Adapted Wasserstein Barycenters of Gaussian Processes

TL;DR

This paper studies barycenters of Gaussian processes under adapted Wasserstein distance, providing theoretical characterizations and properties relevant for stochastic systems with filtration constraints.

Contribution

It introduces Gaussian solutions for adapted Wasserstein barycenter problems, characterizes these barycenters, and explores their existence, uniqueness, and regularity.

Findings

Gaussian solutions exist for the barycenter problem.

Characterization of barycenters via means and covariance operators.

Analysis of existence, uniqueness, and regularity properties.

Abstract

We investigate barycenters of Gaussian process laws in adapted Wasserstein space. The adapted Wasserstein distance refines classical optimal transport by enforcing compatibility of transport plans with the temporal flow of information, and is therefore well suited for stochastic systems with filtration constraints, as common in stochastic control, mathematical finance and sequential decision problems. Within this framework, we consider weighted Fr\'echet means of Gaussian process laws and prove that the associated barycenter problem admits Gaussian solutions. We derive a characterization of adapted Wasserstein barycenters in terms of the means and covariance operators of the underlying processes, and we analyze their existence, uniqueness, and regularity properties under natural assumptions. The Gaussian setting reveals a tractable and structurally rich subclass of adapted transport problems, bridging adapted optimal transport and Bures--Wasserstein geometry. Our results identify adapted Wasserstein barycenters as natural representatives of collections of Gaussian models and suggest new applications in stochastic optimization, robust finance, and sequential statistics.