For a given probability density function p(x) on R^d, we construct a (non-stationary) diffusion process xt, starting at any point x in R^d, such that 1/T ∫_o^T δ(xt-x)dt converges to p(x) almost surely. The ...For a given probability density function p(x) on R^d, we construct a (non-stationary) diffusion process xt, starting at any point x in R^d, such that 1/T ∫_o^T δ(xt-x)dt converges to p(x) almost surely. The rate of this convergence is also investigated. To find this rate, we mainly use the Clark-Ocone formula from Malliavin calculus and the Girsanov transformation technique.展开更多
基金supported by the Simons Foundation (Grant No. 209206)a General Research Fund of the University of Kansas
文摘For a given probability density function p(x) on R^d, we construct a (non-stationary) diffusion process xt, starting at any point x in R^d, such that 1/T ∫_o^T δ(xt-x)dt converges to p(x) almost surely. The rate of this convergence is also investigated. To find this rate, we mainly use the Clark-Ocone formula from Malliavin calculus and the Girsanov transformation technique.