Schr{\"o}dinger-F{\"o}llmer Sampler: Sampling without Ergodicity
read the original abstract
Sampling from probability distributions is an important problem in statistics and machine learning, specially in Bayesian inference when integration with respect to posterior distribution is intractable and sampling from the posterior is the only viable option for inference. In this paper, we propose Schr\"{o}dinger-F\"{o}llmer sampler (SFS), a novel approach for sampling from possibly unnormalized distributions. The proposed SFS is based on the Schr\"{o}dinger-F\"{o}llmer diffusion process on the unit interval with a time dependent drift term, which transports the degenerate distribution at time zero to the target distribution at time one. Comparing with the existing Markov chain Monte Carlo samplers that require ergodicity, no such requirement is needed for SFS. Computationally, SFS can be easily implemented using the Euler-Maruyama discretization. In theoretical analysis, we establish non-asymptotic error bounds for the sampling distribution of SFS in the Wasserstein distance under suitable conditions. We conduct numerical experiments to evaluate the performance of SFS and demonstrate that it is able to generate samples with better quality than several existing methods.
This paper has not been read by Pith yet.
Forward citations
Cited by 3 Pith papers
-
Variational Optimality of F\"ollmer Processes in Generative Diffusions
Föllmer processes are variationally optimal among generative diffusions because they minimize the impact of drift estimation error on path-space KL divergence, rendering different interpolation schedules statistically...
-
Optimal drift optimizer for non-convex optimization
Derives that the conditional terminal law of the optimal controlled process is a Gibbs measure on a proximally penalized energy, giving potential, averaged-gradient, and barycentric drift formulas with terminal-time g...
-
The Ensemble Schr{\"o}dinger Bridge filter for Nonlinear Data Assimilation
The Ensemble Schrödinger Bridge filter adds a diffusion-based analysis step to ensemble prediction, enabling effective nonlinear data assimilation without structural model error or training.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.