Showing 1–1 of 1 results for author: Massa, M

How to quantise probabilities while preserving their convex order

Authors: Marco Massa, Pietro Siorpaes

Abstract: We introduce an algorithm which, given probabilities $μ\leq_{\text{cx}} ν$ in convex order and defined on a separable Banach space $B$, constructs finitely-supported approximations $μ_n \to μ, ν_n\to ν$ which are in convex order $μ_n \leq_{\text{cx}} ν_n$. We provide upper-bounds for the speed of convergence, in terms of the Wasserstein distance. We discuss the (dis)advantages of our algorithm and… ▽ More We introduce an algorithm which, given probabilities $μ\leq_{\text{cx}} ν$ in convex order and defined on a separable Banach space $B$, constructs finitely-supported approximations $μ_n \to μ, ν_n\to ν$ which are in convex order $μ_n \leq_{\text{cx}} ν_n$. We provide upper-bounds for the speed of convergence, in terms of the Wasserstein distance. We discuss the (dis)advantages of our algorithm and its link with the discretisation of the Martingale Optimal Transport problem, and we illustrate its implementation with numerical examples. We study the operation which, given $μ$/$ν$ and some (finite) partition of $B$, outputs $μ_n$/$ν_n$, showing that applied to a probability $γ$ and to all partitions it outputs the set of all probabilities $ζ\leq_{\text{cx}} γ$. △ Less

Submitted 21 June, 2022; originally announced June 2022.