Licence 1Algèbre

Calcul matriciel : addition, multiplication, transposition

50 min15 exercicesSéquence 1.1Licence 1

Vidéo disponible dans la version Premium

Durée : 50 min

Calcul matriciel : addition, multiplication, transposition

### 1. Définitions de base

Une matrice AA de taille m×nm \times n (à coefficients réels) est un tableau de mm lignes et nn colonnes :

A=(aij)1im,1jn=(a11a12a1na21a22a2nam1am2amn)A = (a_{ij})_{1 \leq i \leq m,\, 1 \leq j \leq n} = \begin{pmatrix} a_{11} & a_{12} & \cdots & a_{1n} \\ a_{21} & a_{22} & \cdots & a_{2n} \\ \vdots & & & \vdots \\ a_{m1} & a_{m2} & \cdots & a_{mn} \end{pmatrix}

On note Mm,n(R)\mathcal{M}_{m,n}(\mathbb{R}) l'ensemble des matrices m×nm \times n. Si m=nm=n, on parle de matrice carrée d'ordre nn, et on note Mn(R)\mathcal{M}_n(\mathbb{R}).

Cas particuliers : une matrice n×1n \times 1 est un vecteur colonne, une matrice 1×n1 \times n un vecteur ligne. La matrice nulle 0m,n0_{m,n} a tous ses coefficients égaux à 00. La matrice identité InI_n est carrée d'ordre nn, avec des 11 sur la diagonale et des 00 ailleurs : (In)ij=δij(I_n)_{ij} = \delta_{ij} (symbole de Kronecker).

### 2. Addition de matrices et multiplication par un scalaire

Deux matrices ne peuvent s'additionner que si elles ont la même taille. Si A,BMm,n(R)A, B \in \mathcal{M}_{m,n}(\mathbb{R}), la somme A+BA+B est définie coefficient par coefficient :

(A+B)ij=aij+bij(A+B)_{ij} = a_{ij} + b_{ij}

Pour λR\lambda \in \mathbb{R}, la multiplication par un scalaire est (λA)ij=λaij(\lambda A)_{ij} = \lambda \, a_{ij}.

Propriétés (qui font de Mm,n(R)\mathcal{M}_{m,n}(\mathbb{R}) un espace vectoriel) : pour toutes matrices A,B,CA, B, C de même taille et tous scalaires λ,μ\lambda, \mu :
- A+B=B+AA + B = B + A (commutativité) ;
- (A+B)+C=A+(B+C)(A+B)+C = A+(B+C) (associativité) ;
- A+0m,n=AA + 0_{m,n} = A ;
- λ(A+B)=λA+λB\lambda(A+B) = \lambda A + \lambda B et (λ+μ)A=λA+μA(\lambda+\mu)A = \lambda A + \mu A.

Exemple : A=(1203)A = \begin{pmatrix} 1 & 2 \\ 0 & 3 \end{pmatrix}, B=(4121)B = \begin{pmatrix} 4 & -1 \\ 2 & 1 \end{pmatrix}. Alors A+B=(5124)A+B = \begin{pmatrix} 5 & 1 \\ 2 & 4 \end{pmatrix} et 2A=(2406)2A = \begin{pmatrix} 2 & 4 \\ 0 & 6 \end{pmatrix}.

### 3. Produit matriciel

Le produit ABAB n'est défini que si le nombre de colonnes de AA égale le nombre de lignes de BB. Si AMm,n(R)A \in \mathcal{M}_{m,n}(\mathbb{R}) et BMn,p(R)B \in \mathcal{M}_{n,p}(\mathbb{R}), alors ABMm,p(R)AB \in \mathcal{M}_{m,p}(\mathbb{R}) avec :

(AB)ij=k=1naikbkj(AB)_{ij} = \sum_{k=1}^{n} a_{ik} \, b_{kj}

Autrement dit, le coefficient (i,j)(i,j) de ABAB s'obtient en faisant le produit scalaire de la ii-ième ligne de AA avec la jj-ième colonne de BB.

Exemple résolu : A=(1230)A = \begin{pmatrix} 1 & 2 \\ 3 & 0 \end{pmatrix}, B=(2114)B = \begin{pmatrix} 2 & 1 \\ 1 & 4 \end{pmatrix}.

AB=(1×2+2×11×1+2×43×2+0×13×1+0×4)=(4963)AB = \begin{pmatrix} 1\times 2 + 2\times 1 & 1\times 1 + 2\times 4 \\ 3\times 2 + 0\times 1 & 3\times 1 + 0\times 4 \end{pmatrix} = \begin{pmatrix} 4 & 9 \\ 6 & 3 \end{pmatrix}

Propriétés du produit :
- Associativité : (AB)C=A(BC)(AB)C = A(BC) ;
- Distributivité : A(B+C)=AB+ACA(B+C) = AB+AC et (A+B)C=AC+BC(A+B)C = AC+BC ;
- Élément neutre : AIn=InA=AA\, I_n = I_n\, A = A pour AA carrée d'ordre nn ;
- Non-commutativité : en général ABBAAB \neq BA, et l'un des deux produits peut même ne pas être défini si les tailles ne correspondent pas ;
- Présence de diviseurs de zéro : AB=0AB = 0 n'implique pas A=0A=0 ou B=0B=0. Exemple : (0100)(0100)=(0000)\begin{pmatrix} 0 & 1 \\ 0 & 0 \end{pmatrix}\begin{pmatrix} 0 & 1 \\ 0 & 0 \end{pmatrix} = \begin{pmatrix} 0 & 0 \\ 0 & 0 \end{pmatrix} alors qu'aucun des deux facteurs n'est nul.

### 4. Transposition

La transposée ATA^T (ou AA^{\top}) d'une matrice AMm,n(R)A \in \mathcal{M}_{m,n}(\mathbb{R}) est la matrice de Mn,m(R)\mathcal{M}_{n,m}(\mathbb{R}) obtenue en échangeant lignes et colonnes : (AT)ij=aji(A^T)_{ij} = a_{ji}.

Exemple : si A=(123456)A = \begin{pmatrix} 1 & 2 & 3 \\ 4 & 5 & 6 \end{pmatrix}, alors AT=(142536)A^T = \begin{pmatrix} 1 & 4 \\ 2 & 5 \\ 3 & 6 \end{pmatrix}.

Propriétés de la transposition :
- (AT)T=A(A^T)^T = A ;
- (A+B)T=AT+BT(A+B)^T = A^T + B^T ;
- (λA)T=λAT(\lambda A)^T = \lambda A^T ;
- (AB)T=BTAT(AB)^T = B^T A^T (attention à l'inversion de l'ordre des facteurs) ;
- une matrice carrée AA est dite symétrique si AT=AA^T = A, et antisymétrique si AT=AA^T = -A (dans ce cas, sa diagonale est nécessairement nulle).

Exemple résolu (preuve de (AB)T=BTAT(AB)^T = B^TA^T via les coefficients) : Pour AMm,nA \in \mathcal{M}_{m,n}, BMn,pB \in \mathcal{M}_{n,p} :

((AB)T)ji=(AB)ij=k=1naikbkj=k=1n(AT)ki(BT)jk=k=1n(BT)jk(AT)ki=(BTAT)ji\big((AB)^T\big)_{ji} = (AB)_{ij} = \sum_{k=1}^n a_{ik}b_{kj} = \sum_{k=1}^n (A^T)_{ki}(B^T)_{jk} = \sum_{k=1}^n (B^T)_{jk}(A^T)_{ki} = (B^TA^T)_{ji}

Les deux matrices ont donc les mêmes coefficients : (AB)T=BTAT(AB)^T = B^TA^T. \square

### 5. Puissances de matrices carrées

Pour AA carrée d'ordre nn, on définit A0=InA^0 = I_n, A1=AA^1 = A, et Ak+1=AkAA^{k+1} = A^k \cdot A pour k1k \geq 1. Comme le produit matriciel est associatif, les règles usuelles s'appliquent : AkAl=Ak+lA^k A^l = A^{k+l}. Attention : en général (AB)kAkBk(AB)^k \neq A^k B^k si ABBAAB \neq BA, et (A+B)2A2+2AB+B2(A+B)^2 \neq A^2 + 2AB + B^2 sauf si AA et BB commutent.

### 6. Résumé méthodologique

| Opération | Condition de taille | Résultat |
|---|---|---|
| A+BA+B | même taille m×nm \times n | taille m×nm \times n |
| λA\lambda A | aucune (toute matrice) | même taille que AA |
| ABAB | colonnes de AA = lignes de BB | taille (lignes de AA) ×\times (colonnes de BB) |
| ATA^T | aucune | taille n×mn \times m si AA est m×nm \times n |

Exercices

Soit AA une matrice 2×32 \times 3 et BB une matrice 3×43 \times 4. Quelle est la taille de ABAB ?

Calculer A+BA+B pour A=(2103)A = \begin{pmatrix} 2 & -1 \\ 0 & 3 \end{pmatrix} et B=(1452)B = \begin{pmatrix} 1 & 4 \\ 5 & -2 \end{pmatrix}.

Vrai ou faux : pour toutes matrices carrées AA et BB de même taille, AB=BAAB = BA.

Quelle est la matrice identité I2I_2 ?

Soit A=(123456)A = \begin{pmatrix} 1 & 2 & 3 \\ 4 & 5 & 6 \end{pmatrix}. Quelle est la taille de ATA^T ?

Suivez votre progression

Connectez-vous pour sauvegarder votre avancement et gagner des XP.

Se connecter