- Endomorphisme de $ \mathcal{M}_2(\mathbb{R}) $
Soient $ M = \begin{pmatrix} a & b \\ c & d \end{pmatrix} $ et $ N = \begin{pmatrix} a' & b' \\ c' & d' \end{pmatrix} $ deux matrices de $ \mathcal{M}_2(\mathbb{R}) $, et $ \lambda \in \mathbb{R} $.
On remarque que $ a+d = \text{Tr}(M) $, la trace de la matrice. L'application peut donc s'écrire $ f(M) = M + \text{Tr}(M)I_2 $. \[ f(\lambda M + N) = (\lambda M + N) + \text{Tr}(\lambda M + N)I_2 \] Par linéarité de la trace : \[ f(\lambda M + N) = \lambda M + N + (\lambda \text{Tr}(M) + \text{Tr}(N))I_2 \] \[ f(\lambda M + N) = \lambda(M + \text{Tr}(M)I_2) + (N + \text{Tr}(N)I_2) \] \[ f(\lambda M + N) = \lambda f(M) + f(N) \] $ f $ est une application linéaire et ses valeurs sont dans $ \mathcal{M}_2(\mathbb{R}) $. C'est donc un endomorphisme de $ \mathcal{M}_2(\mathbb{R}) $. - Matrice $ A $ dans la base canonique
Ăvaluons $ f $ sur chaque vecteur de la base canonique $ (J_1, J_2, J_3, J_4) $ :- $ f(J_1) = J_1 + 1 \cdot I_2 = J_1 + (J_1 + J_4) = 2J_1 + J_4 $
- $ f(J_2) = J_2 + 0 \cdot I_2 = J_2 $
- $ f(J_3) = J_3 + 0 \cdot I_2 = J_3 $
- $ f(J_4) = J_4 + 1 \cdot I_2 = J_4 + (J_1 + J_4) = J_1 + 2J_4 $
-
- Nouvelle base de $ \mathcal{M}_2(\mathbb{R}) $
Notons $ \mathcal{B}' = (J_1-J_4, J_2, J_3, I_2) $. Sachant que $ I_2 = J_1 + J_4 $, Ă©crivons la matrice de passage $ P $ de la base canonique vers cette nouvelle famille : \[ P = \begin{pmatrix} 1 & 0 & 0 & 1 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ -1 & 0 & 0 & 1 \end{pmatrix} \] Calculons son dĂ©terminant (par exemple en dĂ©veloppant par rapport aux colonnes centrales) : \[ \det(P) = 1 \times 1 - (-1) \times 1 = 2 \] Puisque $ \det(P) \neq 0 $, la matrice $ P $ est inversible. La famille est donc libre. Ătant constituĂ©e de $ 4 $ vecteurs dans un espace de dimension $ 4 $, c'est une base de $ \mathcal{M}_2(\mathbb{R}) $. - Matrice $ D $ de $ f $ dans cette base
Ăvaluons l'image des vecteurs de $ \mathcal{B}' $ par l'endomorphisme $ f $ :- $ f(J_1-J_4) = f(J_1) - f(J_4) = (2J_1 + J_4) - (J_1 + 2J_4) = J_1 - J_4 $
- $ f(J_2) = J_2 $
- $ f(J_3) = J_3 $
- $ f(I_2) = I_2 + \text{Tr}(I_2)I_2 = I_2 + 2I_2 = 3I_2 $
- Nouvelle base de $ \mathcal{M}_2(\mathbb{R}) $
- Automorphisme de $ \mathcal{M}_2(\mathbb{R}) $
La matrice $ D $ représentant $ f $ dans la base $ \mathcal{B}' $ est diagonale et tous ses coefficients diagonaux sont non nuls ($ \det(D) = 3 \neq 0 $). La matrice $ D $ (et donc $ A $) est inversible. L'endomorphisme $ f $ est par conséquent bijectif : c'est un automorphisme de $ \mathcal{M}_2(\mathbb{R}) $.