A diffusion approach to Stein's method on Riemannian manifolds

  • 2020-03-25 17:03:58
  • Huiling Le, Alexander Lewis, Karthik Bharath, Christopher Fallaize
  • 4

Abstract

We detail an approach to develop Stein's method for bounding integral metricson probability measures defined on a Riemannian manifold $\mathbf{M}$. Ourapproach exploits the relationship between the generator of a diffusion on$\mathbf{M}$ with target invariant measure and its characterising Steinoperator. We consider a pair of such diffusions with different starting points,and investigate properties of solution to the Stein equation based on analysisof the distance process between the pair. Several examples elucidating the roleof geometry of $\mathbf{M}$ in these developments are presented.

 

Quick Read (beta)

loading the full paper ...