The weak topology on \(L^p\) of a vector measure (Q930751): Difference between revisions

The aim of the paper under review is to investigate the weak topology of \(L^p(\mu,X)\), \(1<p<\infty\), where \(\mu\) is countably additive and \(X\) is a real Banach space. The main result is the following: A bounded net \((f_\alpha)\subset L^p(\mu,X)\) converges weakly to \(f\in L^p(\mu,X)\) if and only if \(\int_A f_\alpha d\mu\to\int_A fd\mu\) weakly in \(X\) for each measurable set \(A\). It is essential that \(p>1\); for \(p=1\) it is known (found independently by G. P. Curbera and S. Okada) that the above characterization needs the extra assumption that \(L^1(\mu,X)\) contains no complemented copy of \(\ell_1\). On their way to the main result the authors observe that \(L^p(\mu,X)^\ast\) is order continuous with weak unit, so \(L^p(\mu,X)^\ast\) is weakly compactly generated, or WCG in short. From this it follows that every subspace of \(L^p(\mu,X)\) is WCG. When \(1/p+1/q=1\), defining \[ \Gamma=\{\gamma_{g,x^\ast}:g\in L^q(\mu,X), x^\ast\in X^\ast\}\subset B_{L^p(\mu,X)^\ast},\;\gamma_{g,x^\ast}(f)=\int_\Omega fgd(x^\ast \mu), \] one can observe that the main result is equivalent to the statement that the weak topology of \(L^q(\mu,X)\) coincides on bounded sets with the topology of pointwise convergence on \(\Gamma\). The last part of the paper addresses the problem of finding when \(\Gamma\) is a James boundary. \(\Gamma\) is norming, but it seems to be unclear whether it is always a James boundary. Two sufficient conditions to assure \(\Gamma\) be a James boundary are found.