\langle f, g \\rangle = 2c^f_0c^g_0 - c^f_0c^g_0 + 2c^f_1c^g_1 - c^f_1c^g_2 - c^f_2c^g_0 - c^f_2c^g_1 + 2c^f_2c^g_2$\n",
"4. $\\qquad f(x) = x - \\dfrac{1}{2}; \\quad g(x) = x^2 - x - \\dfrac{1}{6} \\qquad \\ \\ with: \\quad \\langle f, g \\rangle = \\int_0^1 f(x)g(x) \\ dx$\n",
"5. $\\qquad f(x) = \\sin(x); \\quad \\ g(x) = \\cos(x) \\qquad \\qquad \\ \\ with: \\quad \\langle f,g \\rangle = \\int_{-\\pi}^{\\pi} f(x)g(x) \\ dx$\n",
"6. $\\qquad f(x) = \\sin(x); \\quad \\ g(x) = \\cos(x) \\qquad \\qquad \\ \\ with: \\quad \\langle f,g \\rangle = \\int_{0}^{\\pi / 2} f(x)g(x) \\ dx$"
]
},
{
"cell_type": "code",
- "execution_count": 2,
+ "execution_count": null,
"metadata": {},
- "outputs": [
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "Cliquer sur CTRL pour sélectionner plusieurs réponses\n"
- ]
- },
- {
- "data": {
- "application/vnd.jupyter.widget-view+json": {
- "model_id": "daffcdc93f9841568bb81cbcff6cbcad",
- "version_major": 2,
- "version_minor": 0
- },
- "text/plain": [
- "interactive(children=(SelectMultiple(description='Les fonctions sont orthogonales dans les cas:', layout=Layou…"
- ]
- },
- "metadata": {},
- "output_type": "display_data"
- }
- ],
+ "outputs": [],
"source": [
"corrections.Ex3Chapitre9_2()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"**Aide**: Vous pouvez vous aider en exécutant les cellules suivantes, qui permettent de tracer les 3 couples de fonctions considérés dans l'exercice 3"
]
},
{
"cell_type": "code",
- "execution_count": 3,
+ "execution_count": null,
"metadata": {},
- "outputs": [
- {
- "data": {
- "image/png": "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\n",
- "text/plain": [
- "