Isometric factorization of vector measures and applications to spaces of integrable functions (Q2062595)

The factorization of vector measures and, in general, of integration operators associated to vector measures, has proved to be a fundamental tool in the theory of operators on Banach function spaces and, therefore, in general, on Banach lattices. The present paper provides a comprehensive overview of this fact, which constitutes an interesting work from at least two points of view: (1) the work makes masterful use of classical results from the general theory of Banach spaces and, in addition, (2) it uses more recent arguments specific to the theory of Banach function spaces, intimately related to integration maps of vector measures. The general context of the paper is provided by the Davis-Figiel-Johnson-Pełczyński factorization of weakly compact operators through reflexive spaces in its isometric version (due to Lima, Nygaard and Oja [\textit{Å.~Lima} et al., Isr. J. Math. 119, 325--348 (2000; Zbl 0983.46024)]), which is exploited in the paper. The authors thus define a new framework for the study of vector measures. Theorem~3.7 improves an already known result, which states that under the assumption of an injective factorization \(I_m= J \circ T\) of an integration operator \(I_m\) of a vector measure, the canonically defined vector measure \(A \mapsto T(\chi_A)\) gives as a space of integrable functions exactly the same as the measure \(m\), with equivalent norms. In the previous version of this result that was known, an additional (unnecessary) assumption on the non-inclusion of a copy of \(\ell_\infty\) in the factorization space was required. In addition, estimates are given for the norm of the factorization operator. This result can be used in an isometric version of the Davis-Figiel-Johnson-Pełczyński procedure (DFJP-LNO for short) that is explained in Section~2.2. This is given in Theorem~4.1, in which the DFJP-LNO factorization of an integration operator of the associated space \(L_\infty(m)\) is given through a reflexive space, with precise estimates of the norms of the pointwise evaluation of the operators involved. Further interesting consequences are shown, for example in Proposition~4.3, where a factorization via a Lorentz space \(L_{2,1}(\|m\|)\) of the semivariation of the vector measure \(m\) is provided. Theorem 4.5 provides the central result of the paper. In it, the authors show that, starting from a DFJP-LNO factorization \(I_m=J \circ T\) of an integration operator \(I_m\), one obtains a standard isometric factorization such that a vector measure \(\hat m: A \mapsto T(\chi_A)\) is shown to satisfy the isometric identification given by \(\|f\|_{L^1(m)}=\|f\|_{L^1(\hat m)}\) for each \(f \in L^1(m).\) Further consequences are shown, including some interesting facts regarding Bochner representability in the case where \(m\) has finite variation. Corollaries~4.7 (and~4.8) complete the picture: Corollary 4.7. Suppose that \(I_m\) is weakly compact. Then there exist a reflexive Banach space \(Y\), a countably additive vector measure \(\hat m : \Sigma \to Y\) and an injective operator \(J : Y \to X\) such that \(m = J \circ \hat m\) and \(L_1(m) = L_1(\hat m)\) with equal norms.