2

Let $X$ be a Banach space. A series $\sum x_n$ in $X$ is weakly unconditionally Cauchy (or weakly uncontionally convergent) if $\sum |x^*(x_n)| < +\infty$ for every $x^* \in X^*$.

Exercise 3, page 114 of Diestel's Sequences and Series in Banach Spaces reads:

The bounded linear operators from a Banach space $X$ into $\ell_1$ correspond precisely to the sequences $(x_n^*)$ in $X^*$ for which $\sum |x^*_n(x)| < +\infty$, for each $x \in X$, i. e., the weakly unconditionally Cauchy series in $X^*$.

I'm having trouble with the bolded part. More precisely, I don't see why $\sum |x^*_n(x)| < +\infty$ for each $x \in X$ implies that $\sum x^*_n$ is weakly unconditionally Cauchy in $X^*$ (that is, $\sum |x^{**} (x^*_n)| < +\infty$ for every $x^{**} \in X^{**}$).

Thanks in advance :)

  • 3
    I may be off here, but: a given $(x_n^)$ defines a bounded operator taking $x$ to $( x_n^x)$. Proving the other direction for the adjoint of this particular operator restricted to $c_0$ (use exercise 2) shows $(x_n^*)$ is in fact WuC. – David Mitra Nov 25 '20 at 18:26
  • That settles it! Thanks a lot. – Vinícius Morelli Nov 25 '20 at 21:04

1 Answers1

0

Recall that a series $(\sum_{k=1}^{n} x_{k})_{n\in\mathbb{N}}$ in a Banach space $X$ is called weak unconditionally Cauchy if for every $x^{*}\in X^{*}$ the series $(\sum_{k=1}^{n} x^{*}(x_{k}))_{n\in\mathbb{N}}$ in $\mathbb{K}$ is unconditionally Cauchy. Recall that a series in $\mathbb{K}$ is unconditionally Cauchy if and only if it is absolutely convergent.

Claim. Let $X$ be a Banach space and $(x_{n}^{*})_{n\in\mathbb{N}}$ a sequence in $X^{*}$. The following are equivalent.

$({\rm i})$ The series $(\sum_{k=1}^{n} x_{k}^{*} )_{n\in\mathbb{N}}$ in $X^{*}$ is weak unconditionally Cauchy.

$({\rm ii})$ There is a bounded linear operator $T\colon X\to \ell_{1}$ such that $Tx = (x_{n}^{*}(x))_{n\in\mathbb{N}}$ for all $x\in X$.

Proof. $({\rm i}) \Rightarrow ({\rm ii})$. Suppose the series $(\sum_{k=1}^{n} x_{k}^{*} )_{n\in\mathbb{N}}$ in $X^{*}$ is weak unconditionally Cauchy. Let $x\in X$. Since the map $f\mapsto f(x)$ defined on $X^{*}$ is a continuous linear functional on $X^{*}$, the series $(\sum_{k=1}^{n} x_{k}^{*}(x))_{n\in\mathbb{N}}$ is absolutely convergent. Hence $(x_{n}^{*}(x))_{n\in\mathbb{N}}$ belongs to $\ell_{1}$. Define $T\colon X\to \ell_{1}$ by \begin{equation} Tx := (x_{n}^{*}(x))_{n\in\mathbb{N}} . \end{equation} Observe that $T$ is a well-defined linear operator. A standard argument shows that $T$ has a closed graph. We include the argument for completeness. Let $(x_{k})_{k\in\mathbb{N}}$ be a sequence in $X$, $x\in X$ and $(\alpha_{n})_{n\in\mathbb{N}} \in \ell_{1}$ such that $(x_{k})_{k\in\mathbb{N}}$ converges to $x$ in $X$ and $(Tx_{k})_{k\in\mathbb{N}}$ converges to $(\alpha_{n})_{n\in\mathbb{N}}$ in $\ell_{1}$. Let $n\in\mathbb{N}$. Since the sequence $(x_{k})_{k\in\mathbb{N}}$ converges to $x$ in $X$ and $x_{n}^{*}\in X^{*}$ we have \begin{equation} \lim_{k\to\infty} x_{n}^{*}(x_{k}) = x_{n}^{*}(x) . \tag{1} \end{equation} Now define $e_{n}^{*}\colon \ell_{1}\to\mathbb{K}$ by $e_{n}^{*}((\beta_{k})_{k\in\mathbb{N}} ) := \beta_{n}$. Then $e_{n}^{*}\in \ell_{1}^{*}$. As $(Tx_{k})_{k\in\mathbb{N}}$ converges to $(\alpha_{j})_{j\in\mathbb{N}}$ in $\ell_{1}$ and $e_{n}^{*}\in \ell_{1}^{*}$, we have \begin{equation} \lim_{k\to\infty} x_{n}^{*}(x_{k}) = \lim_{k\to\infty} e_{n}^{*}(Tx_{k}) = e_{n}^{*}((\alpha_{j})_{j\in\mathbb{N}} ) = \alpha_{n} . \tag{2} \end{equation} By combining $(1)$ and $(2)$ we obtain \begin{equation} x_{n}^{*}(x) = \alpha_{n} . \tag{3} \end{equation} As $n$ was arbitrary we see from $(3)$ that $Tx = (\alpha_{n})_{n\in\mathbb{N}}$. Hence $T$ has a closed graph. By the closed graph theorem we see that $T$ is bounded. Hence $T$ has the desired properties.

$({\rm ii}) \Rightarrow ({\rm i})$. Assume there is a bounded linear operator $T\colon X\to \ell_{1}$ such that $Tx = (x_{n}^{*}(x))_{n\in\mathbb{N}}$ for all $x\in X$. For each $n\in\mathbb{N}$ define $e_{n}^{*}\colon \ell_{1}\to\mathbb{K}$ by $e_{n}^{*}((\alpha_{j})_{j\in\mathbb{N}}) := \alpha_{n}$. Let $\Phi \colon \ell_{\infty} \to \ell_{1}^{*}$ be the usual isomorphism given by \begin{equation} (\Phi (\beta_{j})_{j\in\mathbb{N}}))(\alpha_{i})_{i\in\mathbb{N}} = \sum_{k\in\mathbb{N}} \alpha_{k} \beta_{k} . \end{equation} Define $S := (T^{*}\circ \Phi)\vert_{c_{0}}$. Observe that $S$ is a bounded linear operator from $c_{0}$ into $X^{*}$. Let $n\in\mathbb{N}$ and $x\in X$. We have \begin{equation} (Se_{n})(x) = (T^{*}(\Phi e_{n}))(x) = (\Phi e_{n})(Tx) = e_{n}^{*}(Tx) = x_{n}^{*}(x) . \end{equation} Hence we have $Se_{n} = x_{n}^{*}$ for each $n\in\mathbb{N}$. Let $x^{**}\in X^{**}$. Then $S^{*}x^{**}\in c_{0}^{*}$. Since the series $(\sum_{k=1}^{n} e_{k})_{N\in\mathbb{N}}$ is weak unconditionally Cauchy in $c_{0}$, the series $(\sum_{k=1}^{n} (S^{*} x^{**})e_{k})_{N\in\mathbb{N}}$ is absolutely convergent. Hence for each $N\in\mathbb{N}$ we have \begin{equation} \sum_{k=1}^{N} |x^{**}(x_{k}^{*})| = \sum_{k=1}^{N} |x^{**}(Se_{k})| = \sum_{k=1}^{N} |(S^{*}x^{**})e_{k}| \leq \sum_{k=1}^{\infty} |(S^{*}x^{**})e_{k}| < \infty . \tag{4} \end{equation} It follows from $(4)$ that the series $(\sum_{k=1}^{n} x^{**}(x_{k}^{*}))_{n\in\mathbb{N}}$ is unconditionally Cauchy. Hence $(\sum_{k=1}^{n} x_{k}^{*})_{n\in\mathbb{N}}$ is weak unconditionally Cauchy.

Dean Miller
  • 9,566