nimbleHMC

Hamiltonian Monte Carlo and Other Gradient-Based MCMC Sampling Algorithms for 'nimble'

Daniel Turek, Perry de Valpine, Christopher Paciorek

Last update: 8 February 2024

Copyright license: GNU General Public License, version 3.0, GNU General Public License, version 2.0, 3-clause BSD License, File License

Software version identifier: 0.1.0, 0.1.1, 0.2.0, 0.2.1

Description

Provides gradient-based MCMC sampling algorithms for use with the MCMC engine provided by the 'nimble' package. This includes two versions of Hamiltonian Monte Carlo (HMC) No-U-Turn (NUTS) sampling, and (under development) Langevin samplers. The 'NUTS_classic' sampler implements the original HMC-NUTS algorithm as described in Hoffman and Gelman (2014) <arXiv:1111.4246>. The 'NUTS' sampler is a modern version of HMC-NUTS sampling matching the HMC sampler available in version 2.32.2 of Stan (Stan Development Team, 2023). In addition, convenience functions are provided for generating and modifying MCMC configuration objects which employ HMC sampling.

Cites work

The No-U-Turn Sampler: Adaptively Setting Path Lengths in Hamiltonian Monte Carlo

This page was built for software: nimbleHMC