\documentclass{article} \usepackage[margin=1in]{geometry} \usepackage{amsmath} \usepackage{amsthm} \usepackage{parskip} \usepackage{amssymb} \usepackage{fancyhdr} \usepackage{enumerate} \usepackage[bookmarksnumbered]{hyperref} \usepackage[english]{babel} \usepackage{dashrule} \newcommand{\R}{\mathbb{R}} \newcommand{\Z}{\mathbb{Z}} \newcommand{\N}{\mathbb{N}} \newcommand{\Q}{\mathbb{Q}} \newcommand{\C}{\mathbb{C}} \renewcommand{\P}{\mathbb{P}} \newcommand{\F}{\mathbb{F}} \newcommand{\half}{\frac{1}{2}} \usepackage{ textcomp } \newcommand{\A}{\mathcal{A}} \renewcommand{\H}{\mathcal{H}} \newcommand{\1}{\mathbb{1}} \newcommand{\teq}[1]{\stackrel{#1}{=}} \newtheorem{lemma}{Lemma} \newtheorem{theorem}[lemma]{Theorem} \newtheorem{definition}[lemma]{Definition} \pagestyle{fancy} \fancyhead[L]{Hand-in 8 Functional Analysis} \fancyhead[R]{Thomas van Maaren (9825827)} \begin{document} \textbf{Exercise 2.} For each $N \in \mathbb{N}$ we define the functional $M_{N}: \ell^{\infty}(\mathbb{Z}) \rightarrow \mathbb{C}$ by $$ M_{N}(x)=\frac{1}{2 N+1} \sum_{|n| \leq N} x_{n}, \quad \forall x \in \ell^{\infty}(\mathbb{Z}) $$ For each $a \in \mathbb{Z}$ we define the operator $T_{a}: \ell^{\infty}(\mathbb{Z}) \rightarrow \ell^{\infty}(\mathbb{Z})$ by $$ \left(T_{a} x\right)_{n}=x_{n+a}, \quad \forall x \in \ell^{\infty}(\mathbb{Z}) . $$ Let also $e \in \ell^{\infty}(\mathbb{Z})$ be defined by $e_{n}=1$ for all $n \in \mathbb{Z}$, and let $V \subset \ell^{\infty}(\mathbb{Z})$ be the subspace of all $x$ that can be written as a finite sum: $$ x=\lambda_{0} e+\sum_{j=1}^{p} \lambda_{j}\left(f_{j}-T_{a_{j}} f_{j}\right) $$ for some ( $x$-dependent) $p \in \mathbb{N}, \lambda_{0}, \ldots, \lambda_{p} \in \mathbb{C}, a_{1}, \ldots, a_{p} \in \mathbb{Z}$, and $f_{1}, \ldots, f_{p} \in \ell^{\infty}(\mathbb{Z})$. (i) [1 p.] Show that for all $x \in \ell^{\infty}(\mathbb{Z})$ and $a \in \mathbb{Z}$, $$ \left|M_{N}\left(x-T_{a} x\right)\right| \leq \frac{2|a|}{2 N+1}\|x\|_{\infty} . $$ (ii) [0,5 p.] For each $x \in V$, show that $\lim _{N \rightarrow \infty} M_{N}(x)$ equals the coefficient $\lambda_{0}$ in any decomposition (1). In consequence, we can define a map $L: V \rightarrow \mathbb{C}$ by $$ L(x)=\lambda_{0} . $$ Show that $L \in V^{\prime}$. (iii) $[0,5$ p. $]$ Show that the functional $L$ has norm $\|L\|=L(e)=1$. (iv) $[1,5$ p. $]$ Show that there exists a functional $M \in\left(\ell^{\infty}(\mathbb{Z})\right)^{\prime}$ such that: (a) $M(x)=L(x)$ for all $x \in V$, (b) $M(e)=\|M\|=1$, (c) $M(x)=M\left(T_{a} x\right)$ for all $a \in \mathbb{Z}$ and $x \in \ell^{\infty}(\mathbb{Z})$. (v) $[0,5$ p. $]$ Show that there exists no $y \in \ell^{1}(\mathbb{Z})$ such that $$ M(x)=\sum_{n \in \mathbb{N}} x_{n} y_{n}, \quad \forall\left(x_{n}\right)_{n \in \mathbb{Z}} \in \ell^{\infty}(\mathbb{Z}) $$ (vi) $[0,5$ p. $]$ Using the result from the previous question show that $\ell^{1}(\mathbb{Z})$ is not reflexive. In the next exercises, all vector spaces are over $\mathbb{C}$. \hdashrule{\textwidth}{.4pt}{1pt} \begin{enumerate}[(i)] \item Observe that if $a\geq 0$ \begin{align} \left|M_N(x-T_ax)\right| &= \frac{1}{2N+1}\left|\sum_{|n|\leq N} (x-T_ax)_n\right|= \frac{1}{2N+1}\left|\sum_{n=-N}^N x_n - \sum_{n=-N+a}^{N+a}x_n \right|\\ &= \frac{1}{2N+1} \left| \sum_{n=-N}^{-N+a-1} x_n+ \sum_{n=-N+a}^N x_n -\sum_{n=-N+a}^{N} x_n - \sum_{n=N+1}^{N+a} x_n\right|\\ &= \frac{1}{2N+1} \left| \sum_{n=-N}^{-N+a-1}x_n-\sum_{n=N+1}^{N+a} x_n\right|\leq \frac{1}{2N+1}\left( \sum_{n=-N}^{-N+a-1} |x_n| + \sum_{n=N+1}^{N+a} |x_n|\right)\\ &\leq \frac{1}{2N+1}\left( \sum_{n=-N}^{-N+a-1} \|x\|_\infty + \sum_{n=N+1}^{N+a} \|x\|_\infty\right)=\frac{1}{2N+1}\left( a \|x\|_\infty + a \|x\|_\infty\right)\\ &= \frac{2a\|x\|_\infty}{2N+1} = \frac{2|a|\|x\|_\infty}{2N+1} \end{align} Now we need to prove it is true if $a<0$. It is easy to prove that $T_a$ is an isometry. For any $a\in \Z$ and $x\in \ell^\infty$ we see that $\|T_a x\|_\infty = \sup_{n\in \Z} |x_{n+a}| = \sup_{n\in \Z} |x_n| = \|x\|_\infty$. If $a<0$ we know that $-a>0$, so \begin{align}\left|M_n(x-T_ax)\right|&=\left|-M_N(T_a x-x)\right| =|M_n(T_ax - T_{-a}T_ax)|\\ &\leq \frac{2(-a) \|T_ax\|_\infty}{2N+1} =\frac{2|a| \|x\|_\infty}{2N+1} \end{align} in the last equality we used that $T_a$ is an isometry. \item We will first determine $\lim_{N\to \infty} M_n(e)$ \begin{align} \lim_{N\to \infty} M_n(e) = \lim_{N\to \infty} \frac{1}{2N+1} \sum_{|n|\leq N} 1 = \lim_{N\to \infty} \frac{2N+1}{2N+1} = 1 \end{align} Because of (i) we know that for any $1\leq j\leq p$ \begin{align} \lim_{N\to \infty} |M_N(f_j - T_{a_j}f_j| \leq \lim_{N\to \infty}\frac{2|a|}{2N+1}\|x\|_\infty = 0 \end{align} hence \begin{align} \lambda_0 &= \lambda_0\cdot 1 + \sum_{j=1}^p \lambda_j \cdot 0= \lambda_0\cdot \lim_{N\to \infty} M_N(e) + \sum_{j=1}^p \lambda_j \cdot \lim_{N\to \infty} M_N(f_j-T_{a_j}f_j)\\ &= \lim_{N\to \infty} M_N\left(\lambda_0 e+\sum_{j=1}^p\lambda_j(f_j-T_{a_j}f_j)\right) = \lim_{N\to \infty} M_N(x) \end{align} To show that $L\in V'$ we need to show that $L$ is linear and bounded. In (iii) we will prove that $\|L\|=1$, hence $L$ is bounded and therefore we only need to show that $L$ is linear. For this we need to prove that $L(x+\kappa x') = L(x) + \kappa L(x')$ for all $x,x'\in V$ and $\kappa\in \F$. We can write $x$ and $x'$ as $\lambda_{0} e+\sum_{j=1}^{p} \lambda_{j}\left(f_{j}-T_{a_{j}} f_{j}\right)$ and $\lambda_{0}' e+\sum_{j=1}^{p'} \lambda_{j}'\left(f_{j}'-T_{a_{j}'} f_{j}'\right)$. Now define $\mu_0=\lambda_0 + \kappa \lambda_0'$ and $\mu_i = \lambda_i,b_i=a_i,g_i=f_i$ for $1\leq i\leq p$ and $\mu_i = \kappa \lambda_{i-p}',b_i=a_{i-p}',g_i=f_{i-p}$ for $p