Langevin Dynamics
Langevin dynamics is a core stochastic method in Bayesian sampling. It constructs a stochastic differential equation by adding random noise to a gradient descent direction, such that the stationary distribution of the resulting process equals the target posterior distribution. The approach is widely used in probabilistic modeling and Bayesian inference.
Timeline (last 90 days)
Traditional Langevin dynamics relies on gradient information of the target distribution's density function and performs well when the density function is smooth and differentiable.
Traditional first-order Langevin algorithms (such as projected Langevin Monte Carlo) require that the log-density of the target distribution be differentiable when handling sampling on constrained domains.
All Facts (2)
Traditional Langevin dynamics relies on gradient information of the target distribution's density function and performs well when the density function is smooth and differentiable.
50%UnverifiedTraditional first-order Langevin algorithms (such as projected Langevin Monte Carlo) require that the log-density of the target distribution be differentiable when handling sampling on constrained domains.
50%