Unbiased Estimation using a Class of Diffusion Processes

We study the problem of unbiased estimation of expectations with respect to (w.r.t.) a given, general probability measure on that is absolutely continuous with respect to a standard Gaussian measure. We focus on simulation associated to a particular class of diffusion processes, sometimes termed the Schr\"odinger-F\"ollmer Sampler, which is a simulation technique that approximates the law of a particular diffusion bridge process on , . This latter process is constructed such that, starting at , one has . Typically, the drift of the diffusion is intractable and, even if it were not, exact sampling of the associated diffusion is not possible. As a result, \cite{sf_orig,jiao} consider a stochastic Euler-Maruyama scheme that allows the development of biased estimators for expectations w.r.t.~. We show that for this methodology to achieve a mean square error of , for arbitrary , the associated cost is . We then introduce an alternative approach that provides unbiased estimates of expectations w.r.t.~, that is, it does not suffer from the time discretization bias or the bias related with the approximation of the drift function. We prove that to achieve a mean square error of , the associated cost is, with high probability, , for any . We implement our method on several examples including Bayesian inverse problems.
View on arXiv