Analysis of Langevin midpoint methods using an anticipative Girsanov theorem
By: Matthew S. Zhang
Potential Business Impact:
Makes computer sampling faster and more accurate.
We introduce a new method for analyzing midpoint discretizations of stochastic differential equations (SDEs), which are frequently used in Markov chain Monte Carlo (MCMC) methods for sampling from a target measure $\pi \propto \exp(-V)$. Borrowing techniques from Malliavin calculus, we compute estimates for the Radon-Nikodym derivative for processes on $L^2([0, T); \mathbb{R}^d)$ which may anticipate the Brownian motion, in the sense that they may not be adapted to the filtration at the same time. Applying these to various popular midpoint discretizations, we are able to improve the regularity and cross-regularity results in the literature on sampling methods. We also obtain a query complexity bound of $\widetilde{O}(\frac{\kappa^{5/4} d^{1/4}}{\varepsilon^{1/2}})$ for obtaining a $\varepsilon^2$-accurate sample in $\mathsf{KL}$ divergence, under log-concavity and strong smoothness assumptions for $\nabla^2 V$.
Similar Papers
Sublinear iterations can suffice even for DDPMs
Machine Learning (CS)
Makes AI create better pictures using less math.
A Simple Analysis of Discretization Error in Diffusion Models
Machine Learning (CS)
Makes AI create better pictures faster.
Poisson Midpoint Method for Log Concave Sampling: Beyond the Strong Error Lower Bounds
Probability
Speeds up computer learning by 3x.