Home > Research > Publications & Outputs > Augmented pseudo-marginal Metropolis-Hastings f...

Electronic data

  • 2009.05318

    Rights statement: The final publication is available at Springer via http://dx.doi.org/10.1007/s11222-022-10083-5

    Accepted author manuscript, 436 KB, PDF document

    Embargo ends: 15/02/23

    Available under license: CC BY-NC: Creative Commons Attribution-NonCommercial 4.0 International License

Links

Text available via DOI:

View graph of relations

Augmented pseudo-marginal Metropolis-Hastings for partially observed diffusion processes

Research output: Contribution to Journal/MagazineJournal articlepeer-review

Published

Standard

Augmented pseudo-marginal Metropolis-Hastings for partially observed diffusion processes. / Golightly, Andrew; Sherlock, Chris.

In: Statistics and Computing, Vol. 32, 21, 30.06.2022.

Research output: Contribution to Journal/MagazineJournal articlepeer-review

Harvard

APA

Vancouver

Golightly A, Sherlock C. Augmented pseudo-marginal Metropolis-Hastings for partially observed diffusion processes. Statistics and Computing. 2022 Jun 30;32:21. Epub 2022 Feb 15. doi: 10.1007/s11222-022-10083-5

Author

Bibtex

@article{6f81b320ff4c4744baa5c80e94e1ce07,
title = "Augmented pseudo-marginal Metropolis-Hastings for partially observed diffusion processes",
abstract = "We consider the problem of inference for nonlinear, multivariate diffusion processes, satisfying It{\^o} stochastic differential equations (SDEs), using data at discrete times that may be incomplete and subject to measurement error. Our starting point is a state-of-the-art correlated pseudo-marginal Metropolis-Hastings algorithm, that uses correlated particle filters to induce strong and positive correlation between successive likelihood estimates. However, unless the measurement error or the dimension of the SDE is small, correlation can be eroded by the resampling steps in the particle filter. We therefore propose a novel augmentation scheme, that allows for conditioning on values of the latent process at the observation times, completely avoiding the need for resampling steps. We integrate over the uncertainty at the observation times with an additional Gibbs step. Connections between the resulting pseudo-marginal scheme and existing inference schemes for diffusion processes are made, giving a unified inference framework that encompasses Gibbs sampling and pseudo marginal schemes. The methodology is applied in three examples of increasing complexity. We find that our approach offers substantial increases in overall efficiency, compared to competing methods.",
keywords = "Stochastic differential equation, Bayesian inference, pseudo-marginal Metropolis-Hastings, data augmentation, linear noise approximation",
author = "Andrew Golightly and Chris Sherlock",
note = "The final publication is available at Springer via http://dx.doi.org/10.1007/s11222-022-10083-5",
year = "2022",
month = jun,
day = "30",
doi = "10.1007/s11222-022-10083-5",
language = "English",
volume = "32",
journal = "Statistics and Computing",
issn = "0960-3174",
publisher = "Springer Netherlands",

}

RIS

TY - JOUR

T1 - Augmented pseudo-marginal Metropolis-Hastings for partially observed diffusion processes

AU - Golightly, Andrew

AU - Sherlock, Chris

N1 - The final publication is available at Springer via http://dx.doi.org/10.1007/s11222-022-10083-5

PY - 2022/6/30

Y1 - 2022/6/30

N2 - We consider the problem of inference for nonlinear, multivariate diffusion processes, satisfying Itô stochastic differential equations (SDEs), using data at discrete times that may be incomplete and subject to measurement error. Our starting point is a state-of-the-art correlated pseudo-marginal Metropolis-Hastings algorithm, that uses correlated particle filters to induce strong and positive correlation between successive likelihood estimates. However, unless the measurement error or the dimension of the SDE is small, correlation can be eroded by the resampling steps in the particle filter. We therefore propose a novel augmentation scheme, that allows for conditioning on values of the latent process at the observation times, completely avoiding the need for resampling steps. We integrate over the uncertainty at the observation times with an additional Gibbs step. Connections between the resulting pseudo-marginal scheme and existing inference schemes for diffusion processes are made, giving a unified inference framework that encompasses Gibbs sampling and pseudo marginal schemes. The methodology is applied in three examples of increasing complexity. We find that our approach offers substantial increases in overall efficiency, compared to competing methods.

AB - We consider the problem of inference for nonlinear, multivariate diffusion processes, satisfying Itô stochastic differential equations (SDEs), using data at discrete times that may be incomplete and subject to measurement error. Our starting point is a state-of-the-art correlated pseudo-marginal Metropolis-Hastings algorithm, that uses correlated particle filters to induce strong and positive correlation between successive likelihood estimates. However, unless the measurement error or the dimension of the SDE is small, correlation can be eroded by the resampling steps in the particle filter. We therefore propose a novel augmentation scheme, that allows for conditioning on values of the latent process at the observation times, completely avoiding the need for resampling steps. We integrate over the uncertainty at the observation times with an additional Gibbs step. Connections between the resulting pseudo-marginal scheme and existing inference schemes for diffusion processes are made, giving a unified inference framework that encompasses Gibbs sampling and pseudo marginal schemes. The methodology is applied in three examples of increasing complexity. We find that our approach offers substantial increases in overall efficiency, compared to competing methods.

KW - Stochastic differential equation

KW - Bayesian inference

KW - pseudo-marginal Metropolis-Hastings

KW - data augmentation

KW - linear noise approximation

U2 - 10.1007/s11222-022-10083-5

DO - 10.1007/s11222-022-10083-5

M3 - Journal article

VL - 32

JO - Statistics and Computing

JF - Statistics and Computing

SN - 0960-3174

M1 - 21

ER -