Search

It is of significant interest in many applications to sample from a high-dimensional target distribution $\pi$ with the density $\pi(\text{d} x) \propto e^{-U(x)} (\text{d} x) $, based on the temporal discretization of the Langevin stochastic differential equations (SDEs). In this paper, we propose an explicit projected Langevin Monte Carlo (PLMC) algorithm with non-convex potential $U$ and super-linear gradient of $U$ and investigate the non-asymptotic analysis of its sampling error in total variation distance. Equipped with time-independent regularity estimates for the corresponding Kolmogorov equation, we derive the non-asymptotic bounds on the total variation distance between the target distribution of the Langevin SDEs and the law induced by the PLMC scheme with order $\mathcal{O}(h |\ln h|)$. Moreover, for a given precision $\epsilon$, the smallest number of iterations of the classical Langevin Monte Carlo (LMC) scheme with the non-convex potential $U$ and the globally Lipshitz gradient of $U$ can be guaranteed by order ${\mathcal{O}}\big(\tfrac{d^{3/2}}{\epsilon} \cdot \ln (\tfrac{d}{\epsilon}) \cdot \ln (\tfrac{1}{\epsilon}) \big)$. Numerical experiments are provided to confirm the theoretical findings., Comment: 31 pages, 6 figures

This article investigates the weak approximation towards the invariant measure of semi-linear stochastic differential equations (SDEs) under non-globally Lipschitz coefficients. For this purpose, we propose a linear-theta-projected Euler (LTPE) scheme, which also admits an invariant measure, to handle the potential influence of the linear stiffness. Under certain assumptions, both the SDE and the corresponding LTPE method are shown to converge exponentially to the underlying invariant measures, respectively. Moreover, with time-independent regularity estimates for the corresponding Kolmogorov equation, the weak error between the numerical invariant measure and the original one can be guaranteed with convergence of order one. In terms of computational complexity, the proposed ergodicity preserving scheme with the nonlinearity explicitly treated has a significant advantage over the ergodicity preserving implicit Euler method in the literature. Numerical experiments are provided to verify our theoretical findings., Comment: 37 pages, 7 figures

In the field of computational finance, it is common for the quantity of interest to be expected values of functions of random variables via stochastic differential equations (SDEs). For SDEs with globally Lipschitz coefficients and commutative diffusion coefficients, the explicit Milstein scheme, relying on only Brownian increments and thus easily implementable, can be combined with the multilevel Monte Carlo (MLMC) method proposed by Giles \cite{giles2008multilevel} to give the optimal overall computational cost $\mathcal{O}(\epsilon^{-2})$, where $\epsilon$ is the required target accuracy. For multi-dimensional SDEs that do not satisfy the commutativity condition, a kind of one-half order truncated Milstein-type scheme without L\'evy areas is introduced by Giles and Szpruch \cite{giles2014antithetic}, which combined with the antithetic MLMC gives the optimal computational cost under globally Lipschitz conditions. In the present work, we turn to SDEs with non-globally Lipschitz continuous coefficients, for which a family of modified Milstein-type schemes without L\'evy areas is proposed. The expected one-half order of strong convergence is recovered in a non-globally Lipschitz setting, where the diffusion coefficients are allowed to grow superlinearly. This helps us to analyze the relevant variance of the multilevel estimator and the optimal computational cost is finally achieved for the antithetic MLMC. The analysis of both the convergence rate and the desired variance in the non-globally Lipschitz setting is highly non-trivial and non-standard arguments are developed to overcome some essential difficulties. Numerical experiments are provided to confirm the theoretical findings., Comment: 39 pages, 4 figures