Another note on copies of \(\ell_\infty\) and \(c_0\) in \(ca(\Sigma,X)\), and the equality \(ca(\Sigma,X)=cca(\Sigma,X)\) (Q1280014)

This note is a complement to the author's paper [Proc. Am. Math. Soc. 109, No. 3, 747-752 (1990; Zbl 0724.46041)]. We introduce the terminology and notation from the introduction of the paper: ``In particular, if \(X\) and \(Y\) are Banach spaces, then \(L(X,Y)\) denotes the space of all bounded linear operators from \(X\) to \(Y\), while \(K(X,Y)\) denotes its closed subspace consisting of compact operators. We write \(X\cong Y\) or \(X\approx Y\) when the spaces \(X\) and \(Y\) are linearly isometric or isomorphic, respectively. The notation like \(X\supset\ell_\infty\) is used to mean that \(X\) contains an isomorphic copy of \(\ell_\infty\). Throughout, we let \(\Sigma\) stand for a \(\sigma\)-algebra on a nonempty set \(S\), and \(X\) for a Banach space. By \(ca(\Sigma, X)\) we denote the Banach space of all countably additive vector measures \(\mu:\Sigma\to X\) with the sup-norm: \(\| \mu\|= \sup_{E\in\Sigma}\|\mu(E)\|\). (The equivalent semivariation norm could be used as well.) The closed subspace of \(ca(\Sigma, X)\), consisting of those measures \(\mu\) whose range, \(\mu(\Sigma)\), is relatively norm compact, will be denoted by \(cca(\Sigma, X)\). If \(m\) is a finite positive measure on \(\Sigma\), then we write \[ ca(\Sigma, m;X)= \{\mu\in ca(\Sigma,X):\mu\ll m\}, \] \[ cca(\Sigma, m;X)= ca(\Sigma,m; X)\cap cca(\Sigma,X); \] these also are closed subspaces of \(ca(\Sigma,X)\). In the case when \(X\) is the space of (real or complex) scalars, we will suppress \(X\) in these notations. Finally, we will say that the \(\sigma\)-algebra \(\Sigma\) is of type (NA) if it admits a nonzero, atomless, finite positive measure, and that it is of type (A) otherwise.'' Typical theorems in this note are: Theorem. Suppose the \(\sigma\)-algebra \(\Sigma\) is of type (NA), and let \(m\) be any nonzero atomless finite positive measure on \(\Sigma\). Then the following are equivalent. (a) \(ca(\Sigma,X)\supset \ell_\infty\). (b) \(ca(\Sigma,X)\supset c_0\). \((\overline{\text{b}})\) \(cca(\Sigma, X)\supset c_0\). (c) \(K(\ell_2,X)\neq L(\ell_2,X)\). \((\overline{\text{c}})\) There exists a sequence \((x_n)\) in \(X\) which does not converge to zero and is such that \[ \sum^\infty_{n=1}| x^*(x_n)|^2<\infty\quad\text{for all }x^*\in X^*. \] (d) There exists an isomorphic embedding \(J:\ell_\infty\to ca(\Sigma,m;X)\subset ca(\Sigma,X)\) such that \(J(c_0)= J(\ell_\infty)\cap cca(\Sigma,m; X)\) and \(J(c_0)\) is complemented in \(cca(\Sigma,m; X)\), hence also in \(cca(\Sigma, X)\). (e) There exist a sequence \((\mu_n)\) in \(cca(\Sigma,X)\) and a measure \(\mu\) in \(ca(\Sigma,X)\setminus cca(\Sigma,X)\) such that \(\mu(E)= \sum^\infty_{n=1}\mu_n(E)\) for every \(E\in\Sigma\), and \((*)\) the series \(\sum^\infty_{n=1} \mu_n(E)\) converges unconditonally in \(X \forall E\in\Sigma\). (f) There exists a sequence \((\mu_n)\) in \(ca(\Sigma,X)\) such that \((*)\) holds but the series \(\sum^\infty_{n=1}\mu_n\) does not converge in \(ca(\Sigma, X)\). (g) There exists a sequence \((\mu_n)\) in \(ca(\Sigma, X)\) such that \((*)\) holds and \(\inf_n\|\mu_n\|> 0\). Theorem. For any Banach space \(X\), the following are equivalent. (1) \(ca(\Sigma,m; X)= cca(\Sigma, m;X)\), or \(L_{w^*}(L_\infty(m),X)= K_{w^*}(L_\infty(m), X)\), for some \(\sigma\)-algebra \(\Sigma\) of type (NA) and a nonzero atomless finite positive measure \(m\) on \(\Sigma\). (2) \(ca({\mathcal B},\lambda; X)= cca({\mathcal B},\lambda; X)\), or \(L_{w^*}(L_\infty(\lambda), X)= K_{w^*}(L_\infty(\lambda), X)\), where \({\mathcal B}\) is the Borel \(\sigma\)-algebra on \([0,1]\) and \(\lambda\) is Lebesgue measure. (3) \(ca(\Sigma, X)= cca(\Sigma, X)\) for every \(\sigma\)-algebra \(\Sigma\). (4) \(ba({\mathcal F},X)= cba({\mathcal F},X)\), or \(L(B({\mathcal F}), X)= K(B({\mathcal F}), X)\), for every algebra of sets \({\mathcal F}\). (5) \(L(C(\Omega),X)= K(C(\Omega), X)\) for every compact space \(\Omega\). (6) \(L(C(I), X)= K(C(I), X)\), where \(I\) is the closed unit interval. (7) \(L(L_\infty(m), X)= K(L_\infty(m), X)\) for every positive measure \(m\). (8) \(L(L_\infty(m), X)= K(L_\infty(m), X)\) for some probability atomless measure \(m\). (9) \(L(\ell_\infty,X)= K(\ell_\infty,X)\).