Fiche de mathématiques
> >

Matrices (I)

Partager :
Ici, \mathbb {K} désigne un corps commutatif.
Tous les \mathbb{K}-ev considérés sont supposés de dimension finie \geq 1

I- Généralités

1- Définitions

Soit (n,p)\in(\mathbb{N}^*)^{2}
Définitions - Notations :
\bullet On appelle matrice à n lignes, p colonnes et à éléments (ou coefficients) dans \mathbb{K} toute application de \lbrace 1,\cdots,n\rbrace \times\lbrace 1,\cdots, p\rbrace dans \mathbb{K} .
\bullet Une application \large\begin{array}{clcl} A : & \lbrace 1,\cdots,n\rbrace \times\lbrace 1,\cdots, p\rbrace  &\longrightarrow &\mathbb{K}\\ &(i,j)           &\longmapsto     &a_{ij}\\ \end{array}est notée sous la forme d'un tableau :
A=(a_{ij})_{\small \left\begin{array}{l}1\leq i\leq n\\1\leq j\leq p\end{array}\right}=(a_{ij})_{1\leq i\leq n , 1\leq j\leq p}=(a_{ij})_{ij}=(a_{ij}) =\begin{pmatrix}a_{11}&a_{12}&a_{13}&\cdots&a_{1p}\\a_{21}&a_{22}&a_{23}&\cdots&a_{2p}\\ \vdots&\vdots&\vdots& &\vdots\\a_{n1}&a_{n2}&a_{n3}&\cdots&a_{np}  \end{pmatrix}
\bullet Le couple (n,p) est appelé le format de la matrice A, n est le nombre de lignes de A, et p est le nombre de colonnes de A.
\bullet Pour (i,j) \in \lbrace 1,\cdots,n\rbrace \times\lbrace 1,\cdots, p\rbrace , le terme a_{ij} située à la i^{\text ème} ligne et j^{\text ème} colonne s'appelle le (i,j)^{\text ème} terme (ou coefficient) de A .
\bullet On dit que :
A est une matrice carrée si et seulement si n=p , on dit alors que A est une matrice carrée d'ordre n .
A est une matrice-colonne (ou matrice unicolonne) si et seulement si p=1 .
A est une matrice-ligne (ou matrice uniligne) si et seulement si n=1 .
\bullet Si A=(a_{ij})_{ij} est carré d'ordre n, les a_{ii} (1\leq i\leq n) sont appelés les éléments diagonaux de A, et (a_{11},\cdots,a_{nn}) est appelé la diagonale de A .
\bullet On note \mathcal{M}_{n,p}(\mathbb{K}) l'ensemble des matrices à n lignes, p colonnes, et à éléments dans \mathbb{K} .
\bullet On note \mathcal{M}_{n}(\mathbb{K}) l'ensemble \mathcal{M}_{n,n}(\mathbb{K}) des matrices carrées d'ordre n à éléments dans \mathbb{K} .


Exemples:
A=\begin{pmatrix}1&-5&-\sqrt{3}\\\displaystyle{\frac{1}{2}}&8&-7\\2&-\displaystyle{\frac{7}{2}}&0\end{pmatrix} est une matrice carré d'ordre 3 à coefficients réels, A\in\mathcal{M}_{3}(\mathbb{R}) .
B=\begin{pmatrix}i&7+2i&0\\\displaystyle{\frac{1}{2}}i&8&j\\2&-\displaystyle{\frac{5}{9}}&0\end{pmatrix} est une matrice carré d'ordre 3 à coefficients dans \mathbb{C} , B\in\mathcal{M}_{3}(\mathbb{C}) .
C=\begin{pmatrix} 1\\\displaystyle{\frac{4}{3}}\\ -\sqrt{5} \\ 0 \\ -1 \end{pmatrix} est une matrice colonne de \mathcal{M}_{5,1}(\mathbb{R}) .
D=\begin{pmatrix} 1+2i&\displaystyle{\frac{-9}{7}}& i & -j & 0 \end{pmatrix} est une matrice ligne de \mathcal{M}_{1,5}(\mathbb{C}) .

Définitions :
Soit A=(a_{ij})_{1\leq i\leq n , 1\leq j\leq p} \in \mathcal{M}_{n,p}(\mathbb{K}) .
\bullet Pour k\in \lbrace 1,\cdots , n\rbrace :
La matrice ligne (a_{k,j})_{1\leq j\leq p}=\begin{pmatrix} a_{k,1}&a_{k,2}& \cdots & a_{k,p} \end{pmatrix} \in\mathcal{M}_{1,p}(\mathbb{K}) est appelée la k^{\text{ème}} ligne de A .
On appelle k^{\text{ème}} vecteur ligne de A le vecteur (a_{k,1}, \cdots ,a_{k,p}), c'est un élément de \mathbb{K}^p .
\bullet Pour m\in \lbrace 1,\cdots , p\rbrace :
La matrice colonne (a_{i,m})_{1\leq i\leq n}=\begin{pmatrix} a_{1,m}\\a_{2,m}\\ \vdots \\ a_{n,m} \end{pmatrix} \in\mathcal{M}_{n,1}(\mathbb{K}) est appelée la m^{\text{ème}} colonne de A .
On appelle m^{\text{ème}} vecteur colonne de A le vecteur (a_{1,m}, \cdots ,a_{n,m}), c'est un élément de \mathbb{K}^n.



2- Opérations sur les matrices

a- Égalité entre deux matrices
L'égalité entre deux matrices est en fait l'égalité entre deux fonctions, par conséquent deux matrices sont égales lorsqu'elles ont la même taille et les mêmes coefficients.
Définition :
Soient A,B \in\mathcal{M}_{n,p}(\mathbb{K}) tels que A=(a_{i,j})_{1\leq i\leq n , 1\leq j\leq p} et B=(b_{i,j})_{1\leq i\leq n , 1\leq j\leq p}
On dit que A et B sont égaux et on note A=B si et seulement si : \forall (i,j)\in\lbrace 1,\cdots,n\rbrace \times\lbrace 1,\cdots, p\rbrace : a_{i,j}=b_{i,j}



b- Addition
Définition :
On appelle addition dans \mathcal{M}_{n,p}(\mathbb{K}) la loi interne, notée +, définie par:
\forall (a_{i,j})_{1\leq i\leq n , 1\leq j\leq p}\in \mathcal{M}_{n,p}(\mathbb{K}) et \forall (b_{i,j})_{1\leq i\leq n , 1\leq j\leq p}\in \mathcal{M}_{n,p}(\mathbb{K}) : (a_{i,j})_{1\leq i\leq n , 1\leq j\leq p}+(b_{i,j})_{1\leq i\leq n , 1\leq j\leq p}=(a_{i,j}+b_{i,j})_{1\leq i\leq n , 1\leq j\leq p}


Remarque :
On ne somme que des matrices de même type.

Exemple :
\begin{pmatrix}1&-5&0\\1&8&-7\end{pmatrix} + \begin{pmatrix}1&4&0\\1&-6&0\end{pmatrix} = \begin{pmatrix}2&-1&0\\2&2&-7\end{pmatrix}

Proporiétés :
La loi + est :
Associative et commutative.
Admet un élément neutre qui est la matrice de \mathcal{M}_{n,p}(\mathbb{K}) dont tous les termes sont nuls, on la note O_{n,p} ou plus simplement O et on la nomme la matrice nulle.
Tout A=(a_{ij}) de \mathcal{M}_{n,p}(\mathbb{K}) admet un opposé noté -A défini par : \forall (i,j) \in \lbrace 1,\cdots,n\rbrace \times\lbrace 1,\cdots, p\rbrace : -A=(-a_{ij})


Remarque:
L'élément neutre de \mathcal{M}_{n}(\mathbb{K}) O_{n,n} sera noté O_n .
Théorème:
(\mathcal{M}_{n,p}(\mathbb{K}),+) est un groupe abélien



c- Multiplication par un scalaire
Définition :
On appelle multiplication par les scalaires la loi externe \mathbb{K}\times\mathcal{M}_{n,p}(\mathbb{K}) \longrightarrow \mathcal{M}_{n,p}(\mathbb{K}), notée par . ou par absence de symbole, définie par :
\forall \lambda \in\mathbb{K} , \forall (a_{ij})_{ij} \in \mathcal{M}_{n,p}(\mathbb{K}) : \lambda . (a_{ij})_{ij}=(\lambda. a_{ij})_{ij}


Exemple:
4\begin{pmatrix}1&-2&0\\1&0&-1\end{pmatrix}=\begin{pmatrix}4&-8&0\\4&0&-4\end{pmatrix}
Propriétés :
Soient A, B \in \mathcal{M}_{n,p}(\mathbb{K}) , soient \lambda , \mu \in \mathbb{K} 
1.A= A 
\lambda .(A+ B) = \lambda.A+\lambda.B 
(\lambda+\mu).A= \lambda.A+\mu.A
(\lambda\mu).A= \lambda.(\mu.A)


Théorème :
(\mathcal{M}_{n,p}(\mathbb{K}),+,.) est un \mathbb{K}-espace vectoriel


Définition :
Pour (n,p)\in(\mathbb{N}^{*})^{2} et (i,j)\in \lbrace 1,\cdots,n\rbrace \times\lbrace 1,\cdots, p\rbrace , on note E_{ij} la matrice de \mathcal{M}_{n,p}(\mathbb{K}) dont le \displaystyle (i,j)^{ème} terme vaut 1 et tous les autres sont nuls. Les matrices E_{ij} sont appelées les matrices élémentaires de \mathcal{M}_{n,p}(\mathbb{K})


Exemple :
Les matrices élémentaires de \mathcal{M}_{2}(\mathbb{K}) sont : E_{11}=\begin{pmatrix}1&0\\0&0\end{pmatrix} , E_{12}=\begin{pmatrix}0&1\\0&0\end{pmatrix} , E_{21}=\begin{pmatrix}0&0\\1&0\end{pmatrix} et E_{22}=\begin{pmatrix}0&0\\0&1\end{pmatrix}
Les matrices élémentaires de \mathcal{M}_{1,3}(\mathbb{K}) sont : E_{11}=\begin{pmatrix}1&0&0\end{pmatrix} , E_{12}=\begin{pmatrix}0&1&0\end{pmatrix} et E_{13}=\begin{pmatrix}0&0&1\end{pmatrix}
Propostion :
(E_{ij})_{(i,j)\in\lbrace 1,\cdots,n\rbrace \times\lbrace 1,\cdots, p\rbrace } est une base de \mathcal{M}_{n,p}(\mathbb{K}), appelée base canonique de \mathcal{M}_{n,p}(\mathbb{K}) .
\dim{\mathcal{M}_{n,p}(\mathbb{K})}=np, en particulier : \dim{\mathcal{M}_{n}(\mathbb{K})}=n^2


Exemple de manipulation vectorielle :
Soient A_{1}=\begin{pmatrix}1&0\\0&1\end{pmatrix}, A_{2}=\begin{pmatrix}1&0\\0&-1\end{pmatrix} , A_{3}=\begin{pmatrix}1&1\\1&1\end{pmatrix} , A_{1}=\begin{pmatrix}0&-1\\1&0\end{pmatrix}
On se propose de montrer que la famille (A_i)_{1\leq i\leq 4} est une base de \mathcal{M}_{2}(\mathbb{R}).
Il est clair que card{(A_i)}_{1\leq i\leq 4}=\dim{\mathcal{M}_{2}(\mathbb{R})}=4.
Il suffit donc de montrer que (A_i)_{1\leq i\leq 4} est libre .
Soient \lambda_{1},\lambda_{2},\lambda_{3},\lambda_{4} \in\mathbb{R} tels que : \lambda_{1}A_{1}+\lambda_{2}A_{2}+\lambda_{3}A_{3}+\lambda_{4}A_{4}=O_{2}
Matriciellement : \begin{pmatrix}\lambda_{1}+\lambda_{2}+\lambda_{3}&\lambda_{3}-\lambda_{4}\\\lambda_{3}+\lambda_{4}&}\lambda_{1}-\lambda_{2}+\lambda_{3}\end{pmatrix} = \begin{pmatrix}0&0\\0&0\end{pmatrix}
En résolvant le système, on obtient : \lambda_{1}=\lambda_{2}=\lambda_{3}=\lambda_{4}=0
Ce qui permet de conclure.

d- Produit
Définition :
Soient A=(a_{ij})_{ij} \in \mathcal{M}_{n,p}(\mathbb{K}), B=(b_{jk})_{jk}\in \mathcal{M}_{p,q}(\mathbb{K}).
On appelle produit de A par B, et on note A\times B ou encore par absence de symbole AB, la matrice de \mathcal{M}_{n,q}(\mathbb{K}) définie par : AB=(c_{ik})_{ik} où :
\forall (i,k)\in \in\lbrace 1,\cdots,n\rbrace \times\lbrace 1,\cdots, q\rbrace : \displaystyle c_{ik}=\sum_{j=1}^{p} a_{ij}b_{jk}


Remarques :
Le produit AB n'est possible que si le nombre de colonnes de A est égal au nombre de lignes de B. Le résultat a alors autant de lignes que A et autant de colonnes que B. 
En général A B \not{=} B A, il se peut même que A B soit défini, mais pas B A.

Exemples :

\begin{pmatrix}4&-8&0\\4&0&-4\\1&3&-3\end{pmatrix} \begin{pmatrix}5&1\\4&0\\2&2\end{pmatrix} = \begin{pmatrix}-12&4\\12&-4\\11&-5\end{pmatrix}

\begin{pmatrix}1&2\\-2&1\end{pmatrix}\begin{pmatrix}0&1\\-1&2\\0&1\end{pmatrix} n'est pas défini, par contre \begin{pmatrix}0&1\\-1&2\\0&1\end{pmatrix} \begin{pmatrix}1&2\\-2&1\end{pmatrix}=\begin{pmatrix}-2&1\\-5&1\\0&5\end{pmatrix}
Proposition :
Pseudo-distributivité à gauche :
\forall A \in\mathcal{M}_{n,p}(\mathbb{K}) , \forall B,C \in\mathcal{M}_{p,q}(\mathbb{K}) : A(B+C)=AB+AC .
Pseudo-distributivité à droite :
\forall A,B \in\mathcal{M}_{n,p}(\mathbb{K}), \forall C \in\mathcal{M}_{p,q}(\mathbb{K}) : (A+B)C=AC+BC .
\forall \lambda \in\mathbb{K} , \forall A\in \mathcal{M}_{n,p}(\mathbb{K}) , \forall B\in \mathcal{M}_{p,q}(\mathbb{K}) : (\lambda A)B=\lambda(AB)=A(\lambda B) .
Pseudo-associativité :
\forall A\in \mathcal{M}_{n,p}(\mathbb{K}), \forall B\in \mathcal{M}_{p,q}(\mathbb{K}), \forall C\in \mathcal{M}_{q,r}(\mathbb{K}) : (AB)C=A(BC) .



3- La \displaystyle \mathbb{K}-algèbre \displaystyle(\mathcal{M}_{n}(\mathbb{K}),+,.,\times) des matrices carrées

a- Présentation
Proposition-Définition :
Le produit matriciel \times définit une loi de composition interne sur \mathcal{M}_{n}(\mathbb{K})
(\mathcal{M}_{n}(\mathbb{K}),+,.,\times) est une \mathbb{K}-algèbre associative, unitaire.
On note I_{n}=\displaystyle\begin{pmatrix}1&&O\\ &\ddots&\\O&&1\end{pmatrix}\in\mathcal{M}_{n}(\mathbb{K}) l'élément neutre du produit matriciel.


Remarque :
La \mathbb{K}-algèbre (\mathcal{M}_{n}(\mathbb{K}),+,.,\times) n'est commutative que pour n=1, dans ce cas, elle est identifiable à (\mathbb{K},+,.,\times).
Définition :
On dit que deux matrices A et B de \mathcal{M}_{n}(\mathbb{K}) commutent ssi : AB=BA.



Remarque :
\forall A\in\mathcal{M}_{n}(\mathbb{K}) : A \times I_n = I_n \times A= A
Donc : I_n commute avec toute matrice de \mathcal{M}_{n}(\mathbb{K})

b- Matrices diagonales et matrices triangulaires
Définition :
Une matrice carrée A=(a_{ij})_{1\leq i,j\leq n} \in \mathcal{M}_{n}(\mathbb{K}) est dite diagonale si et seulement si : \forall (i,j)\in \lbrace 1,\cdots , n\rbrace ^{2} : i\not{=}j \Longrightarrow a_{ij}=0
On note \mathcal{D}_{n}(\mathbb{K}) l'ensemble des matrices diagonales d'ordre n.


Exemples :
\begin{pmatrix}2&0\\0&7\end{pmatrix} \in \mathcal{D}_{2}(\mathbb{R})
\begin{pmatrix}-6&0&0\\0&-i&0\\0&0&101\end{pmatrix} \in \mathcal{D}_{3}(\mathbb{C})
Notation :
On notera diag(\alpha_{i})_{1\leq i\leq n} la matrice diagonale de \mathcal{D}_{n}(\mathbb{K}) dont la diagonale est la famille (\alpha_{i})_{1\leq i\leq n}



Exemples :
\begin{pmatrix}-6&0&0\\0&-i&0\\0&0&101\end{pmatrix} =diag(-6,-i,101) \in\mathcal{D}_{3}(\mathbb{C})
I_n=diag(1,1,\cdots,1) et O_{n}=diag(0,0,\cdots,0)
Définition :
Soit la matrice carrée A=(a_{ij})_{1\leq i,j\leq n} \in \mathcal{M}_{n}(\mathbb{K}).
A est dite triangulaire supérieure (ou: trigonale supérieure) si et seulement si :\forall (i,j)\in \lbrace 1,\cdots , n\rbrace ^{2} : i>j \Longrightarrow a_{ij}=0
On note \mathcal{T}_{n}^{+}(\mathbb{K}) l'ensemble des matrices triangulaires supérieures d'ordre n.

A est dite triangulaire inférieure (ou: trigonale inférieure) si et seulement si :\forall (i,j)\in \lbrace 1,\cdots , n\rbrace ^{2} : i<j \Longrightarrow a_{ij}=0
On note \mathcal{T}_{n}^{-}(\mathbb{K}) l'ensemble des matrices triangulaires inférieures d'ordre n.



Exemples :
\begin{pmatrix}-6&2&i+2\\0&-i&3\\0&0&j\end{pmatrix} \in\mathcal{T}_{3}^{+}(\mathbb{C})
\begin{pmatrix}-6&0&0\\3&-7&0\\-11&0&4\end{pmatrix} \in\mathcal{T}_{3}^{-}(\mathbb{R})

Proposition :
\mathcal{D}_{n}(\mathbb{K})=\mathcal{T}_{n}^{+}(\mathbb{K})\cap \mathcal{T}_{n}^{-}(\mathbb{K})



Théorème :
\mathcal{D}_{n}(\mathbb{K}) ,\mathcal{T}_{n}^{+}(\mathbb{K}) et \mathcal{T}_{n}^{-}(\mathbb{K}) sont des sous-espaces vectoriels de \mathcal{M}_{n}(\mathbb{K})
\mathcal{D}_{n}(\mathbb{K}) ,\mathcal{T}_{n}^{+}(\mathbb{K}) et \mathcal{T}_{n}^{-}(\mathbb{K}) sont des sous-algèbres de la \mathbb{K}-algèbre \mathcal{M}_{n}(\mathbb{K})



c- Puissances d'une matrice carrée
Définition :
Soit A\in\mathcal{M}_{n}(\mathbb{K})
Partant de la définition du produit de matrices carrées, on peut considérer le cas particulier des puissances d'une matrice tel que :
A^{0}=I_{n}
A^{1}=A
A^{2}=A\times A
A^{3}=A\times A^{2}=A\times A\times A
\vdots
A^{n}=A\times A^{n-1}=\underbrace{A\times A\times \cdots \times A}_{\displaystyle n fois}



Exemple :
Soit A\in\mathcal{M}_{2}(\mathbb{R}) tel que : A=\begin{pmatrix}1&1\\0&1\end{pmatrix}
calculer A^{n} ( \forall n\in \mathbb{N})

Par calcul simple, A^{2}=\begin{pmatrix}1&2\\0&1\end{pmatrix}, A^{3}=\begin{pmatrix}1&3\\0&1\end{pmatrix}
On en déduit que A^{k}=\begin{pmatrix}1&k\\0&1\end{pmatrix} pour k\in\mathbb{N}
Montrons maintenant par récurrence que \forall n\in \mathbb{N} : A^{n}=\begin{pmatrix}1&n\\0&1\end{pmatrix}
Pour n=0 : A^{0}=I_{2} (vérifié)
Supposons que pour n\in\mathbb{N} : A^{n}=\begin{pmatrix}1&n\\0&1\end{pmatrix} et montrons que A^{n+1}=\begin{pmatrix}1&n+1\\0&1\end{pmatrix}
A^{n+1}=A^{n}\times A=\begin{pmatrix}1&n\\0&1\end{pmatrix}\begin{pmatrix}1&1\\0&1\end{pmatrix}=\begin{pmatrix}1&n+1\\0&1\end{pmatrix}

Résultat : \displaystyle \boxed{\forall n\in\mathbb{N} : A^{n}=\begin{pmatrix}1&n\\0&1\end{pmatrix}}

Remarque :
Attention : (A+B)^{2}=A^{2}+AB+BA+B^{2}
Théorème :
Soit (A,B)\in (\mathcal{M}_{n}(\mathbb{K}))^{2} tel que A et B commutent, on a alors \forall m\in\mathbb{N}
(AB)^{m}=A^{m}B^{m}
\displaystyle (A+B)^{m}=\sum_{k=0}^{m} {m\choose k}  A^{k} B^{m-k} (binôme de Newton)
\displaystyle A^{m}-B^{m}=(A-B)\sum_{k=0}^{m-1} A^{k} B^{m-1-k}


Remarque :
AB=0 n'implique pas que A=0 ou B=0
Par conséquence, A^{2}=I_{2} possède dans \mathcal{M}_{2}(\mathbb{R}) d'autres solutions que I_{2} et -I_{2}
Par exemple : \begin{pmatrix}1&0\\0&-1\end{pmatrix}, \begin{pmatrix}-1&0\\0&1\end{pmatrix}
Définition :
On dit qu'une matrice A\in \mathcal{M}_{n}(\mathbb{K}) est idempotente si A^{2}=A.
On dit qu'une matrice A\in \mathcal{M}_{n}(\mathbb{K}) est nilpotente s'il existe m\in\mathbb{N}^{*} tel que A^{m}=O_{n} .


Exemples :
A=\begin{pmatrix}2&-1\\2&-1\end{pmatrix} est idempotente, en effet : A^{2}=\begin{pmatrix}2&-1\\2&-1\end{pmatrix}\begin{pmatrix}2&-1\\2&-1\end{pmatrix}=\begin{pmatrix}2&-1\\2&-1\end{pmatrix}.
B=\begin{pmatrix}0&1&2\\0&0&3\\0&0&0\end{pmatrix} est nilpotente.
En effet : B^{2}=\begin{pmatrix}0&1&2\\0&0&3\\0&0&0\end{pmatrix}\begin{pmatrix}0&1&2\\0&0&3\\0&0&0\end{pmatrix}=\begin{pmatrix}0&0&3\\0&0&0\\0&0&0\end{pmatrix} et B^{3}=\begin{pmatrix}0&1&2\\0&0&3\\0&0&0\end{pmatrix}\begin{pmatrix}0&0&3\\0&0&0\\0&0&0\end{pmatrix}=\begin{pmatrix}0&0&0\\0&0&0\\0&0&0\end{pmatrix}
Proposition :
Soit A\in\mathcal{D}_{n}(\mathbb{K}) tel que A=diag((x_{i})_{1\leq i\leq n} avec x_i\in\mathbb{K} \forall i \in \lbrace 1,\cdots, n\rbrace
On a alors : \forall k\in\mathbb{N} : A^{k}=diag((x_{i}^{k})_{1\leq i\leq n}



d- Matrices inversibles
Définition :
Une matrice A de \mathcal{M}_{n}(\mathbb{K}) est dite inversible si et seulement s'il existe A^{'} \in \mathcal{M}_{n}(\mathbb{K}) telle que AA^{'}=A^{'}A=I_{n}
Si A est inversible, alors A^{'} est unique et appelée inverse de A, et notée A^{-1}
On note \mathcal{GL}_{n}(\mathbb{K}) l'ensemble des matrices inversibles de \mathcal{M}_{n}(\mathbb{K})



Exemple :
La matrice I_n est inversible et : I_{n}^{-1}=I_n
Proposition :
(\mathcal{GL}_{n}(\mathbb{K}),\times) est un groupe appelé groupe linéaire d'ordre n.

Théorème : (dit d'inversibilité)
Pour A\in\mathcal{M}_{n}(\mathbb{K}), les propositions suivantes sont équivalantes:
(i) A est inversible
(ii)A est inversible à droite, ie: \exists B\in\mathcal{M}_{n}(\mathbb{K}) : AB=I_{n}
(iii) A est inversible à gauche ie: \exists C\in\mathcal{M}_{n}(\mathbb{K}) : CA=I_{n}
De plus si tel est le cas A^{-1}=B=C


Lemme :
Soient A,B \in\mathcal{M}_{n,p}(\mathbb{K}).
Si : \forall X\in\mathcal{M}_{p,1}(\mathbb{K}) : AX=BX alors A=B


Remarques :
Pour calculer l'inverse d'une matrice A\in\mathcal{M}_{n}(\mathbb{K}), on prend les deux matrices colonnes X=\begin{pmatrix}x\\y\\z\end{pmatrix} \in\mathcal{M}_{n,1}(\mathbb{K}) et Y=\begin{pmatrix}x^{'}\\y^{'}\\z^{'}\end{pmatrix} \in\mathcal{M}_{n,1}(\mathbb{K}), et on résout le système AX = Y, où X est la matrice colonne inconnue (ie. On cherche x, y et z en fonction de x^{'}, y^{'} et z^{'}) car :AX=Y \Longrightarrow X=A^{-1}Y.
Il existe aussi la méthode de "pivot de Gauss" pour calculer l'inverse d'une matrice, on exposera cette méthode plus tard.

Exemple :
Soit A=\begin{pmatrix}1&2\\1&3\end{pmatrix}
Soit X=\begin{pmatrix}x\\y\end{pmatrix} et Y=\begin{pmatrix}x^{'}\\y^{'}\end{pmatrix} tels que: AX=Y
\begin{pmatrix}1&2\\1&3\end{pmatrix}\begin{pmatrix}x\\y\end{pmatrix}=\begin{pmatrix}x^{'}\\y^{'}\end{pmatrix} \Longleftrightarrow \begin{cases} x^{'}=x+2y\text{     &   }\\  y^{'}=x+3y \text{     &    }   \end{cases}\Longleftrightarrow \begin{cases}x=3x^{'}-2y^{'} \text{     &   }\\  y=-x^{'}+y^{'} \text{    &    }   \end{cases}\Longleftrightarrow \begin{pmatrix}x\\y\end{pmatrix}=\begin{pmatrix}3&-2\\-1&1\end{pmatrix}\begin{pmatrix}x^{'}\\y^{'}\end{pmatrix}\Longleftrightarrow X=A^{-1}Y
Donc : \boxed{A^{-1}=\begin{pmatrix}3&-2\\-1&1\end{pmatrix}}

Proposition :
Soit (a_i)_{1\leq i\leq n} une famille d'éléments dans \mathbb{K} et soit la matrice A\in \mathcal{D}_{n}(\mathbb{K}) telle que A=diag(a_{i})_{1\leq i\leq n}. Les propositions suivantes sont équivalentes:
(i) A est inversible
(ii) \forall i\in\lbrace 1,\cdots,n\rbrace : a_i\not[=}0
De plus, dans ce cas : \displaystyle A^{-1}=diag(\frac{1}{a_i})_{1\leq i\leq n}

Soit la matrice A\in \mathcal{T}_{n}^{+}(\mathbb{K}) (respectivement A\in \mathcal{T}_{n}^{-}(\mathbb{K})). On a équivalence entre:
(i) A est inversible
(ii) Les coefficients diagonaux de A sont tous non nuls.
De plus, si tel est le cas, A^{-1}\in \mathcal{T}_{n}^{+}(\mathbb{K}) (respectivement A^{-1}\in \mathcal{T}_{n}^{-}(\mathbb{K}))



4- Transposition

a- Définition et propriétés:
Définition:
Pour toute matrice A=(a_{ij})_{ 1\leq i\leq n , 1\leq j\leq p}=\begin{pmatrix}a_{11}&\cdots&a_{1p}\\ \vdots&&\vdots \\ a_{n1}&\cdots&a_{np} \end{pmatrix} de \mathcal{M}_{n,p}(\mathbb{K}), on appelle transposée de A la matrice notée ^t A de \mathcal{M}_{p,n}(\mathbb{K}) définie par :
^t A =(a_{ij})_{1\leq j\leq p , 1\leq i\leq n}= \begin{pmatrix}a_{11}&\cdots&a_{n1}\\ \vdots& &\vdots\\ a_{1p}&\cdots&a_{np} \end{pmatrix} (Ainsi le coefficient d'indice (j,i) de ^t A est égal au coefficient d'indice (i,j) de A)


Remarque :
Les colonnes et lignes de ^{t}A correspondent respectivement aux lignes et colonnes de A.

Exemple :
A=\begin{pmatrix}0&1&2\\0&0&3\end{pmatrix}, donc : ^t A=\begin{pmatrix}0&0\\1&0\\2&3\end{pmatrix}

Proposition :
L'application \begin{array}{clcl} T :  \mathcal{M}_{n,p}(\mathbb{K})  \longrightarrow \mathcal{M}_{p,n}(\mathbb{K})\\ M           \longmapsto      ^{t}M\\ \end{array} est un isomorphisme de \mathbb{K}-espaces vectoriels.


Proposition : (Propriétés)
\forall A\in\mathcal{M}_{n,p}(\mathbb{K}) : ^{t}(^{t}A)=A
\forall (\alpha,\beta)\in\mathbb{K}^{2} , \forall (A,B)\in (\mathcal{M}_{n,p}(\mathbb{K}))^{2} : ^{t}(\alpha A+\beta B)=\alpha ^{t}A+\beta ^{t}B
\forall A\in\mathcal{M}_{n,p}(\mathbb{K}) , \forall B\in\mathcal{M}_{p,n}(\mathbb{K}) : ^{t}(AB)=^{t}B ^{t}A
\forall A\in\mathcal{GL}_{n}(\mathbb{K}) : ^{t}A\in\mathcal{GL}_{n}(\mathbb{K}) et (^{t}A)^{-1}=^{t}(A^{-1})



b- Matrices carrées symétriques et antisymétriques
Définitions:
i)
Une matrice A de \mathcal{M}_{n}(\mathbb{K}) est dite symétrique si et seulement si : ^{t}A=A
On note \mathcal{S}_{n}(\mathbb{K}) l'ensemble des matrices symétriques d'ordre n à coefficients dans \mathbb{K}
ii)
Une matrice A de \mathcal{M}_{n}(\mathbb{K}) est dite antisymétrique si et seulement si : ^{t}A=-A
On note \mathcal{A}_{n}(\mathbb{K}) l'ensemble des matrices antisymétriques d'ordre n à coefficients dans \mathbb{K}


Proposition :
\mathcal{S}_{n}(\mathbb{K}) est un sous-espace vectoriel de \mathcal{M}_{n}(\mathbb{K}) de dimension \dfrac{n(n+1)}{2}
\mathcal{A}_{n}(\mathbb{K}) est un sous-espace vectoriel de \mathcal{M}_{n}(\mathbb{K}) de dimension \dfrac{n(n-1)}{2}


Proposition :
\forall(A,B)\in (\mathcal{S}_{n}(\mathbb{K}))^{2} : AB\in\mathcal{S}_{n}(\mathbb{K}) \Longleftrightarrow AB=BA (ie. A et B commutent)
\forall A\in \mathcal{S}_{n}(\mathbb{K}) \cap \mathcal{GL}_{n}(\mathbb{K}) : A^{-1}\in \mathcal{S}_{n}(\mathbb{K})


Théorème :
Les sous-espaces vectoriels \mathcal{S}_{n}(\mathbb{K}) et \mathcal{A}_{n}(\mathbb{K}) sont supplémentaires dans \mathcal{M}_{n}(\mathbb{K})



5- Matrices semblables et matrices équivalentes

Définitions :

Soient A,B \in\mathcal{M}_{n,p}(\mathbb{K}), A et B sont dites équivalentes et on note A eq B si et seulement s'il existe deux matrices inversibles P\in\mathcal{M}_{n}(\mathbb{K}) et Q\in\mathcal{M}_{p}(\mathbb{K}) telles que: A=PBQ.
Soient A,B \in\mathcal{M}_{n}(\mathbb{K}), A et B sont dites semblables et on note A\sim B si et seulement s'il existe P\in\mathcal{M}_{n}(\mathbb{K}) inversible telle que : A=PBP^{-1}


Proposition :
Les relations eq et \sim sont des relations d'équivalence respectivement dans \mathcal{M}_{n,p}(\mathbb{K}) et \mathcal{M}_{n}(\mathbb{K})




II- Matrices et applications linéaires

1- Représentations matricielles

a- Matrice colonne des composantes d'un vecteur
Définition :
Soient E un \mathbb{K}-ev , n=\dim{(E)} , \mathcal{B}=(e_{1},\cdots,e_{n}) une base de E , x\in E , (x_{1},\cdots,x_{n}) les composantes de x dans la base \mathcal{B} : \displaystyle x=\sum_{k=1}^{n} x_{k}e_{k}
La matrice colonne \begin{pmatrix} x_{1}\\x_{2} \\ \vdots \\x_{n} \end{pmatrix} s'appelle la matrice-colonne des composantes de x dans \mathcal{B}, on la note : Mat_{\mathcal{B}}(x)


Remarques :
Puisque les composantes d'un vecteur dépendent de la base choisie, il est nécessaire de préciser celle-ci.
Toutes les matrices des composantes d'un vecteur appartiennent à \mathcal{M}_{n,1}(\mathbb{K}) .

Exemples :
Soit dans \mathbb{R}_{3}[X] , le polynôme P=2X^{3}+8X+3, la matrice-colonne des composantes de P dans la base canonique \mathcal{B}=(1,X,X^{2},X^{3}) est : Mat_{\mathcal{B}}(P)=\begin{pmatrix}3\\8 \\0 \\2 \end{pmatrix}
Soit dans \mathcal{M}_{2,3}(\mathbb{R}), la matrice A=\begin{pmatrix}2&-1&1\\2&-3&5\end{pmatrix}, la matrice-colonne de A dans la base canonique \mathcal{B}=(E_{11},E_{12},E_{13},E_{21},E_{22},E_{23}) est : Mat_{\mathcal{B}}(A)=\begin{pmatrix}2\\-1\\1 \\2\\-3\\5 \end{pmatrix}

Théorème :
En conservant les notations de la définition; L'application \begin{array}{clcl} M_{\mathcal{B}} :  E  \longrightarrow \mathcal{M}_{n,1}(\mathbb{K})\\ x          \longmapsto      Mat_{\mathcal{B}}(x)\\ \end{array} est un isomorphisme de \mathbb{K}-espace vectoriel.



b- Matrice des composantes d'une famille de vecteurs
Définition :
Soient E un \mathbb{K}-ev , n=\dim{(E)} , \mathcal{B}=(e_{1},\cdots,e_{n}) une base de E, p \in\mathbb{N}^{*}, \mathcal{F}=(V_{1},\cdots, V_{p}) une famille finie de p éléments de E, et, pour chaque j de \lbrace 1,\cdots , p\rbrace, (a_{1j},\cdots,a_{nj}) les composantes de V_{j} dans \mathcal{B} : \forall j\in\lbrace 1,\cdots , p\rbrace , V_{j}=\displaystyle\sum_{k=1}^{n} a_{kj}e_{k}
La matrice (a_{ij})_{1\leq i\leq n  , 1\leq j \leq p } = \begin{pmatrix}a_{11}&\cdots&a_{1p}\\ \vdots&&\vdots \\ a_{n1}&\cdots&a_{np} \end{pmatrix} de \mathcal{M}_{n,p}(\mathbb{K}) s'appelle la matrice de la famille \mathcal{F} relativement à la base \mathcal{B} et est notée Mat_{\mathcal{B}}(\mathcal{F}) ou Mat_{\mathcal{B}}(V_{1},\cdots, V_{p})


Remarque :
Dans le cas p=1, on retrouve la notion de matrice des composantes d'un vecteur.

Exemples :
Considérons l'espace vectoriel \mathbb{R}^{3} muni de sa base canonique \mathcal{B}=((1,0,0),(0,1,0),(0,0,1))
Soient x_{1}=(1,2,3), x_{2}=(2,0,1), x_{3}=(1,0,-1) et x_{4}=(-1,1,1) quatre vecteurs de \mathbb{R}^{3}
On a : Mat_{\mathcal{B}}(x_{1},x_{2},x_{3},x_{4})=\begin{pmatrix}1&2&1&-1\\2&0&0&1  \\ 3&1&-1&1 \end{pmatrix} \in \mathcal{M}_{3,4}(\mathbb{R})

Soit E un \mathbb{K}-ev de dimension n muni d'une base \mathcal{B} connue, il est clair que : Mat_{\mathcal{B}}(\mathcal{B})=I_{n}

c- Matrice d'une application linéaire
Définition :
Soient  \begin{cases} E,F  \text{deux} \mathbb{K}-ev, p=\dim{(E)}, n=\dim{(F)} \\ \mathcal{B}=(e_{1},\cdots,e_{p}) \text{une base de  } E , \mathcal{C}=(f_{1},\cdots,f_{n}) \text{une base de } F \\f\in\mathcal{L}(E,F)   \end{cases} .
Pour chaque j de \lbrace 1,\cdots , p \rbrace , notons (a_{1j},\cdots,a_{nj}) les composantes de f(e_{j}) dans \mathcal{C} : f(e_{j})=\displaystyle \sum_{k=1}^{n} a_{kj} f_{k}
On appelle matrice de f relativement aux bases \mathcal{B} et \mathcal{C}, et on note Mat_{\mathcal{B},\mathcal{C}}(f), la matrice de \mathcal{M}_{n,p}(\mathbb{K}) définie par : Mat_{\mathcal{B},\mathcal{C}}(f)= (a_{ij})_{1\leq i\leq n  , 1\leq j \leq p } = \begin{pmatrix}a_{11}&\cdots&a_{1p}\\ \vdots&&\vdots \\ a_{n1}&\cdots&a_{np} \end{pmatrix}



Remarque :
Lorsque A=Mat_{\mathcal{B},\mathcal{C}}(f), on dit que f est représentée par A dans les bases \mathcal{B} et \mathcal{C}, ou que A représente f dans \mathcal{B} et \mathcal{C}
La matrice représentative de f dépend du choix des bases \mathcal{B} et \mathcal{C}, il est donc nécessaire de préciser celles-ci.


Théorème :
Soient E et F deux \mathbb{K}-espaces vectoriels munis respectivement des bases \mathcal{B} et \mathcal{C}
L'application \begin{array}{clcl} M_{\mathcal{B},\mathcal{C}} : \mathcal{L}(E,F)  \longrightarrow \mathcal{M}_{n,p}(\mathbb{K})\\ u         \longmapsto      Mat_{\mathcal{B},\mathcal{C}}(u)\\ \end{array} est un isomorphisme de \mathbb{K}-espaces vectoriels.



Définition : (Cas Endomorphisme)
Soient E un \mathbb{K}-ev, n=\dim{(E)}, \mathcal{B}=(e_{1},\cdots,e_{n}) une base de E, f\in\mathcal{L}(E).
On appelle matrice de f relativement à la base \mathcal{B}, et on note Mat_{\mathcal{B}}(f), la matrice de \mathcal{M}_{n}(\mathbb{K}) définie par : Mat_{\mathcal{B}}(f)=Mat_{\mathcal{B},\mathcal{B}}(f)



Exemples :
Soit f : \mathbb{R}^{3}\longrightarrow \mathbb{R}^{2} l'application linéaire définie par : f(x,y,z)=(x+2y-z,x-y)
Formons la matrice de f relativement aux bases canoniques \mathcal{B}=(e_{1},e_{2},e_{3}) et \mathcal{B}^{'}=(u_{1},u_{2}) respectivement de \mathbb{R}^{3} et \mathbb{R}^{2}
On a :
f(e_{1})=f(1,0,0)=(1,1)=u_{1}+u_{2}
f(e_{2})=f(0,1,0)=(2,-1)=2u_{1}-u_{2}
f(e_{3})=f(0,0,1)=(-1,0)=-u_{1}
On en déduit:
Mat_{\mathcal{B},\mathcal{B}^{'}}(f)=\begin{pmatrix}1&2&-1\\ 1&-1&0 \end{pmatrix} \in \mathcal{M}_{2,3}(\mathbb{R})

Soit f\in\mathcal{L}(\mathbb{R}^{3}) tel que : f(x,y,z)=(y+z;z+x,x+y) et \mathcal{B}=(e_{1},e_{2},e_{3}) la base canonique de \mathbb{R}^{3}
On a : f(e_{1})=(0,1,1) , f(e_{2})=(1,0,1) , f(e_{3})=(1,1,0)
Donc : Mat_{\mathcal{B}}(f)=\begin{pmatrix}0&1&1\\ 1&0&1\\1&1&0 \end{pmatrix} \in \mathcal{M}_{3}(\mathbb{R})

2- Utilisation du calcul matriciel pour quelques problèmes relatifs aux applications linéaires

a- Image d'un vecteur par une application linéaire - Noyau et image
Théorème :
Soient E et F deux \mathbb{K}-espaces vectoriels munis respectivement des bases \mathcal{B} et \mathcal{C}, u\in\mathcal{L}(E,F)
La matrice de u dans les bases \mathcal{B} et \mathcal{C} est l'unique matrice A\in\mathcal{M}_{n,p}(\mathbb{K}) vérifiant :
\forall x\in E \forall y\in F : y=u(x) \Longleftrightarrow Y=AX avec X=Mat_{\mathcal{B}}(x) et Y=Mat_{\mathcal{C}}(y)


Définitions :
Soit A\in\mathcal{M}_{n,p}(\mathbb{K})
On appelle noyau de A le sev de \mathcal{M}_{p,1}(\mathbb{K}) noté Ker(A), défini par: Ker(A)=\lbrace X\in\mathcal{M}_{p,1}(\mathbb{K}) ,  AX=0\rbrace
On appelle image de A le sev de \mathcal{M}_{n,1}(\mathbb{K}) noté Im(A), défini par: Im(A)=\lbrace Y\in\mathcal{M}_{n,1}(\mathbb{K}) , \exists X\in\mathcal{M}_{p,1}(\mathbb{K}) , Y= AX\rbrace = \lbrace AX , X\in\mathcal{M}_{p,1}(\mathbb{K})\rbrace


Théorème : (dit de correspondance)
Soient E et F deux \mathbb{K}-espaces vectoriels munis respectivement des bases \mathcal{B} et \mathcal{C}, u\in\mathcal{L}(E,F), A=Mat_{\mathcal{B},\mathcal{C}}(u)
Il y a correspondance entre Ker(u) et Ker(A) et entre Im(A) et Im(u)



Cela veut dire que les éléments de Ker(A) sont les représentations matricielles des éléments de Ker(u) et que les éléments de Im(A) sont les représentations matricielles de Im(u) (relativement aux bases \mathcal{B} et \mathcal{C})

Exemple :
Soit E un \mathbb{R}-ev muni d'une base \mathcal{B}=(e_{1},e_{2},e_{3})
Soit u un endomorphisme de \mathcal{L}(E) dont la matrice dans \mathcal{B} est : A=Mat_{\mathcal{B}}(u)=\begin{pmatrix}1&1&2\\ 1&-1&0\\1&0&1 \end{pmatrix}
Soit x=x_{1}e_{1}+x_{2}e_{2}+x_{3}e_{3} \in E
On a: Mat_{\mathcal{B}}(u(x))=AX=\begin{pmatrix}x_{1}+x_{2}+2x_{3}\\x_{1}-x_{2}\\x_{1}+x_{3} \end{pmatrix}
On peut étudier Ker(u) en étudiant Ker(A), c'est-à-dire en résolvant l'équation matricielle AX=0_{3,1}
AX=0_{3,1}\Longleftrightarrow \begin{cases}x_{1}+x_{2}+2x_{3}=0  \\ x_{1}-x_{2}=0  \\x_{1}+x_{3}=0  \end{cases}\Longleftrightarrow \begin{cases}x_{1}=x_{2}  \\ x_{1}=-x_{3}    \end{cases}
Ker(A)=\lbrace \alpha \begin{pmatrix}1\\1\\-1 \end{pmatrix}  /  \alpha\in\mathbb{R} \rbrace
Ainsi, Ker(u)=\lbrace \alpha(e_{1}+e_{2}-e_{3}) / \alpha \in\mathbb{R} \rbrace
Donc : Ker(u)=Vect(e_{1}+e_{2}-e_{3})

b- Compositions d'applications linéaires
Proposition :
Soient \displaystyle \begin{cases} E,F,G \text{trois} \mathbb{K}-\text{ev} \\ \mathcal{B},\mathcal{C},\mathcal{D} \text{des bases de} E,F,G \text{respectivement} \\ f\in\mathcal{L}(E,F) , g\in\mathcal{L}(F,G) \end{cases}
On a: Mat_{\mathcal{B},\mathcal{D}}(gof)=(Mat_{\mathcal{C},\mathcal{D}}(g)) (Mat_{\mathcal{B},\mathcal{C}}(f))


Corollaire :
Soit E un \mathbb{K}-ev muni d'une base \mathcal{B}, on a:
\forall f,g\in\mathcal{L}(E) : Mat_{\mathcal{B}}(fog)=(Mat_{\mathcal{B}}(f)) (Mat_{\mathcal{B}}(g))
\forall f\in\mathcal{L}(E) : Mat_{\mathcal{B}}(f^{m})=(Mat_{\mathcal{B}}(f))^{m} pour tout m\in\mathbb{N}


Proposition :
Soient E un \mathbb{K}-ev de dimension n muni d'une certaine base \mathcal{B}, f\in\mathcal{L}(E), A la représentation matricielle de f dans la base \mathcal{B}, on a:
f est nilpotant \Longleftrightarrow A est nilpotante
f est un projecteur \Longleftrightarrow A^{2}=A
f est une symétrie \Longleftrightarrow A^{2}=I_{n}



c- Bijection et inversibilité
Théorème :
Soient E et F deux \mathbb{K}-espaces vectoriels de m^eme dimension munis respectivement des bases \mathcal{B} et \mathcal{C}, f\in\mathcal{L}(E,F), A=Mat_{\mathcal{B},\mathcal{C}}(f). Les propositions suiventes sont équivalantes:
i- f est un isomorphisme.
ii- A est inversible.
De plus, dans ce cas : Mat_{\mathcal{C},\mathcal{B}}(f^{-1})=A^{-1}


Corollaire :
Soient E un \mathbb{K}-espace vectoriel de dimension n muni d'une base \mathcal{B}, f\in\mathcal{L}(E). les propositions suivantes sont équivalantes:
i- f est un automorphisme
ii- Mat_{\mathcal{B}}(f)\in\mathcal{GL}_{n}(\mathbb{K})
De plus, si tel est le cas : Mat_{\mathcal{B}}(f^{-1})=[Mat_{\mathcal{B}}(f)]^{-1}



Exemple :
Soit E un \mathbb{R}-espace vectoriel muni de la base \mathcal{B}=(e_{1},e_{2},e_{3}) et soit F un \mathbb{R}-espace vectoriel muni de la base \mathcal{C}=(u_{1},u_{2},u_{3}).
Soit f:E\rightarrow F l'application linéaire définie par sa représentation matricielle :
Mat_{\mathcal{B},\mathcal{C}}(f)=\begin{pmatrix}1&0&-1\\0&0&2\\1&-1&-1 \end{pmatrix}
Montrons que f est un isomorphisme, pour cela, on montre que Mat_{\mathcal{B},\mathcal{C}}(f) est inversible.
Soient X=\begin{pmatrix}x_{1}\\x_{2}\\x_{3} \end{pmatrix} et Y=\begin{pmatrix}y_{1}\\y_{2}\\y_{3} \end{pmatrix} de \mathcal{M}_{3,1}(\mathbb{R}) telles que : AX=Y
Mat_{\mathcal{B},\mathcal{C}}(f)X=Y\Longleftrightarrow \begin{pmatrix}1&0&-1\\0&0&2\\1&-1&-1 \end{pmatrix}\begin{pmatrix}x_{1}\\x_{2}\\x_{3}\end{pmatrix}=\begin{pmatrix}y_{1}\\y_{2}\\y_{3} \end{pmatrix} \Longleftrightarrow  \begin{cases}x_{1}-x_{3}=y_{1}  \\ 2x_{3}=y_{2}  \\x_{1}-x_{2}-x_{3}=y_{3}  \end{cases}\Longleftrightarrow  \begin{cases}x_{3}=\displaystyle{\frac{y_{2}}{2}}  \\ x_{1}-x_{3}=y_{1}  \\x_{1}-x_{2}-x_{3}=y_{3}  \end{cases}\Longleftrightarrow  \begin{cases}x_{3}=\displaystyle{\frac{y_{2}}{2}}  \\ x_{1}=y_{1}+\displaystyle{\frac{y_{2}}{2}}  \\x_{2}=-y_{3}+y_{1}+\displaystyle{\frac{y_{2}}{2}}-\displaystyle{\frac{y_{2}}{2}}   \end{cases}
\Longleftrightarrow\boxed{\begin{cases}  x_{1}=y_{1}+\displaystyle{\frac{y_{2}}{2}} \\ x_{2}=y_{1}-y_{3}  \\x_{3}=\displaystyle{\frac{y_{2}}{2}}  \end{cases}}
Donc : [Mat_{\mathcal{B},\mathcal{C}}(f)]^{-1}=\begin{pmatrix}1&\displaystyle{\frac{1}{2}}&0\\1&0&-1\\0&\displaystyle{\frac{1}{2}}&0\end{pmatrix}
On en déduit que Mat_{\mathcal{B},\mathcal{C}}(f) est inversible, donc f est bijective.
Déterminons f^{-1} :
On a: [Mat_{\mathcal{B},\mathcal{C}}(f)]^{-1}=Mat_{\mathcal{C},\mathcal{B}}(f^{-1})
Donc : Mat_{\mathcal{C},\mathcal{B}}(f^{-1})=\begin{pmatrix}1&\displaystyle{\frac{1}{2}}&0\\1&0&-1\\0&\displaystyle{\frac{1}{2}}&0\end{pmatrix}
Soit (x,y,z)\in  F :
Mat_{\mathcal{C},\mathcal{B}}(f^{-1})\begin{pmatrix}x\\y\\z\end{pmatrix}=\begin{pmatrix}1&\displaystyle{\frac{1}{2}}&0\\1&0&-1\\0&\displaystyle{\frac{1}{2}}&0\end{pmatrix}\begin{pmatrix}x\\y\\z\end{pmatrix}=\begin{pmatrix}x+\displaystyle{\frac{1}{2}}y\\x-z\\\displaystyle{\frac{1}{2}}y\end{pmatrix}
Enfin : \boxed{f^{-1}:F\rightarrow E \text{ telle que }  f^{-1}(x,y,z)=(x+\displaystyle{\frac{1}{2}}y,x-z,\displaystyle{\frac{1}{2}}y)}
Remarquons que pour montrer que f est bijective, il suffit de montrer que l'équation matrcielle Mat_{\mathcal{B},\mathcal{C}}(f)\text{. }X=0 n'admet que X=0 comme solution, donc le noyau de f est se réduit au vecteur nul, donc f est injective, et par raison de dimension ( dim(E)=dim(F) ), f est bijective. La méthode présentée ci-dessus permet en plus de ça de déterminer f^{-1}.

d- Quelques endomorphismes particuliers
Proposition : (Homothétie vectorielle)
Soit E un \mathbb{K}-ev de dimension n .
Dans toute base de E, la matrice de l'homothétie vectorielle de rapport a\in\mathbb{K} est aI_{n}.


Proposition : (Symétrie et projection)
Soit E un \mathbb{K}-ev de dimension n, soient F et G deux sev supplémentaires de E respectivement de dimensions r et n-r.
La matrice de la projection p sur F parallèlement à G dans une base \mathcal{B} adaptée à la supplémentarité de F et G est :Mat_{\mathcal{B}}(p)=\begin{pmatrix}I_{r}&O\\O&O_{n-r}\end{pmatrix}
La matrice de la symétrie s sur F parallèlement à G dans une base \mathcal{B} adaptée à la supplémentarité de F et G est :Mat_{\mathcal{B}}(s)=\begin{pmatrix}I_{r}&O\\O&-I_{n-r}\end{pmatrix}



III- Changement de base

1- Matrice de passage

Définition :
Soit E un \mathbb{K}-espace vectoriel de dimension n muni de deux bases \mathcal{B}=(e_{1},\cdots,e_{n}) et \mathcal{B}^{'}=(e_{1}^{'},\cdots,e_{n}^{'}).
On appelle matrice de passage de la base \mathcal{B} à la base \mathcal{B}^{'} la matrice de \mathcal{M}_{n}(\mathbb{K}) notée P_{\mathcal{B}\mathcal{B}^{'}} ou plus simplement P telle que : P_{\mathcal{B}\mathcal{B}^{'}}=Mat_{\mathcal{B}}(\mathcal{B}^{'})= Mat_{\mathcal{B}}(e_{1}^{'},\cdots,e_{n}^{'})


Exemple :
Soit E=\mathbb{R}_{3}[X] muni de sa base canonique \mathcal{B}= (1,X,X^{2},X^{3})
La famille \mathcal{C}=(1,1+X,(1+X)^{2},(1+X)^{3}) est une base de E (démonstration simple et laissée en exercice)
Ecrivons la matrice de passage P_{\mathcal{B},\mathcal{C}} :
On sait que : 1=1+0X+0X^{2}+0X^{3}
1+X=1+1X+0X^{2}+0X^{3}
(1+X)^{2}=1+2X+1X^{2}+0X^{3}
(1+X)^{3}=1+3X+3X^{2}+1X^{3}
Donc : P_{\mathcal{B}\mathcal{C}}=Mat_{\mathcal{B}}(\mathcal{C})=\begin{pmatrix}1&1&1&1\\0&1&2&3\\0&0&1&3\\0&0&0&1 \end{pmatrix}

Proposition :
Soit E un \mathbb{K}-espace vectoriel.
Pour toutes bases \mathcal{B}, \mathcal{B}^{'} de E : P_{\mathcal{B}\mathcal{B}^{'}}=Mat_{\mathcal{B}^{'}\mathcal{B}}(Id_{E})


Théorème: (Caractérisation des bases)
Soit E un \mathbb{K}-espace vectoriel de dimension n muni d'une base \mathcal{B}, et soit \mathcal{B}^{'} une famille de n vecteurs de E.
Alors \mathcal{B}^{'} est une base de E si et seulement si P_{\mathcal{B}\mathcal{B}^{'}} \in\mathcal{GL}_{n}(\mathbb{K})


Proposition :
Soit E un \mathbb{K}-espace vectoriel de dimension n, \mathcal{B}, \mathcal{B}^{'}, \mathcal{B}^{''} des bases de E. On a:
P_{\mathcal{B}\mathcal{B}^{''}}=P_{\mathcal{B}\mathcal{B}^{'}}P_{\mathcal{B^{'}}\mathcal{B}^{''}}
P_{\mathcal{B}\mathcal{B}}=I_{n}
P_{\mathcal{B}\mathcal{B}^{'}} est inversible et [P_{\mathcal{B}\mathcal{B}^{'}}]^{-1}=P_{\mathcal{B}^{'}\mathcal{B}}



2- Nouvelles représentations par changement de base

Théorème : (Nouvelles composantes d'un vecteur)
Soient E un \mathbb{K}-espace vectoriel de dimension n, \mathcal{B},\mathcal{B}^{'} deux bases de E, x\in E, X=Mat_{\mathcal{B}}(x), X^{'}=Mat_{\mathcal{B^{'}}}(x) et P=X=Mat_{\mathcal{B}}(\mathcal{B^{'}}).
On a : X=PX^{'} et X^{'}=PX


Théorème : (Nouvelle représentation d'une application linéaire)
Soient E et F deux \mathbb{K}-espaces vectoriels, \mathcal{B},\mathcal{B}^{'} deux bases de E, \mathcal{C},\mathcal{C}^{'} deux bases de F, u\in \mathcal{L}(E,F), A=Mat_{\mathcal{B},\mathcal{C}}(u), A^{'}=Mat_{\mathcal{B^{'}},\mathcal{C^{'}}}(u), P=Mat_{\mathcal{B}}(\mathcal{B^{'}}) et Q=Mat_{\mathcal{C}}(\mathcal{C^{'}}).
On a : A^{'}=Q^{-1}AP


Théorème : (Nouvelle représentation d'un endomorphisme)
Soient E un \mathbb{K}-espace vectoriel, \mathcal{B},\mathcal{B}^{'} deux bases de E, u\in \mathcal{L}(E), A=Mat_{\mathcal{B}}(u), A^{'}=Mat_{\mathcal{B^{'}}}(u) et P=Mat_{\mathcal{B}}(\mathcal{B^{'}}).
On a : A^{'}=P^{-1}AP


Exemple:
Soit E un \mathbb{R}-ev muni d'une base \mathcal{B}=(e_{1},e_{2},e_{3}) et soit f\in\mathcal{L}(E) définie par : Mat_{\mathcal{B}}(f)=A avec A=\begin{pmatrix}2&1&1\\-3&-2&-1\\3&5&4 \end{pmatrix}.
Soit \mathcal{C}=(u_{1},u_{2},u_{3}) une famille d'éléments dans E définie par : \begin{cases}u_{1}=-e_{2}+e_{3}  \\ u_{2}=e_{1}-e_{2}+e_{3}  \\u_{3}=-e_{1}+e_{2}-2e_{3}  \end{cases}.
Il est simple de vérifier que \mathcal{C} est libre (laissé en exercice), donc \mathcal{C} est une autre base de E.
La matrice de passage P de \mathcal{B} à \mathcal{C} est donc : P=Mat_{\mathcal{B}}(\mathcal{C})=\begin{pmatrix}0&1&-1\\-1&-1&1\\1&1&-2 \end{pmatrix}.
Son inverse P^{-1} peut etre calculé en écrivant les éléments de la base \mathcal{B} en fonction des éléments de la base \mathcal{C} : \begin{cases}e_{1}=-u_{1}+u_{2}  \\ e_{2}=-u_{1}-u_{2}-u_{3}  \\e_{3}=-u_{2}-u_{3}  \end{cases}, on a alors: P^{-1}=\begin{pmatrix}-1&-1&0\\1&-1&-1\\0&-1&-1 \end{pmatrix}.
Formons maintenant la matrice de f dans la base \mathcal{C} en calculant f(u_{1}) , f(u_{2}) et f(u_{3})
On obtient aisément que: f(u_{1})=-u_{1}, f(u_{2})=2u_{2} et f(u_{3})=3u_{3} (en utilisant le calcul matriciel Y=AX : laissé en exercice)
la matrice de f dans la base \mathcal{C} qu'on notera D est alors: D=Mat_{\mathcal{C}}(f)=diag(-1,2,3)
On a, par formule de changement de base : A=PDP^{-1}, ce qui permet le calcul des puissances de la matrice A car:
\forall n\in\mathbb{N},
A^{n}=(PDP^{-1})^{n}=\underbrace{PDP^{-1}.PDP^{-1} . PDP^{-1} \cdots PDP^{-1}}_{\text{n fois}}=PDI_{n}DI_{n}DI_{n} \cdots I_{n}DP^{-1}=PD^{n}P^{-1}
Or, D^{n} = (diag(-1,2,3))^{n}=diag((-1)^{n},2^{n},3^{n}).
On obtient enfin:
\forall n\in\mathbb{N} : A^{n}=\begin{pmatrix}0&1&-1\\-1&-1&1\\1&1&-2 \end{pmatrix}\begin{pmatrix}(-1)^{n}&0&0\\0&2^{n}&0\\0&0&3^{n} \end{pmatrix}\begin{pmatrix}-1&-1&0\\1&-1&-1\\0&-1&-1 \end{pmatrix}=(-1)^{n}\begin{pmatrix}0&0&0\\1&1&0\\-1&-1&0 \end{pmatrix}+2^{n}\begin{pmatrix}1&-1&-1\\-1&1&1\\1&-1&-1 \end{pmatrix}+3^{n}\begin{pmatrix}0&1&1\\0&-1&-1\\0&2&2 \end{pmatrix}

3- La trace

a- Trace d'une matrice carrée
Définition :
On appelle trace d'une matrice carrée A=(a_{ij})\in\mathcal{M}_{n}(\mathbb{K}) le scalaire: tr(A)=\displaystyle\sum_{i=1}^{n}a_{ii}



Exemples:
\forall n\in\mathbb{N}^{*} : tr(I_{n})=n
Soit A=\begin{pmatrix}2&1&-1\\-1&3&1\\1&0&-2 \end{pmatrix}, tr(A)=3

Proposition :
Pour toutes matrices carrées A et B de même ordre et pour tous scalaires a,b \in\mathbb{K} :
tr(aA+bB)=atr(A)+btr(B)
tr(^{t}A)=tr(A)
Autrement dit, l'application tr: \mathcal{M}_{n}(\mathbb{K}) \longrightarrow \mathbb{K} est une forme linéaire invariante par transposition.



Théorème :
\forall A\in\mathcal{M}_{n,p}(\mathbb{K}) , \mathcal{M}_{p,n}(\mathbb{K}) : tr(AB)=tr(BA)


Proposition :
\forall (A,B)\in (\mathcal{M}_{n}(\mathbb{K}))^{2} : A\sim B \Longrightarrow tr(A)=tr(B)


b- Trace d'un endomorphisme
Définition :
Soient E un \mathbb{K}-ev de dimension finie et f\in\mathcal{L}(E)
La trace de f est la trace d'une matrice représentative de f dans une base quelconque de E


En effet, Soient \mathcal{B} et \mathcal{B^{'}} deux bases de E.
Notons, P=Mat_{\mathcal{B}}(\mathcal{B^{'}}) , A=Mat_{\mathcal{B}}(f) et A^{'}=Mat_{\mathcal{B^{'}}}(f)
On sait que : A=PA^{'}P^{-1}
On a alors : tr(A)=tr(PA^{'}P^{-1})=tr(P(A^{'}P^{-1}))=tr((A^{'}P^{-1})P)=tr(A^{'})
Donc, la trace de la matrice représentative de l'endomorphisme f est indépendante de la base choisie.

Théorème :
La trace définit une forme linéaire sur \mathcal{L}(E), vérifiant : \forall (f,g)\in(\mathcal{L}(E))^{2} : tr(fog)=tr(gof)



IV- Rang d'une matrice

1- Définition et propriétés

Définition :
Soit A\in\mathcal{M}_{n,p}(\mathbb{K}). On appelle rang de A, et on note rg(A) le rang de la famille des colonnes de A dans \mathcal{M}_{n,1}(\mathbb{K})


Théorème :
Si \mathcal{F}=(x_{1},\cdots, x_{p}) une famille de vecteurs d'un \mathbb{K}-ev E et si A est la matrice de \mathcal{F} dans une certaine base \mathcal{B} de E alors : rg\mathcal{F} = rg(A)


Théorème :
Si u est une application linéaire d'un \mathbb{K}-ev E vers un K-ev F et si A est la matrice de u relative à des bases \mathcal{B} et \mathcal{C} respectivement de E et F alors: rg(A)=rg(u).


Proposition :
\forall A\in\mathcal{M}_{n,p}(\mathbb{K}) : rg(A)\leq Min(n,p)
\forall A\in\mathcal{M}_{n}(\mathbb{K}) : rg(A)=n \Longleftrightarrow A\in \mathcal{GL}_{n}(\mathbb{K})
\forall A\in\mathcal{M}_{n,p}(\mathbb{K}) : \begin{cases}\forall P\in \mathcal{GL}_{p}(\mathbb{K}), \text{   } rg(AP)=rg(A)  \\ \forall Q\in \mathcal{GL}_{n}(\mathbb{K}), \text{   } rg(QA)=rg(A)   \end{cases}


Proposition :
Soient A\in\mathcal{M}_{n,p}(\mathbb{K}) , r=rg(A). Alors A est équivalente à la matrice J_{n,p,r} définie par:
J_{n,p,r}=\begin{pmatrix}I_{r}&O_{r,p-r}\\O_{n-r,r}&O_{n-r,p-r} \end{pmatrix}


Remarque :
J_{n,p,0}=0

Proposition :
\forall (A,B)\in (\mathcal{M}_{n,p}(\mathbb{K}))^{2} : AeqB\Longleftrightarrow rg(A)=rg(B)


Proposition :
\forall A\in \mathcal{M}_{n,p}(\mathbb{K}) : rg(^{t}A)=rg(A)



2- Opérations élémentaires sur les matrices

Soient (n,p)\in (\mathbb{N}-\lbrace 0,1 \rbrace)^{2} , A=(a_{ij})\in \mathcal{M}_{n,p}(\mathbb{K})
On appelle opérations élémentaires sur les colonnes de A les transformations suivantes (où C_{j} désigne la j^{\text{ème}} colonne de A, 1\leq j\leq p) :
Échange de deux colonnes de A entres elles.
Remplacement d'une colonne C_{j} de A par aC_{j}a\in\mathbb{K}^{*}.
Remplacement d'une colonne C_{j} de A par C_{j}+aC_{k}a\in\mathbb{K} et k\in\lbrace 1,\cdots,p\rbrace -\lbrace j \rbrace.
On définit de manière analogue les opérations élémentaires sur les lignes de A( qui sont les opérations élémentaires sur les colonnes de la transposée de A).

Proposition :
Soit A\in \mathcal{M}_{n,p}(\mathbb{K}).
Les opérations élémentaires sur les colonnes et sur les lignes de A conservent le rang de A


Exemples :
A=\begin{pmatrix}1&2&1\\1&3&2\\1&1&0\end{pmatrix}
rg(A)=rg\begin{pmatrix}1&2&1\\1&3&2\\1&1&0\end{pmatrix}=rg\begin{pmatrix}1&2&1\\0&0&1\\0&-1&-1\end{pmatrix}=rg\begin{pmatrix}1&2&1\\0&0&1\\0&0&0\end{pmatrix}=2
Opérations effectuées:
Opération 1 : L_{2}\longrightarrow L_{2}-L_{1} et L_{3}\longrightarrow L_{3}-L_{1}
Opération 2 : L_{3}\longrightarrow L_{3}+L_{2}


B=\begin{pmatrix}1&1&1\\-1&1&-1\\2&1&1\\1&2&-1\end{pmatrix}
rg(B)=rg\begin{pmatrix}1&1&1\\-1&1&-1\\2&1&1\\1&2&-1\end{pmatrix}=rg\begin{pmatrix}1&0&0\\-1&2&0\\2&-1&-1\\1&1&-2\end{pmatrix}=3
Opération effectuée:
C_{2}\longrightarrow C_{2}-C_{1} et C_{3}\longrightarrow C_{3}-C_{1}


Remarque :
Les opérations élémentaires doivent être effectuées successivement et non simultanément.

Contre-exemple :
I_{2}=\begin{pmatrix}1&0\\0&1\end{pmatrix} devient, en effectuant L_{1} \longrightarrow L_{1}+L_{2} et L_{2}\longrightarrow L_{2}+L_{1} :
Simultanément : \begin{pmatrix}1&1\\1&1\end{pmatrix} (Faux)
Successivement : \begin{pmatrix}2&1\\1&1\end{pmatrix}

3- Méthode de Gauss

Soit A\in\mathcal{M}_{n,p}(\mathbb{K}).
Si la 1^{\text{ère}} ligne de A est nulle, la matrice de \mathcal{M}_{n-1,p}(\mathbb{K}) obtenue en supprimant dans A la 1^{\text{ère}} ligne a le même rang que A. On peut donc supposer que la 1^{\text{ère}} ligne de A est non nulle.
Par permutation de colonnes, on se ramène à une matrice de même rang que A, et dont le (1,1)^{\text{ème}} terme est non nul. En multipliant la 1^{\text{ère}} colonne par l'inverse de cet élément, on se ramène à une matrice A_{1}=(a_{ij})_{ij} telle que a_{11}=1.
Pour chaque j de \lbrace 2,\cdots, p \rbrace, le remplacement de la colonne C_{j} par C_{j}-a_{1j}C_{1} fait apparaître une matrice A_{2}, de même rang que A, et dont la 1^{\text{ère}} ligne est (1,0,\cdots,0) :
A_{1}=\begin{pmatrix}1&a_{12}&a_{13}&\cdots&a_{1p}\\a_{21}&a_{22}&a_{23}&\cdots&a_{2p}\\ \vdots&\vdots&\vdots& &\vdots\\a_{n1}&a_{n2}&a_{n3}&\cdots&a_{np} \end{pmatrix} \longrightarrow A_{2}=\begin{pmatrix}1&0&0&\cdots&0\\a_{21}&a_{22}-a_{12}a_{21}&a_{23}-a_{13}a_{21}&\cdots&a_{2p}-a_{1p}a_{21}\\ \vdots&\vdots&\vdots& &\vdots\\a_{n1}&a_{n2}-a_{12}a_{n1}&a_{n3}-a_{13}a_{n1}&\cdots&a_{np}-a_{1p}a_{n1}\end{pmatrix}

En réitérant le procédé sur la matrice à n-1 lignes et p-1 colonnes située en bas à droite dans A_{2}, on arrive, au bout d'un nombre fini d'opérations élémentaires sur les colonnes de A et suppressions d'éventuelles lignes ou colonnes nulles, à une matrice R (qui a donc le même rang que A) de la forme:

R=\displaystyle\begin{pmatrix}1& &O\\ &\ddots &\\ & &1\\  &\cdots & \end{pmatrix}
Il est clair que, puisque les colonnes de R forment une famille libre, le rang de R est le nombre de colonnes de R (qui n'est pas nécessairement le nombre de colonnes de A)

Utilisation : (Calcul de l'inverse)
Soit A\in \mathcal{GL}_{n}(\mathbb{K}). On cherche à calculer A^{-1}. On effectue pour cela, en suivant la méthode de Gauss :
[puce ]Des opérations élémentaires uniquement sur les lignes ( ou uniquement les colonnes) de A qui transforme A en I_{n}.
Ces mêmes opérations à I_{n}, la matrice obtenue est A^{-1}.


Exemple :
A=\begin{pmatrix}1&1&-1\\1&1&0\\2&1&1\end{pmatrix}

\left(\begin{array}{ccc|ccc} 1&1&-1 &1&0&0\\ 1&1&0 &0&1&0\\ 2&1&1 &0&0&1\\ \end{array}\right)

\left(\begin{array}{ccc|ccc} 1&1&-1 &1&0&0\\ 0&0&1 &-1&1&0\\ 0&-1&3 &-2&0&1\\ \end{array}\right)

\left(\begin{array}{ccc|ccc} 1&1&-1 &1&0&0\\ 0&-1&3 &-2&0&1\\ 0&0&1 &-1&1&0\\ \end{array}\right)

\left(\begin{array}{ccc|ccc} 1&1&-1 &1&0&0\\ 0&1&-3 &2&0&-1\\ 0&0&1 &-1&1&0\\ \end{array}\right)

\left(\begin{array}{ccc|ccc} 1&1&0 &0&1&0\\ 0&1&0 &-1&3&-1\\ 0&0&1 &-1&1&0\\ \end{array}\right)

\left(\begin{array}{ccc|ccc} 1&0&0 &1&-2&1\\ 0&1&0 &-1&3&-1\\ 0&0&1 &-1&1&0\\ \end{array}\right)
On en déduit que : A^{-1}=\begin{pmatrix}1&-2&1\\-1&3&-1\\-1&1&0\end{pmatrix}
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 1675 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 !