Provides lightweight, dependency-minimal implementations of
Langevin diffusion based Markov chain Monte Carlo samplers, including
the Unadjusted Langevin Algorithm (ULA) and the Metropolis-Adjusted
Langevin Algorithm (MALA). The core sampling loops are written in C++
via 'Rcpp' and 'RcppArmadillo' for performance, while exposing a simple
R-level interface where the user supplies the gradient of the negative
log-density (and, for MALA, the negative log-density itself). Intended
as a building block for Bayesian inference and stochastic optimization
rather than a full probabilistic programming framework. Methods follow
Roberts and Tweedie (1996)