2015
DOI: 10.1016/j.jcp.2014.09.037
|View full text |Cite
|
Sign up to set email alerts
|

Extra Chance Generalized Hybrid Monte Carlo

Abstract: We study a method, Extra Chance Generalized Hybrid Monte Carlo, to avoid rejections in the Hybrid Monte Carlo method and related algorithms. In the spirit of delayed rejection, whenever a rejection would occur, extra work is done to find a fresh proposal that, hopefully, may be accepted. We present experiments that clearly indicate that the additional work per sample carried out in the extra chance approach clearly pays in terms of the quality of the samples generated.

Help me understand this report
View preprint versions

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
36
0

Year Published

2017
2017
2024
2024

Publication Types

Select...
6
2
1

Relationship

0
9

Authors

Journals

citations
Cited by 30 publications
(36 citation statements)
references
References 29 publications
(76 reference statements)
0
36
0
Order By: Relevance
“…In order to suppress the random walk behaviour induced by the introduction of the accept-reject step needed to ensure the detailed balance it has been suggested to modify the Metropolis acceptreject step [17,18]. The resulting probability distribution satisfies the fixed point equation but drops the detailed balance condition.…”
Section: Look Ahead Hmcmentioning
confidence: 99%
See 1 more Smart Citation
“…In order to suppress the random walk behaviour induced by the introduction of the accept-reject step needed to ensure the detailed balance it has been suggested to modify the Metropolis acceptreject step [17,18]. The resulting probability distribution satisfies the fixed point equation but drops the detailed balance condition.…”
Section: Look Ahead Hmcmentioning
confidence: 99%
“…Using the notation introduced in section 2.2 the new LAHMC algorithm [17,18] consists in the following steps:…”
Section: Look Ahead Hmcmentioning
confidence: 99%
“…In future work, we will examine the bias vs. efficiency trade-offs of our neglect of the shadow work, and the extent to which these can be mitigated by choice of Langevin integrator, or by using reduced-momentum-flipping variants of Hamiltonian Monte Carlo. [62][63][64] While it has been argued more generically that the contribution of "shadow work" in nonequilibrium simulations can be neglected without introducing much bias, 53 this is likely highly dependent on the specific choice of integrator used, so we recommend caution if other integrators are considered.…”
Section: Nonequilibrium Candidate Monte Carlomentioning
confidence: 99%
“…This proof is based on the proof of stationarity given in [8]. Let x = (q, p, z), and let P (x) be the pdf of the target distribution.…”
Section: Proof Of Stationaritymentioning
confidence: 99%