A generalized guided Monte Carlo algorithm

By A. Horowitz. 1991

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Stochastic Gradient Hamiltonian Monte Carlo (2014, background): “Our revised momentum update can also be viewed as akin to partial momentum refreshment (Horowitz, 1991; Neal, 1993), which also corresponds to second-order Langevin dynamics.”
Handbook of Markov Chain Monte Carlo (2011, neutral): “An alternative that can suppress random walk behaviour even when trajectories consist of just one leapfrog step is to only partially refresh the momentum between trajectories, as proposed by Horowitz (1991).”
Building a better probabilistic model of images by factorization (2011, methodology): “The underlying technique is introduced in [12], and is clearly presented in [24] (Sections 5.”
Improving Asymptotic Variance of MCMC Estimators: Non-reversible Chains are Better (2004, methodology): “ slow, diﬀusive movement via a random walk that is typical of reversible Markov chains. Examples include “overrelaxation” methods (Adler 1981; Neal 1998, 2003) and the “guided Monte Carlo” methods of Horowitz (1991) and Gustafson (1998). However, non-reversible Markov chains have often been avoided ”
Hamiltonian Monte Carlo Without Detailed Balance (2014, background): “Previous attempts to address this have combined multiple Markov steps that individually satisfy detailed balance into a composite step that does not (Horowitz, 1991), with limited success (Kennedy & Pendleton, 1991).”
Bayesian Sampling Using Stochastic Gradient Thermostats (2014, methodology): “Hybrid Monte Carlo (HMC) [6, 10], avoid this high degree of correlation by combining dynamic systems with the Metropolis step.”
MCMC Using Hamiltonian Dynamics (2011, background): “g steps. An alternative that can suppress random walk behaviour even when trajectories consist of just one leapfrog step is to only partially refresh the momentum between trajectories, as proposed by Horowitz (1991). Suppose that the kinetic energy has the typical form K(p) = pTM−1p/2. The following”
ANALYSIS OF A NONREVERSIBLE MARKOV CHAIN SAMPLER (2000, methodology): “In this paper, we analyze a nonreversible Markov chain that does a onedimensional walk, as an abstraction of these practical sampling methods, particularly that of Horowitz [21].”
The Bouncy Particle Sampler: A Nonreversible Rejection-Free Markov Chain Monte Carlo Method (2015, methodology): “This strategy is akin to the partial momentum refreshment strategy for HMC methods [12], [20, Section 4.”
The Geometric Foundations of Hamiltonian Monte Carlo (2014, background): “Under this artificial limitation, partial momentum refreshment schemes [36,66],”