Inverser \[A=\left(\begin{array}{ccc} -3&1&0\\ 2&0&1\\ 1&2&-1 \end{array} \right)\]
en utilisant des opérations élémentaires sur les lignes et les colonnes.
en utilisant la comatrice.
On effectue les mêmes opérations sur les lignes de \(A\) et de \(I_3\).
Chaque opération sur les lignes est la multiplication à gauche par une matrice inversible. Donc \(A\) est inversible et
On calcule \(\mathop{\rm det}(A)=9\) et \(\mathop{\mathrm{com}}(A)=\begin{pmatrix} -2&3&4\\1&3&7\\1&3&-2 \end{pmatrix}\) donc \(A^{-1}=\dfrac{{\mathop{\mathrm{com}}(A)}^{\mathrm{T}}}{\mathop{\rm det}(A)}=\dfrac19 \begin{pmatrix} -2 & 1 & 1 \\ 3 & 3 & 3 \\ 4 & 7 & -2 \end{pmatrix}\).
Montrer que les matrices suivantes sont inversibles et calculer leur inverse :
\(A=\left(\begin{array}{cc} -1&1\\ 2&1 \end{array} \right)\)
\(B=\left(\begin{array}{cc} 1&i\\ i&1 \end{array} \right)\)
\(C=\left( \begin {array}{ccc} 1&0&1\\-1&0&1 \\0&1&-1\end {array} \right)\)
\(D=\left( \begin {array}{ccc} 0&-2&1\\1&0&1 \\0&-1&2\end {array} \right)\)
\(E=\left( \begin {array}{ccc} 1&1&0\\0&1&2 \\0&0&1\end {array} \right)\)
\(F=\left( \begin {array}{cccc} 0&1&0&0\\0&0&1&0 \\0&0&0&1\\1&0&0&0\end {array} \right)\)
\(A^{-1}= \left( \begin {array}{cc} -1/3&1/3\\2/3&1/3 \end {array} \right)\)
\(B^{-1}=\left( \begin {array}{cc} 1/2&-i/2\\-i/2&1/2 \end {array} \right)\)
\(C^{-1}=\left( \begin {array}{ccc} 1/2&-1/2&0\\1/2&1/2&1 \\1/2&1/2&0\end {array} \right)\)
\(D^{-1}=\left( \begin {array}{ccc} 1/3&1&-2/3\\-2/3&0&1/3 \\-1/3&0&2/3\end {array} \right)\)
\(E^{-1}= \left( \begin {array}{ccc} 1&-1&2\\0&1&-2 \\0&0&1\end {array} \right)\)
\(F^{-1}=\left( \begin {array}{cccc} 0&0&0&1\\1&0&0&0 \\0&1&0&0\\0&0&1&0\end {array} \right)\)
On considère la matrice \(A\in\mathfrak{M}_{2}\left(\mathbb{C}\right)\) donnée par \(A=\left(\begin{array}{cc} 1&2i \\ -i&1 \end{array}\right)\).
Montrer que \(A^2-2A-I_2=0\). On dit que \(X^2-2X-1\) est un polynôme annulateur de \(A\).
En déduire que \(A\) est inversible et calculer son inverse.
Retrouver ce résultat par un calcul direct.
On montre par un calcul direct que \(A^2-2A-I_2=0\).
L’égalité précédente amène : \(A\left(A-2I_2\right)=I_2\). \(A\) est donc inversible et sa matrice inverse est : \(\left( \begin {array}{cc} -1&2\,i\\-i&-1 \end {array} \right)\)
On retrouve ce résultat en utilisant la comatrice ou en passant par un système.
Soit \(A=\left( \begin {array}{ccc} 1&0&1\\-1&0&2 \\0&1&-1\end {array} \right)\)
Montrer que le polynôme que \(P=X^3-3X+3\) est un polynôme annulateur de \(A\).
En déduire que \(A\) est inversible et calculer son inverse.
Retrouver ce résultat par un calcul direct.
Même déroulement que l’exercice précédent. On trouve \(A^{-1} = I_3 - \dfrac13 A^2 =\dfrac{1}{3}\left(\begin{array}{ccc} 2&-1&0\\1&1&3\\1&1&0 \end{array}\right)\).
Soient deux matrices carrées \(A,B \in \mathfrak{M}_{n}(\mathbb{\mathbb{K} })\) vérifiant \(AB=0\). Montrer que si \(A\) est inversible, alors \(B = 0\).
Si \(A\) est inversible alors on peut écrire : \[AB=0 \Rightarrow A^{-1}\times AB= A^{-1} \times 0 \Rightarrow B=0.\]
Soit une matrice \(A \in \mathfrak{M}_{n}(\mathbb{\mathbb{R} })\) antisymétrique. On pose \(M=I+A\).
Soit une matrice colonne \(X\in\mathfrak{M}_{n,1}(\mathbb{\mathbb{R} })\). Calculer la matrice \({X}^{\mathrm{T}}AX\)
En déduire que la matrice \(M\) est inversible.
Si \(A=\left(a_{ij}\right)\) alors \({X}^{\mathrm{T}}AX=\sum_{i=1}^n \sum_{j=1}^n a_{ij} xi x_j\). Comme \(A\) est antisymétrique, pour tout \(i\neq j\) \(a_{ji} xi x_j=-a_{ij} xi x_j\) et \(a_{ii}=0\). Alors \(\boxed{{X}^{\mathrm{T}}AX=0}\).
Soit \(X\in \mathfrak{M}_{n,1}(\mathbb{\mathbb{R} })\). Alors \[MX=0 \Rightarrow {X}^{\mathrm{T}} MX=0 \Rightarrow {X}^{\mathrm{T}} \left(I+A\right)X=0 \Rightarrow {X}^{\mathrm{T}} X+ {X}^{\mathrm{T}} AX =0 \Rightarrow {X}^{\mathrm{T}} X=0\] en vertu de la première question. Mais si \(X=\left(x_i\right)\), \({X}^{\mathrm{T}} X=\sum_{i=1}^n {x_i^2}\) et \({X}^{\mathrm{T}} X=0\) implique que \(\forall i\in\llbracket 1,n\rrbracket\), \(x_i=0\) et que \(X=0\). On en déduit que \(M\) est inversible.
Déterminer l’inverse de la matrice carrée \(A= \begin{pmatrix} 1 & & &\mathbb{O}\\ a & \ddots & & \\ & \ddots & \ddots & \\ \mathbb{O}& & a &1 \end{pmatrix}\)
Soit \(e=\left(e_1,\dots,e_n\right)\) une base de \(\mathbb{K}^n\) et \(f=\left(f_1,\dots,f_n\right)\) la famille de vecteurs de \(\mathbb{K}^n\) admettant \(A\) comme matrice dans la base \(e\). On a donc :\[f_1=e_1+a e_2,\, f_2=e_2+a e_3,\, \dots,\, f_{n-2}=e_{n-2}+a e_{n-1},\, f_{n-1}=e_{n-1}+ a e_n,\, f_n=e_n .\] On en déduit que : \[\begin{aligned} e_n&=&f_n\\ e_{n-1}&=&f_{n-1}-a f_n\\ e_{n-2}&=&f_{n-2}-a f_{n-1}+a^2 f_n\\ \vdots &=& \vdots\\ e_2&=&f_2-af_3+a^2 f_4-a^3 f_5+\dots+\left(-1\right)^{n-2} a^{n-2}f_n\\ e_1&=&f_1-af_2+a^2 f_3-a^3 f_4+\dots+\left(-1\right)^{n-1}a^{n-1}f_n\\\end{aligned}\] donc \(f\) est une base de \(\mathbb{K}^n\), \(A\) est inversible et \[A^{-1}=\textrm{ Mat}_{f}\left(e\right)=\begin{pmatrix} 1&0&\dots&\dots&0\\ -a&1&\vdots&\vdots&\vdots\\ a^2&-a&\ddots&&\\ -a^3&a^2&&\ddots&\\ \vdots &\vdots&&&\vdots\\ (-1)^{n-2}a^{n-2} &(-1)^{n-3}a^{n-3}&\dots&1&0\\ (-1)^{n-1}a^{n-1} &(-1)^{n-2}a^{n-2}&\dots&-a&1 \end{pmatrix}\]
Soit \(A\in\mathfrak{M}_{n}\left(\mathbb{K}\right)\) telle que \(I_n+A\) est inversible. Soit \(B=\left(I_n-A\right)\left(I_n+A\right)^{-1}\).
Montrer que \(B=\left(I_n+A\right)^{-1}\left(I_n-A\right)\).
Montrer que \(I_n+B\) est inversible et exprimer \(A\) en fonction de \(B\).
On a \(\left(I_n+A\right)\left(I_n-A\right) = \left(I_n-A\right)\left(I_n+A\right)\), donc \(I_n-A = \left(I_n+A\right)^{-1}\left(I_n-A\right)\left(I_n-A\right)\) donc \(B = \left(I_n-A\right) \left(I_n+A\right)^{-1}= \left(I_n+A\right)^{-1}\left(I_n-A\right)\), ce qu’il fallait vérifier.
\(B = \left(2I_n -I_n-A\right)\left(I_n+A\right)^{-1} = 2\left(I_n+A\right)^{-1} - \left(I_n+A\right)\left(I_n+A\right)^{-1} = 2\left(I_n+A\right)^{-1} - I_n\). D’où \(I_n+B = 2\left(I_n+A\right)^{-1}\), ce qui montre bien que \(I_n+B\) est inversible. De plus \(\left(I_n+B\right)^{-1} = \dfrac12 \left(I_n+A\right)\), d’où \(A = 2\left(I_n+B\right)^{-1} - I_n\).
Soit \(A\) une matrice carrée nilpotente de taille \(n\in\mathbb{N}^*\). Montrer que la matrice \((I_n-A)\) est inversible.
On suppose que \(A\) est nilpotente d’ordre \(p\in\mathbb{N}^*\). On a donc \(A^k=0\) pour tout \(k\geqslant p\) et \(A^k\neq 0\) pour tout \(k<p\). Mais \[\left(I_n-A\right)\left(I_n+A+A^2+\dots+A^{p-1}\right)=\left(I_n-A\right)+\left(A-A^2\right)+\dots+\left(A^{p-1} -A^p \right) = I_n\] par télescopage et car \(A^p=0\). Donc \(I_n-A\) est inversible d’inverse \(I_n+A+A^2+\dots+A^{p-1}\).
Soit une matrice \(U\) triangulaire supérieure telle que tous les éléments de la diagonale soient non-nuls. Montrer que la matrice \(U\) est inversible. (On montrera que \(UX=0\Longrightarrow X=0\) )
Soit une matrice colonne \(X\in \mathfrak{M}_{n,1}(\mathbb{R})\) telle que \(UX=0\). On obtient un système d’équations triangulaire sur les coordonnées de \(X\) qui se résout de proche en proche en partant de la dernière équation et on obtient finalement que \(X=0\). Par conséquent, \(U\) est inversible.
Une autre façon de voir est de considérer que la matrice est celle d’un endomorphisme \(u\) de \(\mathbb{R}_{n-1}[X]\) dans la base \((X^k)\). L’hypothèse \(U\) triangulaire supérieure avec tous les éléments de la diagonale non-nuls se traduit par : \(\forall k\in \{0,\ldots,n-1 \},\deg(u(X^k)) = k\). Donc \(u\) transforme la base \((X^k)\) en une famille échelonnée en degrés, donc une base de \(\mathbb{R}_{n-1}[X]\). Donc \(u\) est bijectif, donc \(U\) est inversible.
On considère la matrice \(M = ((m_{i,j}))_{1\leqslant i, j \leqslant n} \in \mathfrak{M}_{n}(\mathbb{\mathbb{R} })\) avec \(m_{ij} = \begin{cases} \dfrac{i}{j} & \textrm{ si } i \neq j \\ 0 & \textrm{ si } i = j \end{cases}\). Calculer \(M^2\) et \(M^{-1}\).
Faire d’abord le calcul pour \(n = 3\). En notant \(M^2 = ((a_{i,j}))\), on tire que \(a_{i,j} = \sum_{k=1}^n m_{i,k}m_{k,j} = \sum_{k\not\in \{i, j\}} \dfrac{i}{j}\) \[a_{ij} = \begin{cases} (n-2)\dfrac{i}{j} & \textrm{ si } i \neq j \\ (n-1) & \textrm{ si } i = j \end{cases}\] On remarque que \(M^2 = (n-2)M + (n-1)I_n\) d’où l’on tire que \(M\) est inversible et que \(M^{-1} = \dfrac{1}{n-1}.(M - (n-2) I_n)\).
Inverser la matrice suivante :
\[A =\begin{pmatrix} 1&6&5&4&3&2\\ 2&1&6&5&4&3\\ 3&2&1&6&5&4\\ 4&3&2&1&6&5\\ 5&4&3&2&1&6\\ 6&5&4&3&2&1 \end{pmatrix}\]
Avec la forme de la matrice \(A\), on peut faire le pari que la matrice inverse peut s’écrire, si elle existe, \[A^{-1} =\begin{pmatrix} x&y&z&t&u&v\\ v&x&y&z&t&u\\ u&v&x&y&z&t\\ t&u&v&x&y&z\\ z&t&u&v&x&y\\ y&z&t&u&v&x \end{pmatrix}.\] L’égalité \(AA^{-1}=I_6\) se traduit, pour la première colonne de \(AA^{-1}\), par \[\left\lbrace \begin{array}{rrrrrrrr} x&+6v&+5u&+4t&+3z&+2y&=&1\\ 2x&+v&+6u&+5t&+4z&+3y&=&0\\ 3x&+2v&+u&+6t&+5z&+4y&=&0\\ 4x&+3v&+2u&+t&+6z&+5y&=&0\\ 5x&+4v&+3u&+2t&+z&+6y&=&0\\ 6x&+5v&+4u&+3t&+2z&+y&=&0 \end{array}\right.\] Il n’est pas difficile de voir que la (les ?) solutions de ce système fournit une solution pour les autres colonnes.
On considère une matrice \(A \in \mathfrak{M}_{n}(\mathbb{\mathbb{C} })\), \(A = (a_{i,j})_{1\leqslant i, j \leqslant n}\) à diagonale dominante : \[\forall i \in [\kern-0.127em[ 1, n ]\kern-0.127em]^2,\quad \lvert a_{ii} \rvert > \sum_{j \neq i } \lvert a_{ij} \rvert\] Montrer que la matrice \(A\) est inversible. Cette propriété est connue sous le nom de lemme d’Hadamard.
Supposons que \(A\) ne soit pas inversible. Alors il existe \(X=\left(x_i\right)\in \mathfrak{M}_{n,1}(\mathbb{\mathbb{C} })\) non nul tel que \(AX=0\). Comme \(X\) est non nul, on sait que \(\alpha=\max_{i\in\llbracket 1,n\rrbracket} \left|x_i\right| \neq 0\). De plus, on a encore \(A\left(X/\alpha\right)=0\). On peut donc supposer que \(\max_{i\in\llbracket 1,n\rrbracket} \left|x_i\right| =1\). On note \(i_0\) l’indice \(i\in\llbracket 1,n\rrbracket\) tel que \(\left|x_i\right|=1\) . L’égalité \(AX=0\) amène pour tout \(i\in\llbracket 1,n\rrbracket\), \(\sum_{j=1}^n a_{ij} x_j=0\) ou encore \(-a_{ii}x_i =\sum_{j=1,j\neq i}^n a_{ij} x_j\). En particulier, pour \(i=i_0\), cette égalité devient en passant à la valeur absolue et en utilisant l’inégalité triangulaire : \[\left|a_{i_0i_0}\right|=\left|a_{i_0i_0} x_{i_0}\right|= \left|\sum_{j=1,j\neq i_0}^n a_{i_0j}x_j\right|\leqslant\sum_{j=1,j\neq i_0}^n \left|a_{i_0j}\right|\left|x_j\right|\leqslant \sum_{j=1,j\neq i_0}^n \left|a_{i_0j}\right|\] car \(\max_{i\in\llbracket 1,n\rrbracket} \left|x_i\right| =1\). Mais comme la matrice est à diagonale dominante, on a aussi \(\lvert a_{i_0i_0} \rvert > \sum_{j \neq i_0 } \lvert a_{i_0j} \rvert\) et on aboutit à une contradiction. Par conséquent \(A\) est inversible.
Soit \(n\in\mathbb N\), Démontrer qu’il existe un unique polynôme \(U_n(X)\in\mathbb Z[X]\), de degré \(n\), vérifiant \[\forall \vartheta \in \mathbb R,\; \sin \vartheta\,. U_n(\cos \vartheta) = \sin {(n+1)\vartheta}.\] Démontrer que \[U_{n+1}(X) + U_{n-1}(X) = 2XU_n(X)\qquad\qquad(*)\] Démontrer que \[\forall n\geqslant1,\;\forall p\geqslant1,\; U_{n+p} = U_p U_n - U_{p-1} U_{n-1}\qquad\qquad(**).\]
Soit \[B_n(x) = \begin{pmatrix} 2x & 1 & 0 & \ldots & 0 \\ 1 & 2x & 1 & \ddots & \vdots \\ 0 & \ddots & \ddots & \ddots & 0 \\ \vdots & \ddots & 1 & 2x & 1 \\ 0 & \ldots & 0 & 1 & 2x \\ \end{pmatrix}.\] Démontrer que pour \(x\neq \cos\dfrac{k\pi}{n+1}, B_n(x)\) est inversible et son inverse est la matrice symétrique définie par \[b'_{i,j} = (-1)^{i+j} \dfrac{U_{i-1}(x) U_{n-j}(x)}{U_n(x)} \textrm { pour } i\leqslant j\] et \[b'_{i,j} = (-1)^{i+j} \dfrac{U_{j-1}(x) U_{n-i}(x)}{U_n(x)} \textrm { pour } i\geqslant j.\]
Existence : On l’établit par récurrence. On peut prendre \(U_0(X) = 1, \quad U_1(X) = 2X\),
Unicité : La différence de deux tels polynômes s’annulerait sur \([-1;1]\) ce qui donne l’unicité.
Enfin, en utilisant trois fois \(\sin a\sin b = {\scriptstyle 1\over\scriptstyle 2} \left(\cos(a-b) - cos(a+b)\right)\), on a d’une part : \[\sin^2\vartheta U_{n+p}(\cos\vartheta) = \sin\vartheta\sin(n+p+1)\vartheta = {\scriptstyle 1\over\scriptstyle 2}\cos(n+p)\vartheta - {\scriptstyle 1\over\scriptstyle 2}\cos(n+p+2)\vartheta,\] et d’autre part : \[\begin{aligned} \sin^2\vartheta U_{p}(\cos\vartheta)U_{n}(\cos\vartheta)-\sin^2\vartheta U_{p-1}(\cos\vartheta)U_{n-1}(\cos\vartheta) &= \sin(p+1)\vartheta \sin(n+1)\vartheta - \sin p\vartheta \sin n\vartheta \\ &= \dfrac12\cos(p-n)\vartheta - \dfrac12\cos(p+n+2)\vartheta \\ & \qquad - \dfrac12\cos(p-n)\vartheta + \dfrac12\cos(p+n)\vartheta \\ &= \dfrac12\cos(p+n)\vartheta - \dfrac12\cos(p-n)\vartheta\end{aligned}\] On a bien établi (**)
Soit \(1\leqslant i<j\leqslant n\) (en posant \(U_{-1}(x) = 0\) dans le cas où \(j=n\)) \[\begin{aligned} \sum_{k=1}^n b'_{i,k}b_{k,j} &= b'_{i, j-1} b_{j-1, j} + b'_{i, j} b_{j, j} + b'_{i, j+1} b_{j+1, j} = b'_{i, j-1} b_{j-1, j} + 2xb'_{i, j} + b'_{i, j+1} \\ & = \dfrac{(-1)^{i+j-1}U_{i-1}U_{n-j+1} + (-1)^{i+j} 2x U_{i-1}U_{n-j} + (-1)^{i+j+1}U_{i-1}U_{n-j-1} }{U_n(x)}\\ &= (-1)^{i+j-1}\dfrac{U_{i-1}\left( U_{n-j+1} - 2x U_{n-j} + U_{n-j-1}\right) }{U_n(x)}\\ & = 0.\end{aligned}\] d’après \((*)\) où l’on remplace \(n\) par \(n-j\).
Pour \(i=j\) (en posant \(U_{-1}(x) = 0\) dans le cas où \(j=n\) et donc \(b'_{n, n+1} = 0\)) \[\begin{aligned} \sum_{k=1}^n b'_{i,k}b_{k,i} &= b'_{i, i-1} b_{i-1, i} + b'_{i, i} b_{i, i} + b'_{i, i+1} b_{i+1, i} = b'_{i, i-1} b_{i-1, i} + 2xb'_{i, i} + b'_{i, i+1} \\ &= -\dfrac{U_{i-2}U_{n-i} + 2x U_{i-1}U_{n-i} - U_{i-1}U_{n-i-1} }{U_n(x)}\\ &= \dfrac{1 }{U_n(x)}\left( -U_{i-2}U_{n-i} + U_{i-1}\left( 2x U_{n-i} + U_{n-i-1}\right)\right) \\ &= \dfrac{1 }{U_n(x)}\left( -U_{i-2}U_{n-i} + U_{i-1}U_{n-i+1}\right) = \dfrac{1 }{U_n(x)} U_{n-i+1+i-1} \textrm{ d'après } (**)\\ & = \dfrac{1 }{U_n(x)} U_n = 1.\end{aligned}\]
Inverser les matrices suivantes :
\(\begin{pmatrix}0&&{(1)}\\ &\ddots \\ {(1)}&&0\end{pmatrix}\)
\(\begin{pmatrix}a&&{(b)}\\&\ddots \\ {(b)}&&a\end{pmatrix}\)
\(\begin{pmatrix}1 &1 & &{(0)}\\ &\ddots &\ddots \\ & &\ddots &1 \\ {(0)}& & &1 \\\end{pmatrix}\)
\(\begin{pmatrix}1 &\bar\alpha &\bar\alpha ^2 \\ \alpha &1 &\bar\alpha \\ \alpha ^2 &\alpha &1 \\\end{pmatrix}\), \(\alpha \in \mathbb{C}\)
\(\begin{pmatrix}{(0)}& &a_n \\ &\ddots \\ a_{1} & &{(0)}\\\end{pmatrix}\)
\(\begin{pmatrix}1+\frac 1{\lambda _{1}} & &{(1)} \\ &\ddots \\ {(1)} & &1+\frac 1{\lambda _n} \\\end{pmatrix}\)
\(\dfrac{A + (2-n)I}{n-1}\).
\(\dfrac 1{(a-b)(a+(n-1)b)}\begin{pmatrix} {a+(n-2)b} & &{(-b)} \\ &\ddots \\ {(-b)} & &{a+(n-2)b} \\\end{pmatrix}\).
\(\begin{pmatrix}1 &-1 &1 &\dots&\pm 1 \\ &\ddots &\ddots &\ddots &\vdots \\ & &\ddots &\ddots &1 \\ & & &\ddots &-1 \\ {(0)}& & & &1 \\\end{pmatrix}\).
\(\dfrac 1{1-\alpha \bar\alpha }\begin{pmatrix}1 &-\bar\alpha &0 \\ -\alpha &1+\alpha \bar\alpha &-\bar\alpha \\ 0 &-\alpha &1 \\\end{pmatrix}\).
\(\begin{pmatrix}{(0)} & &1/a_n \\ &\ddots \\ 1/a_{1} & &{(0)} \\\end{pmatrix}\).
\(\mathop{\rm diag}\nolimits(\lambda _i) - \dfrac 1{1 + \lambda _{1} +\dots+ \lambda _n}(\lambda _i\lambda _j)\).
Soient \(A = \begin{pmatrix}1 &1/2 &1/3 \\ 1/2 &1/3 &1/4 \\ 1/3 &1/4 &1/5 \\\end{pmatrix}\) et \(B = \begin{pmatrix}1 &0.5 &0.33 \\ 0.5 &0.33 &0.25 \\ 0.33 &0.25 &0.20 \\\end{pmatrix}\). Calculer \(A^{-1}\) et \(B^{-1}\).
\(A^{-1} = \begin{pmatrix} 9 & -36 & 30 \\ -36 & 192 & -180 \\ 30 & -180 & 180 \\\end{pmatrix}\), \(B^{-1} = \frac{100}{63}\begin{pmatrix} 35 & -175 & 161 \\ -175 & 911 & -850 \\ 161 & -850 & 800 \\\end{pmatrix} \approx \begin{pmatrix} 55.6 & -277.8 & 255.6 \\ -277.8 & 1446.0 & -1349.2 \\ 255.6 & -1349.2 & 1269.8 \\\end{pmatrix}\).
Soit \(\mathcal D = \{ A=(a_{ij}) \in \mathfrak{M}_{n}(\mathbb{\mathbb{R}}) \text{ tq }\forall i,j\), \(a_{ij} \geq 0\) et \(\forall i\), \(\sum_{j=1}^n a_{ij} = 1\}\).
Montrer que \(\mathcal D\) est stable par multiplication.
Déterminer les matrices \(A \in \mathcal D\) inversibles telles que \(A^{-1} \in \mathcal D\).
Si \(A,B \in \mathcal D\) et \(AB = I\), alors pour \(i \neq j\) : \(\forall k\), \(a_{ik}b_{kj} = 0\).
Soit \(a_{i1} \neq 0\) : alors \(b_{1j} = 0\) pour tout \(j \neq i\), donc \(a_{i1} = b_{1i} = 1\).
Donc chaque colonne de \(A\) contient \(n-1\) fois \(0\) et une fois \(1\). \(A\) étant inversible, c’est une matrice de permutation.
Soit \(A = (a_{ij}) \in \mathfrak{M}_{n}(\mathbb{\mathbb{K}})\). On dit que \(A\) est centro-symétrique si pour tous \(i,j\) : \({a_{n+1-i,n+1-j} = a_{ij}}\). Montrer que si \(A\) et \(B\) sont centro-symétriques, il en est de même de \(AB\). Montrer que si \(A\) est centro-symétrique et inversible alors \(A^{-1}\) est aussi centro-symétrique.
Soit \(\mathbb{K}\) ayant au moins trois éléments et \(A \in \mathfrak{M}_{n}(\mathbb{\mathbb{K}})\). Montrer qu’il existe \(U,V \in GL_n(\mathbb{K})\) telles que \(A = U+V\). Donner un contre-exemple si \(\mathop{\rm card}\nolimits(\mathbb{K})=2\).
Existe-t-il une base de \(\mathfrak{M}_{n}(\mathbb{\mathbb{C}})\) constituée de matrices inversibles ?
Oui : \(I_n+E_{ij}\), \(i,j\in \llbracket 1,n\rrbracket\).
Soit \(M \in \mathfrak{M}_{n}(\mathbb{\mathbb{R}})\) antisymétrique.
Montrer que \(I+M\) est inversible (si \(MX = 0\), calculer \({ }^t(MX)(MX)\)).
Soit \(A = (I-M)(I+M)^{-1}\). Montrer que \({ }^tA = A^{-1}\).
Soit \(M \in \mathfrak{M}_{n}(\mathbb{\mathbb{Q}})\). Comparer les énoncés : \((1)\Leftrightarrow M\) est inversible dans \(\mathfrak{M}_{n}(\mathbb{\mathbb{Q}})\) et \((2)\Leftrightarrow M\) est inversible dans \(\mathfrak{M}_{n}(\mathbb{\mathbb{C}})\).
Soit \(A \in \mathcal M _{n+1}(\mathbb{Q})\) telle que \(a_{ij} = \binom{j-1}{i-1}\). Interpréter \(A\) comme la matrice d’un endomorphisme simple de \(\mathbb{Q}_n[X]\). En déduire la matrice \(A^{-1}\).
Soit \(A \in \mathfrak{M}_{n}(\mathbb{\mathbb{K}})\) telle que \(a_{ij} = (-1)^{n-j}\binom{n-j}{i-1}\).
Interpréter \(A\) comme la matrice d’un endomorphisme de \(\mathbb{K}_{n-1}[X]\).
En déduire \(A^3\).
\(\varphi (P) = (-X-1)^{n-1}P\left(-\dfrac 1{X+1}\right)\).
\(I\).
Que dire des morphismes de groupe \(\varphi :GL_n(\mathbb{R})\to \mathbb{Z}/p\mathbb{Z}\) ?
Soit \(\varphi\) un tel morphisme. Alors pour toute matrice \(M\in GL_n(\mathbb{R})\) on a \(\dot{0} = p\varphi (M) = \varphi (M^p)\), donc \(\varphi\) s’annule sur toute matrice qui est une puissance \(p\)-ème. Notons \(P(i,j,\alpha )\) la matrice de l’opération élémentaire \(L_i \leftarrow L_i + \alpha L_j\), qui est aussi la matrice de l’opération élémentaire \(C_j \leftarrow C_j + \alpha C_i\). Toute matrice \(M \in GL_n(\mathbb{R})\) peut être transformée, à l’aide de ces seules opérations élémentaires, en une matrice \(M' = \mathop{\rm diag}\nolimits(1,\dots,1,\det(M))\) par une adaptation de l’algorithme de Gauss. Comme \(P(i,j,\alpha ) = P(i,j,\alpha /p)^p\) et \(\det(M) = \pm (|\det(M)|^{1/p})^p\), on obtient : \(\varphi (M) = \dot{0}\) si \(\det(M) > 0\) et \(\varphi (M) = \varphi (\mathop{\rm diag}\nolimits(1,\dots,1,-1)) = x\) si \(\det(M) < 0\). Réciproquement, la fonction \(\varphi\) ainsi définie est effectivement un morphisme de groupe si et seulement si \(2x = \dot{0}\), soit \(x=\dot{0}\) pour \(p\) impair, et \(x\in \{ \dot{0},\dot q\}\) pour \(p=2q\).
Soit \(A\in \mathfrak{M}_{n}(\mathbb{\mathbb{R}})\) définie par \(A=\begin{pmatrix} 2 & -1 & & & (0) \\ -1 & 2 & -1 & & \\ & -1 & \ddots & \ddots & \\ & & \ddots & \ddots & -1 \\ (0) & & & -1 & 2 \\\end{pmatrix}\).
Montrer que, pour tout \(v\in \mathbb{R}^n\), si les coefficients de \(Av\) sont tous positifs alors les coefficients de \(v\) sont tous positifs.
Montrer que \(A\) est inversible et que son inverse est à coefficients positifs.
Soit \(p\) le plus petit indice tel que \(v_p=\mathop{\min}\limits_{1\leq i\leq n}v_i\). Si \(p=1\) on obtient \(2v_1\geq v_2\geq v_1\) et donc \(v_1\geq 0\). De même si \(p=n\). On suppose que \(2\leq p\leq n-1\). On a alors \(2v_p\geq v_{p-1}+v_{p+1}\geq 2v_p\) avec égalité si et seulement si \(v_{p-1}=v_{p+1}=v_p\), ce qui est impossible d’après la définition de \(p\). Donc \(v_p > 0\).
Soit \(v\) tel que \(Av=0\). D’après la question précédente on sait que, pour tout \(i\), \(v_i\geq 0\). Soit \(p\) le plus petit indice tel que \(v_p=\mathop{\max}\limits_{1\leq i\leq n}v_i\). Si \(p=1\) alors \(0=2v_1-v_2\geq v_1\) et donc \(v_1=0\) puis \(v=0\). De même si \(p=n\). On suppose que \(2\leq p\leq n-1\). On a alors \(2v_p=v_{p-1}+v_{p+1}\), ce qui entraîne \(v_{p-1}=v_p\) et contredit la définition de \(p\). On en déduit que \(v=0\) et donc que \(A\) est inversible. Soit \(v_i\) la \(i\)-ème colonne de \(A^{-1}\). Tous les coefficients de \(Av_i\) sont positifs, donc les coefficients de \(v_i\) sont tous positifs.