Limit Theorems in Warsserstein Distance for Empirical Measures of Diffusion Processes on Riemannian Manifolds

Abstract: Let

M

be a compact connected Riemannian manifold possibly with a boundary, let

V i n C^{2} (M)

such that

m u (d x) : = e^{V (x)} d x

is a probability measure, and let

{l a m b d a_{i}}_{i g e 1}

be all non-trivial eigenvalues of

- L

with Neumann boundary condition if the boundary exists. Then the empirical measures

{m u_{t}}_{t > 0}

of the diffusion process generated by

L

(with reflecting boundary if the boundary exists) satisfy

lim_{t o infty} �ig{t mathbb E^x [W_2(mu_{t},mu)^2]�ig}= sum_{i=1}^inftyfrac 2 {lambda_i^2} ext{ uniformly in } xin M,

where

m a t h b b E^{x}

denotes the expectation for the diffusion process starting at point

x

,

W_{2}

is the

L^{2}

-Warsserstein distance induced by the Riemannian metric. The limit is finite if and only if

d l e 3

, and in this case we derive the following central limit theorem:

lim_{t oinfty} sup_{xin M} Big|mathbb P^x(t W_2(mu_{t},mu)^2<a)- mathbb PBig( sum_{k=1}^infty frac{2xi_k^2}{lambda_k^2}<aBig)Big|=0, age 0,

where

m a t h b b P^{x}

is the probability with respect to

m a t h b b E^{x}

, and

{x i_{k}}_{k g e 1}

are i.i.d. standard Gaussian random variables. Moreover, when

d g e 4

we prove that the main order of

m a t h b b E^{x} [W_{2} (m u_{t}, m u)^{2}]

is

t^{- f r a c 2 d - 2}

as

t o i n f t y

. Moreover, when

d g e 4

the main order of

m a t h b b E^{x} [W_{2} (m u_{t}, m u)^{2}]

is

t^{- f r a c 2 d - 2}

as

t o i n f t y

. Finally, we establish the long-time large deviation principle for

{W_{2} (m u_{t}, m u)^{2}}_{t g e 0}

with a good rate function given by the information with respect to

m u

.

This page was built for publication: Limit Theorems in Warsserstein Distance for Empirical Measures of Diffusion Processes on Riemannian Manifolds