Intrinsic coupling on Riemannian manifolds and polyhedra (Q1767541)

This paper aims to simplify Cranston's stochastic proof of gradient estimates for harmonic functions on a Riemannian manifold. The paper starts from a central limit theorem (CLT) for geometric random walks and gives an elementary construction of couplings between Brownian motions on a Riemannian manifold. It also shows that cut locus phenomena are inessential for Kendall's and Cranston's stochastic proof of gradient estimates for harmonic functions on a Riemannian manifold with lower curvature bound. It is mentioned that -- since the method only uses an asymptotic quadruple inequality and a CLT -- this approach might work on certain non-smooth spaces, and this is illustrated for Riemannian polyhedra. The author also recovers the classical heat kernel gradient estimate which is well known in the smooth setting.