If $D \in \mathcal{M}_n(\mathbb{C})$ is a diagonal matrix whose diagonal coefficients are $d_0, \ldots, d_{n-1}$, show that $\|D\| = \max_{0 \leq i \leq n-1} |d_i|$.
Let $A, B \in \mathcal{M}_n(\mathbb{C})$. Suppose that there exists a unitary matrix $U \in \mathcal{M}_n(\mathbb{C})$ such that $B = UAU^{-1}$. Show that $\|A\| = \|B\|$.
Throughout this part, $\mathcal{A}$ is a subalgebra of $\mathcal{M}_{n}(\mathbb{R})$ strictly contained in $\mathcal{M}_{n}(\mathbb{R})$ and we denote by $d$ its dimension. The trace of any matrix $M$ of $\mathcal{M}_{n}(\mathbb{R})$ is denoted $\operatorname{tr}(M)$. Show that the map defined on $\mathcal{M}_{n}(\mathbb{R}) \times \mathcal{M}_{n}(\mathbb{R})$ by $(A, B) \mapsto \langle A \mid B \rangle = \operatorname{tr}(A^{\top} B)$ is an inner product on $\mathcal{M}_{n}(\mathbb{R})$.
For any matrix $B \in \mathcal { M } _ { N } ( \mathbb { R } )$, we set $\| B \| = \sup _ { \| x \| = 1 } \| B x \|$. After justifying the existence of $\| B \|$, show that $B \mapsto \| B \|$ is a norm on $\mathcal { M } _ { N } ( \mathbb { R } )$ satisfying $$\forall x \in \mathbb { R } ^ { N } \quad \| B x \| \leq \| B \| \| x \|$$
Let $A \in \mathcal { S } _ { N } ( \mathbb { R } )$ be a matrix with eigenvalues (not necessarily distinct) $\lambda _ { 1 } , \ldots , \lambda _ { N }$. Show that $$\| A \| = \max _ { 1 \leq i \leq N } \left| \lambda _ { i } \right|$$
Let $A \in \mathcal { S } _ { N } ^ { + } ( \mathbb { R } )$. For all $x \in \mathbb { R } ^ { N }$, we set $\| x \| _ { A } = \langle x , A x \rangle ^ { 1 / 2 }$. a) Show that the map $x \mapsto \| x \| _ { A }$ is a norm on $\mathbb { R } ^ { N }$. b) Show that there exist constants $C _ { 1 }$ and $C _ { 2 }$ strictly positive, which we will express in terms of the eigenvalues of $A$, such that $$\forall x \in \mathbb { R } ^ { N } \quad C _ { 1 } \| x \| \leq \| x \| _ { A } \leq C _ { 2 } \| x \| .$$
We keep the notations from Parts II and III. We denote $e _ { k } = x _ { k } - \tilde { x }$ and $e _ { 0 } = x _ { 0 } - \tilde { x }$. We recall that $I _ { N }$ is the identity matrix of order $N$. Show that $$\left\| e _ { k } \right\| _ { A } = \min \left\{ \left\| \left( I _ { N } + A Q ( A ) \right) e _ { 0 } \right\| _ { A } \mid Q \in \mathbb { R } [ X ] , \operatorname { deg } ( Q ) \leq k - 1 \right\}$$
We keep the notations from Parts II and III. We denote $e _ { k } = x _ { k } - \tilde { x }$ and $e _ { 0 } = x _ { 0 } - \tilde { x }$. We recall that $I _ { N }$ is the identity matrix of order $N$, and $\| \cdot \|$ denotes the matrix norm defined in question 2. Show that $$\left\| e _ { k } \right\| _ { A } \leq \left\| e _ { 0 } \right\| _ { A } \min \left\{ \left\| I _ { N } + A Q ( A ) \right\| \mid Q \in \mathbb { R } [ X ] , \operatorname { deg } ( Q ) \leq k - 1 \right\}$$ (One may use the properties of $A ^ { 1 / 2 }$ demonstrated in question 6.)
Throughout part II, $A \in \mathcal{M}_n(\mathbb{R})$ is a strictly positive matrix. Show that, for all $x \in \mathbb{R}^n$, $$\left\{\begin{array}{l} x \geqslant 0 \Longrightarrow Ax \geqslant 0 \\ x \geqslant 0 \text{ and } x \neq 0 \Longrightarrow Ax > 0. \end{array}\right.$$
Throughout part II, $A \in \mathcal{M}_n(\mathbb{R})$ is a strictly positive matrix satisfying $\rho(A) = 1$. We consider an eigenvalue $\lambda \in \mathbb{C}$ of $A$ with modulus 1 and $x$ an eigenvector associated with $\lambda$. We assume that $|x| < A|x|$. Show that there exists $\varepsilon > 0$ such that $A^2|x| - A|x| > \varepsilon A|x|$.
We set, for all $n \geq 0$ and all $x \in \mathbf{R}$, $P_n(x) = \sum_{k=0}^{n} \frac{x^k}{k!}$ where $k!$ denotes the factorial of $k$. Let $A \in \operatorname{Sym}^+(p)$. (a) Show that for all $(i,j) \in \llbracket 1,p \rrbracket^2$, we have $$\lim_{n \rightarrow +\infty} P_n[A]_{ij} = \exp\left(A_{ij}\right)$$ (b) Show that $\exp[A] \in \operatorname{Sym}^+(p)$. (c) Let $u \in \mathbf{R}^p$. Show that $\exp[A] \odot \left(uu^T\right) \in \operatorname{Sym}^+(p)$.
We denote $\mathcal{B}_{n}(\mathbb{R})$ the set of doubly stochastic matrices in $\mathcal{M}_{n}(\mathbb{R})$, that is the set of matrices $M = \left(m_{i,j}\right)_{1 \leqslant i,j \leqslant n}$ whose coefficients are all non-negative and such that $\sum_{j=1}^{n} m_{i,j} = \sum_{j=1}^{n} m_{j,i} = 1$ for every $i \in \llbracket 1, n \rrbracket$. We denote $f : \left|\, \begin{array}{ccc} \mathcal{M}_{n}(\mathbb{R}) & \rightarrow & \mathbb{R} \\ M & \mapsto & \sum_{1 \leqslant i,j \leqslant n} m_{i,j}\left(\lambda_{i}(A) - \lambda_{j}(B)\right)^{2}. \end{array}\right.$ Justify that $f$ admits a minimum on $\mathcal{B}_{n}(\mathbb{R})$.
We denote $\mathcal{B}_{n}(\mathbb{R})$ the set of doubly stochastic matrices in $\mathcal{M}_{n}(\mathbb{R})$, that is the set of matrices $M = \left(m_{i,j}\right)_{1 \leqslant i,j \leqslant n}$ whose coefficients are all non-negative and such that $\sum_{j=1}^{n} m_{i,j} = \sum_{j=1}^{n} m_{j,i} = 1$ for every $i \in \llbracket 1, n \rrbracket$. We denote $f : \left|\, \begin{array}{ccc} \mathcal{M}_{n}(\mathbb{R}) & \rightarrow & \mathbb{R} \\ M & \mapsto & \sum_{1 \leqslant i,j \leqslant n} m_{i,j}\left(\lambda_{i}(A) - \lambda_{j}(B)\right)^{2}. \end{array}\right.$ Let $(i,j,k) \in \llbracket 1,n \rrbracket^{3}$ such that $j \geqslant i$ and $k \geqslant i$. Show that, for $M \in \mathcal{M}_{n}(\mathbb{R})$ and for $x \in \mathbb{R}^{+}$, $$f\left(M + xE_{ii} + xE_{jk} - xE_{ik} - xE_{ji}\right) - f(M) = 2x\left(\lambda_{i}(A) - \lambda_{j}(A)\right)\left(\lambda_{k}(B) - \lambda_{i}(B)\right) \leqslant 0$$
We denote $\mathcal{B}_{n}(\mathbb{R})$ the set of doubly stochastic matrices in $\mathcal{M}_{n}(\mathbb{R})$, that is the set of matrices $M = \left(m_{i,j}\right)_{1 \leqslant i,j \leqslant n}$ whose coefficients are all non-negative and such that $\sum_{j=1}^{n} m_{i,j} = \sum_{j=1}^{n} m_{j,i} = 1$ for every $i \in \llbracket 1, n \rrbracket$. We denote $f : \left|\, \begin{array}{ccc} \mathcal{M}_{n}(\mathbb{R}) & \rightarrow & \mathbb{R} \\ M & \mapsto & \sum_{1 \leqslant i,j \leqslant n} m_{i,j}\left(\lambda_{i}(A) - \lambda_{j}(B)\right)^{2}. \end{array}\right.$ Let $n \geqslant 2$ and $M = \left(m_{i,j}\right)_{1 \leqslant i,j \leqslant n} \in \mathcal{B}_{n}(\mathbb{R})$ a matrix different from the identity. We denote $i$ the smallest integer belonging to $\llbracket 1,n \rrbracket$ such that $m_{i,i} \neq 1$. Show that there exists a matrix $M^{\prime} = \left(m_{i,j}^{\prime}\right)_{1 \leqslant i,j \leqslant n} \in \mathcal{B}_{n}(\mathbb{R})$ such that $f\left(M^{\prime}\right) \leqslant f(M)$ and $m_{j,j}^{\prime} = 1$ for every $j \in \llbracket 1,i \rrbracket$.
We denote $\mathcal{B}_{n}(\mathbb{R})$ the set of doubly stochastic matrices in $\mathcal{M}_{n}(\mathbb{R})$, that is the set of matrices $M = \left(m_{i,j}\right)_{1 \leqslant i,j \leqslant n}$ whose coefficients are all non-negative and such that $\sum_{j=1}^{n} m_{i,j} = \sum_{j=1}^{n} m_{j,i} = 1$ for every $i \in \llbracket 1, n \rrbracket$. We denote $f : \left|\, \begin{array}{ccc} \mathcal{M}_{n}(\mathbb{R}) & \rightarrow & \mathbb{R} \\ M & \mapsto & \sum_{1 \leqslant i,j \leqslant n} m_{i,j}\left(\lambda_{i}(A) - \lambda_{j}(B)\right)^{2}. \end{array}\right.$ Deduce that $$\min\left\{f(M) \mid M \in \mathcal{B}_{n}(\mathbb{R})\right\} = f\left(I_{n}\right)$$