Statistical Independence and the Brockwell Transform -- From an Integral Equation Perspective (2404.07558v1)

Abstract: Statistical independence is a notion ubiquitous in various fields such as in statistics, probability, number theory and physics. We establish the stability of independence for any pair of random variables by their corresponding Brockwell transforms (Brockwell, 2007) beyond the non-atomic condition that is naturally imposed on their distributions, thereby generalizing the proposition originated by Cai et al. (2022). A central novelty in our work is to formulate the problem as a possibly new type of mathematical inverse problem, which aims to claim that an integral equation has a unique solution in terms of its stochastic kernel -- not under-determined contrary to the usual cases, and also to design a recursive constructive scheme, combined with the property of the quantile function, that solves the aforementioned integral equation in an iterative manner.