Paperlane

no. 16 · Tuesday 17 November 2026

On Pairwise Independent and Strongly Correlated Random Variables

Dmitrii Gvozdev and Alexander Kolesnikov

math.PR

Abstract

Characterizing joint distributions that maximize the centered third-order moment under prescribed marginal distributions and pairwise independence has various statistical applications. This problem can be formulated as a multistochastic optimal transport problem. In this work, we generalize the results on the $(3,2)$-multistochastic Monge-Kantorovich problem to cost functions of the form $c(x,y,z)=f(x)f(y)f(z)$, where $f$ is defined on $[0,1]$ through dyadic expansions: $f(x)=\sum_{i\geq 1} a_i x_i$, with $x_i\in\{0,1\}$. Under summability and domination conditions on the coefficients, we show that an optimal transport plan is concentrated on the Sierpinski tetrahedron. We also characterize mixtures of geometric coefficient sequences through Hausdorff moment sequences and construct a recursive dual solution with its integral representation for costs with geometric coefficients.

fields
Probability
pdf ↗ page arXiv:2611.20365 ↗