Fiche de mathématiques
> >

Matrices (II)

Partager :
Ici, \mathbb{K} désigne un corps commutatif, en général: \mathbb{K}=\mathbb{R} ou \mathbb{C}.

I- Décomposition en blocs

Dans toute cette partie, soient :
n,p,q\in\mathbb{N}^{*} , A=(a_{ij})_{ij}\in\mathcal{M}_{n,p}(\mathbb{K}).
l,m\in\mathbb{N}^{*} ,(n_{1},\ldots,n_{l})\in(\mathbb{N}^{*})^{l} , (p_{1},\ldots,p_{m})\in (\mathbb{N}^{*})^{m} tels que \displaystyle \sum_{t=1}^{l} n_{t}=n et \displaystyle \sum_{t=1}^{m} p_{t}=p.
n_{0}=p_{0}=0
\gamma_{k}=\displaystyle\sum_{i=0}^{k}n_{i} , \forall k\in\lbrace 0,\ldots,l\rbrace
\theta_{r}=\displaystyle\sum_{j=0}^{r}p_{j} , \forall r\in\lbrace 0,\ldots,m\rbrace

1- Définition

Définition :
La matrice A est dite écrite par bloc (ou partitionnée) si A est écrite "divisée" d'une manière cohérente en matrices rectangulaires de dimensions inférieures appelées blocs, on écrit:
A=\begin{pmatrix} B_{11}&B_{12}&\ldots&B_{1m}\\B_{21}&B_{22}&\ldots&B_{2m}\\\vdots&\vdots&&\vdots\\B_{l1}&B_{l2}&\ldots&B_{lm}\end{pmatrix}

tel que: pour (k,r)\in\lbrace 1,\ldots,l\rbrace\times\lbrace 1,\ldots,m\rbrace, la matrice B_{kr}=\begin{pmatrix} a_{\gamma_{k-1}+1\theta_{r-1}+1} &\ldots&a_{\gamma_{k-1}+1\theta_{r}}\\ \vdots &&\vdots\\a_{\gamma_{k}\theta_{r-1}+1}&\ldots&a_{\gamma_{k}\theta_{r}}\end{pmatrix} \in\mathcal{M}_{n_{k}p_{r}}(\mathbb{K}) appelée: le (k,r)^{\text{ème}} bloc dans la décomposition de A en blocs suivant le découpage (n_{1},\ldots,n_{l}) pour les lignes et (p_{1},\ldots,p_{m}) pour les colonnes.


Exemples :
\begin{pmatrix} x \\X\end{pmatrix} \in\mathcal{M}_{n+1,1}(\mathbb{K}), pour x\in\mathbb{K} et X\in\mathcal{M}_{n,1}(\mathbb{K})
\begin{pmatrix}A&B\\C&D\end{pmatrix} \in\mathcal{M}_{n+p}(\mathbb{K}), pour A\in\mathcal{M}_{n}(\mathbb{K}), B\in\mathcal{M}_{n,p}(\mathbb{K}), C\in\mathcal{M}_{p,n}(\mathbb{K}) et D\in\mathcal{M}_{p}(\mathbb{K})

Remarque :
Si A est une matrice carrée, nous n'utiliserons que les décompositions en blocs pour lesquelles l=m et (n_{1},\ldots,n_{l})=(p_{1},\ldots,p_{m}). Et dans ce cas, les blocs sont appelés les blocs diagonaux de la décomposition de A en blocs, ce sont des matrices carrées.

2- Opérations par blocs

Proposition : (Addition et multiplication par scalaire)
Soient \lambda\in\mathbb{K}, A,B\in\mathcal{M}_{n,p}(\mathbb{K}).
Si A et B sont décomposées par blocs avec le même découpage, alors \lambda A + B admet la décomposition en blocs avec le même découpage avec:
\lambda A+B=\lambda\begin{pmatrix}A_{11}&\ldots&A_{1m}\\\vdots&&\vdots\\ A_{l1}&\ldots&A_{lm}\end{pmatrix}+\begin{pmatrix}B_{11}&\ldots&B_{1m}\\\vdots&&\vdots\\ B_{l1}&\ldots&B_{lm}\end{pmatrix}=\begin{pmatrix}\lambda A_{11}+B_{11}&\ldots&\lambda A_{1m}+B_{1m}\\\vdots&&\vdots\\ \lambda A_{l1}+B_{l1}&\ldots&\lambda A_{lm}+B_{lm}\end{pmatrix}


Exemple :
Soient A,A^{'}\in\mathcal{M}_{n}(\mathbb{K}), B,B^{'}\in\mathcal{M}_{n,p}(\mathbb{K}), C,C^{'}\in\mathcal{M}_{p,n}(\mathbb{K}) et D,D^{'}\in\mathcal{M}_{p}(\mathbb{K}) :
\begin{pmatrix}A&B\\C&D\end{pmatrix} +\begin{pmatrix}A^{'}&B^{'}\\C^{'}&D^{'}\end{pmatrix} =\begin{pmatrix}A+A^{'}&B+B^{'}\\C+C^{'}&D+D^{'}\end{pmatrix}

Proposition : (Produit)
Soient A\in\mathcal{M}_{n,p}(\mathbb{K}) , B\in\mathcal{M}_{p,q}(\mathbb{K}) telles que: A=\begin{pmatrix}A_{11}&\ldots&A_{1m}\\\vdots&&\vdots\\ A_{l1}&\ldots&A_{lm}\end{pmatrix} et B=\begin{pmatrix}B_{11}&\ldots&B_{1m^{'}}\\\vdots&&\vdots\\ B_{l^{'}1}&\ldots&B_{l^{'}m^{'}}\end{pmatrix} des décompositions en blocs de A et B telles que:
l^{'}=m et (n_{1}^{'},\ldots,n_{l^{'}}^{'})=(p_{1},\ldots,p_{m})

Alors AB admet la décomposition en blocs: AB= \begin{pmatrix}\displaystyle\sum_{j=1}^{l^{'}} A_{1j}B_{j1} & \ldots&\displaystyle\sum_{j=1}^{l^{'}} A_{1j}B_{jm^{'}}\\\vdots&&\vdots\\\displaystyle\sum_{j=1}^{l^{'}} A_{lj}B_{j1}&\ldots&\displaystyle\sum_{j=1}^{l^{'}} A_{lj}B_{jm^{'}}\end{pmatrix}


Exemple :
Soient: A,B,C,D,A^{'},B^{'},C^{'},D^{'} \in\mathcal{M}_{n}(\mathbb{K})
\begin{pmatrix}A&B\\C&D\end{pmatrix}\begin{pmatrix}A^{'}&B^{'}\\C^{'}&D^{'}\end{pmatrix}=\begin{pmatrix}AA^{'}+BC^{'}&AB^{'}+BD^{'}\\CA^{'}+DC^{'}&CB^{'}+DD^{'}\end{pmatrix}\in\mathcal{M}_{2n}(\mathbb{K})

Proposition: (Transposition)
Soit A\in\mathcal{M}_{n,p}(\mathbb{K}) telle que: A=\begin{pmatrix}A_{11}&\ldots&A_{1m}\\\vdots&&\vdots\\ A_{l1}&\ldots&A_{lm}\end{pmatrix} une décomposition en blocs de cette dernière.
On a: ^{t}A=\begin{pmatrix}^{t}A_{11}&\ldots&^{t}A_{l1}\\\vdots&&\vdots\\ ^{t}A_{1m}&\ldots&^{'}A_{lm}\end{pmatrix}


Exemple :
Soient : A,B,C,D \in\mathcal{M}_{n}(\mathbb{K})
^{t}\begin{pmatrix}A&B\\C&D\end{pmatrix}=\begin{pmatrix}^{t}A&^{t}C\\^{t}B&^{t}D\end{pmatrix}

3- Matrices triangulaires par blocs

Définitions:
Une matrice carrée A est dite triangulaire supérieure par blocs si et seulement si elle admet une décomposition en blocs: A=\begin{pmatrix} A_{11}&\ldots&A_{1l}\\&\ddots&\vdots\\O&&A_{ll}\end{pmatrix} telle que A_{11},\ldots,A_{ll} sont des matrices carrées et que les blocs sous la diagonale sont tous des matrices nulles.
Une matrice carrée A est dite triangulaire inférieure par blocs si et seulement si elle admet une décomposition en blocs: A=\begin{pmatrix} A_{11}&&O\\\vdots&\ddots&\\A_{l1}&\ldots&A_{ll}\end{pmatrix} telle que A_{11},\ldots,A_{ll} sont des matrices carrées et que les blocs en dessus de la diagonale sont tous des matrices nulles.
Une matrice carrée A est dite diagonale par blocs si et seulement si elle admet une décomposition en blocs: A=\begin{pmatrix} A_{11}&&O\\&\ddots&\\O&&A_{ll}\end{pmatrix} telle que A_{11},\ldots,A_{ll} sont des matrices carrées et que les blocs non diagonaux sont tous des matrices nulles. Dans ce cas, on peut noter: A=diag(A_{11},\ldots,A_{ll})



Proposition:
Le détermiant d'une matrice triangulaire (inférieure ou supérieure) par blocs est égal au produit des déterminant des blocs diagonaux.




II- Exponentielle de matrice

Prérequis : Séries d'applications et espaces vectoriels normés
Soit n\in\mathbb{N}^{*}

Proposition :
La série d'applications \displaystyle \sum_{k\geq 0}(\begin{array}{clcl}  &\mathcal{M}_{n}(\mathbb{K})  &\rightarrow &\mathcal{M}_{n}(\mathbb{K})\\ &A&\mapsto     &\dfrac{1}{k!}A^{k}\\\end{array}) converge normalement sur toute partie bornée de \mathcal{M}_{n}(\mathbb{K}).


Preuve :
On sait qu'il existe au moins une norme ||.|| d'algèbre sur \mathcal{M}_{n}(\mathbb{K}), il est simple de vérifier (récurrence immédiate) que: \forall A \in \mathcal{M}_{n}(\mathbb{K}), \forall k\in\mathbb{N}^{*} : ||A^{k}||\leq ||A||^{k}
Soit X une partie bornée de \mathcal{M}_{n}(\mathbb{K}), il existe M\in\mathbb{R}_{+} tel que: \forall A\in X : ||A||\leq M
On a alors :
\forall k\in\mathbb{N}^{*}, \forall A\in X : ||\frac{1}{k!} A^{k}||\leq \frac{1}{k!}||A||^{k}\leq \frac{M^{k}}{k!}
Et puisque la série numérique \displaystyle\sum_{k\geq 0} \frac{M^{k}}{k!} converge, la série d'applications \displaystyle \sum_{k\geq 0}(\begin{array}{clcl}  &\mathcal{M}_{n}(\mathbb{K})  &\rightarrow &\mathcal{M}_{n}(\mathbb{K})\\ &A&\mapsto     &\frac{1}{k!}A^{k}\\\end{array}) converge normalement sur X.

Définition :
On appelle exponentielle, et on note exp, l'application de \mathcal{M}_{n}(\mathbb{K}) dans \mathcal{M}_{n}(\mathbb{K}) définie par :
\forall A\in \mathcal{M}_{n}(\mathbb{K}) : exp(A)=\displaystyle\sum_{k=0}^{+\infty} \frac{1}{k!}A^{k}


Proposition :
\forall A,B\in\mathcal{M}_{n}(\mathbb{K}) : AB=BA\Longrightarrow exp(A+B)=exp(A)exp(B)=exp(B)exp(A).
\forall A\in \mathcal{M}_{n}(\mathbb{K}) : exp(A)\in \mathcal{GL}_{n}(\mathbb{K}) et (exp(A))^{-1}=exp(-A).
\forall A\in \mathcal{M}_{n}(\mathbb{K}) , \forall P\in \mathcal{GL}_{n}(\mathbb{K}) : exp(P^{-1}AP)=P^{-1}exp(A)P.

Publié le
ceci n'est qu'un extrait
Pour visualiser la totalité des cours vous devez vous inscrire / connecter (GRATUIT)
Inscription Gratuite se connecter
Merci à
Panter Correcteur
pour avoir contribué à l'élaboration de cette fiche


Vous devez être membre accéder à ce service...

Pas encore inscrit ?

1 compte par personne, multi-compte interdit !

Ou identifiez-vous :


Rester sur la page

Inscription gratuite

Fiches en rapport

parmi 1674 fiches de maths

Désolé, votre version d'Internet Explorer est plus que périmée ! Merci de le mettre à jour ou de télécharger Firefox ou Google Chrome pour utiliser le site. Votre ordinateur vous remerciera !