A splitting Hamiltonian Monte Carlo method for efficient sampling

05/30/2021
by   Lei Li, et al.
0

We propose a splitting Hamiltonian Monte Carlo (SHMC) algorithm, which can be numerically and computationally efficient when combined with the random mini-batch strategy. By splitting the "effective potential energy" U∝ -β^-1logρ into two parts U = U_1 + U_2, one makes a proposal using the "easy-to-sample" part U_1, followed by probabilistically accepting that proposal by a Metropolis rejection step using U_2. The splitting allows efficient sampling from systems with singular potentials (or distributions with degenerate points) and/or with multiple potential barriers. In our SHMC algorithm, the proposal using U_1 is generated by the Hamiltonian dynamics, which can be potentially more efficient than the overdamped Langevin dynamics. We also use random batch strategies to reduce the computational cost to 𝒪(1) per time step in generating the proposals for problems arising from many-body systems and Bayesian inference, and prove that the errors of the Hamiltonian induced by the random batch approximation is 𝒪(√(Δ t)) in the strong and 𝒪(Δ t) in the weak sense, where Δ t is the time step. Numerical experiments are conducted to verify the theoretical results and the computational efficiency of the proposed algorithms in practice.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
07/15/2022

Split Hamiltonian Monte Carlo revisited

We study Hamiltonian Monte Carlo (HMC) samplers based on splitting the H...
research
07/10/2018

Multi-stage splitting integrators for sampling with modified Hamiltonian Monte Carlo methods

Modified Hamiltonian Monte Carlo (MHMC) methods combine the ideas behind...
research
08/29/2020

Random Batch Algorithms for Quantum Monte Carlo simulations

Random batch algorithms are constructed for quantum Monte Carlo simulati...
research
11/09/2020

Symmetrically processed splitting integrators for enhanced Hamiltonian Monte Carlo sampling

We construct integrators to be used in Hamiltonian (or Hybrid) Monte Car...
research
01/19/2012

Split HMC for Gaussian Process Models

In this paper, we discuss an extension of the Split Hamiltonian Monte Ca...
research
09/08/2017

Roll-back Hamiltonian Monte Carlo

We propose a new framework for Hamiltonian Monte Carlo (HMC) on truncate...
research
07/15/2021

Decentralized Bayesian Learning with Metropolis-Adjusted Hamiltonian Monte Carlo

Federated learning performed by a decentralized networks of agents is be...

Please sign up or login with your details

Forgot password? Click here to reset