Metropolis-adjusted Subdifferential Langevin Algorithm

2025/07/09 by Ning, Ning
#FOS: Computer and information sciences #Methodology (stat.ME)

paper · doi:10.48550/arxiv.2507.06950

Abstract

The Metropolis-Adjusted Langevin Algorithm (MALA) is a widely used Markov Chain Monte Carlo (MCMC) method for sampling from high-dimensional distributions. However, MALA relies on differentiability assumptions that restrict its applicability. In this paper, we introduce the Metropolis-Adjusted Subdifferential Langevin Algorithm (MASLA), a generalization of MALA that extends its applicability to distributions whose log-densities are locally Lipschitz, generally non-differentiable, and non-convex. We evaluate the performance of MASLA by comparing it with other sampling algorithms in settings where they are applicable. Our results demonstrate the effectiveness of MASLA in handling a broader class of distributions while maintaining computational efficiency.

Citations

• Subgradient Langevin Methods for Sampling from Non-smooth Potentials
• Convergence of Dirichlet Forms for MCMC Optimal Scaling with Dependent Target Distributions on Large Graphs
• Stochastic Langevin Differential Inclusions with Applications to Machine Learning
• Non-convex Learning via Replica Exchange Stochastic Gradient MCMC
• Conservative set valued fields, automatic differentiation, stochastic gradient method and deep learning
• Analysis of Langevin Monte Carlo via convex optimization
• Non-convex Optimization for Machine Learning
• Efficient Bayesian computation by proximal Markov chain Monte Carlo: when Langevin meets Moreau
• On the Exponentially Weighted Aggregate with the Laplace Prior
• Optimization Methods for Large-Scale Machine Learning
• Proximal Markov chain Monte Carlo algorithms
• Proximal Markov chain Monte Carlo algorithms
• The Bayesian Lasso
• Weak convergence and optimal scaling of random walk Metropolis algorithms

Related

• Proximal Markov chain Monte Carlo algorithms
• Proximal Markov chain Monte Carlo algorithms
• autoMALA: Locally adaptive Metropolis-adjusted Langevin algorithm
• Nonconvex sampling with the Metropolis-adjusted Langevin algorithm
• Particle Metropolis-adjusted Langevin algorithms