On the multilinear Hausdorff problem of moments (Q395147)

The paper considers a weaker version of the classical Hausdorff multivariate moment problem. In particular, a characterization of all multimoment sequences of a regular Borel polymeasure on \([0,1]^n\) is given in terms of the Bernstein coefficients of the \(n\)-linear functional associated to the multisequence (Theorem~2). This result is a nontrivial generalization of the classical solution to the univariate moment problem on \([0,1]\) due to Hausdorff. The authors also discuss how their ideas combined with some properties of polymeasures on compact sets can be used to derive a solution to the classical multivariate problem (end of Section~2). Moreover, recent developments on integral representations of orthogonally additive polynomials on Banach lattices are employed to give necessary and sufficient conditions such that a multisequence which solves the classical Hausdorff multivariate moment problem is also a solution to the so-called strong Hausdorff multilinear moment problem (Theorem~3). This article contains an application to weakly harmonizable stochastic processes (Section~4). The authors propose a characterization of the covariance function of a second-order weakly Hausdorff harmonizable stochastic process, i.e., such that the support of the defining stochastic measure is contained in \([0,1]\).