Second dual projection characterizations of three classes of \(L_0\)-closed, convex, bounded sets in \(L_1\) (Q2481819): Difference between revisions

The bidual of \(L^1 = L^1(\lambda)\) (\(\lambda\) a finite positive measure) has the decomposition \((L^1)^{**} = L^1 \oplus_1 L^1_s\); in other words, there is a projection \(P:(L^1)^{**} \to L^1\) such that \(\| x \| = \| Px \| + \| x - Px \| \) for every \(x \in (L^1)^{**}\). \(P\) is called the Yosida--Hewitt projection. In [Trans.\ Mosc.\ Math.\ Soc.\ 34, 127--148 (1978; Zbl 0425.46019)], \textit{A.\,V.\thinspace Bukhvalov} and \textit{G.\,Lozanovski} proved that, if \(C\) is a convex subset of \(L^1 (\lambda)\), bounded and closed in norm, then \(C\) is closed in measure if and only if \(P(\overline{C}^{w^*})=C\). \textit{G.\,Godefroy} [Trans.\ Am.\ Math.\ Soc.\ 286, 227--249 (1984; Zbl 0521.46012); Isr.\ J.\ Math.\ 61, No.\,3, 301--331 (1988; Zbl 0661.43003)] used this result to study subspaces of \(L^1\) whose unit ball is closed in measure. Further results were obtained by him and the reviewer in [Math.\ Scand.\ 66, No.\,2, 249--263 (1990; Zbl 0687.46010)] and led them, with \textit{N.\,J.\thinspace Kalton}, to a characterization of subspaces \(X\) of \(L^1\) which embed into \(\ell_1\) [J.~Reine Angew.\ Math.\ 471, 43--75 (1996; Zbl 0842.46008)], involving compactness in measure and local convexity of the topology of convergence in measure on the unit ball \(B_X\) of \(X\). In the paper under review, the authors generalize some of the results of this last paper to general closed bounded convex subsets \(C\) of \(L^1\), instead of \(B_X\) (and add new results). They prove that, if \(C\) is endowed with the topology of convergence in measure \(\tau_\lambda\), \(W=\overline{C}^{w^*}\) with the weak* topology of \((L^1)^{**}\), then, for the restriction of the Yosida--Hewitt projection \(P_{\mid W} :W \to C\), one has:{\parindent=6mm \begin{itemize}\item[(1)] \(P_{\mid W}\) is weak* to \(\tau_\lambda\) sequentially continuous if and only if \(C\) is a Schur set, i.e., every sequence in \(C\) which is weakly convergent in \(C\), is actually norm convergent; \item[(2)] \(P_{\mid W}\) is a weak* to \(\tau_\lambda\) compactness preserving mapping if and only if \(C\) is \(\tau_\lambda\)-compact; \item[(3)] \(P_{\mid W}\) is a weak* to \(\tau_\lambda\) continuous if and only if \(C\) is a good set (and if and only if \(C\) is a pluriweak-to-measure continuity set).\end{itemize}} \(S \subset L^1\) is said to be a good set if, whenever \(\{f_a\}_a\) is a net in \(S\) such that \(f_a \to_a f + \nu\) weak*, with \(f \in L^1\) and \(\nu \in L^1_s\), then \(f_a \to_a f\) in measure. Interesting examples illustrate the paper.