The zonoid algebra, generalized mixed volumes, and random determinants (Q2136565)

From MaRDI portal
scientific article
Language Label Description Also known as
English
The zonoid algebra, generalized mixed volumes, and random determinants
scientific article

    Statements

    The zonoid algebra, generalized mixed volumes, and random determinants (English)
    0 references
    0 references
    0 references
    0 references
    0 references
    11 May 2022
    0 references
    As the authors say, this paper links the three topics of zonoids, mixed volumes, and random determinants under a general new framework. The starting point is a notion of tensor product of two (origin-symmetric) zonoids, originally due to [\textit{G. Aubrun} and \textit{C. Lancien}, Positivity 20, No. 1, 1--23 (2016; Zbl 1335.52015)]. After introducing formal differences of zonoids to obtain vector spaces of zonoids, and associating with a multilinear map between Euclidean spaces a canonical multilinear map of the corresponding vector spaces of zonoids, the authors are able to introduce an exterior product of zonoids. On the way, continuity investigations play an important role. Given \(k\le m\) zonoids \(Z_1,\dots,Z_k\) in an \(m\)-dimensional Euclidean vector space \(V\), the exterior product \(Z_1\wedge\dots\wedge Z_k\) is defined as a zonoid in the exterior power \(\Lambda^k V\). This leads to a graded, associative and commutative real algebra, which is called the zonoid algebra. This algebra inherits a duality notion from the Hodge star operator. Remarkably, the Hodge dual of \(Z^{\wedge(m-1)}\) is a multiple of the projection body of \(Z\). Some surprising connections to mixed volumes are revealed. For example, the mixed volume of zonoids \(Z_1,\dots,Z_m\subset V\) is, up to an explicit factor, the mean width of \(Z_1\wedge\dots\wedge Z_m\), and the \(d\)th intrinsic volume of the zonoid \(Z\subset V\) is, up to an explicit factor, the mean width of \(Z^{\wedge d}\). Applications are made to expectations of absolute random determinants \(|\det M|\). A result of [\textit{R. A. Vitale}, Ann. Appl. Probab. 1, No. 2, 293--300 (1991; Zbl 0739.15015)], where the columns of the random matrix \(M\) are i.i.d. random vectors, can be extended to the case where the columns are no longer identically distributed, and even to some cases where the independence is relaxed. Interesting conclusions are drawn from these extensions, by applying the Aleksandrov-Fenchel inequalities for mixed volumes. Similar constructions are based on the complex wedge product. Among other applications, an analogue of mixed volumes comes up, called the mixed J-volume. In particular, this leads to special angular valuations on polytopes in \({\mathbb R}^{2n}\).
    0 references
    0 references
    0 references
    0 references
    0 references
    convex body
    0 references
    zonoid
    0 references
    mixed volume
    0 references
    random determinant
    0 references
    0 references
    0 references