×

Augmented pseudo-marginal Metropolis-Hastings for partially observed diffusion processes. (English) Zbl 1482.62011

Summary: 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.

MSC:

62-08 Computational methods for problems pertaining to statistics
PDFBibTeX XMLCite
Full Text: DOI arXiv

References:

[1] Andrieu, C.; Doucet, A.; Holenstein, R.; L’Ecuyer, P.; Owen, AB, Particle Markov chain Monte Carlo for efficient numerical simulation, Monte Carlo and Quasi-Monte Carlo methods 2008, 45-60 (2009), Heidelberg: Spinger, Heidelberg · Zbl 1184.65001 · doi:10.1007/978-3-642-04107-5_3
[2] Andrieu, C.; Doucet, A.; Holenstein, R., Particle Markov chain Monte Carlo methods (with discussion), J. R. Statist. Soc. B, 72, 3, 1-269 (2010) · Zbl 1411.65020 · doi:10.1111/j.1467-9868.2009.00736.x
[3] Andrieu, C.; Roberts, GO, The pseudo-marginal approach for efficient computation, Annal. Statistics, 37, 697-725 (2009) · Zbl 1185.60083 · doi:10.1214/07-AOS574
[4] Arkin, A.; Ross, J.; McAdams, HH, Stochastic kinetic analysis of developmental pathway bifurcation in phage \(\lambda \)-infected Escherichia coli cells, Genetics, 149, 1633-1648 (1998)
[5] Arnaudon, A., van der Meulen, F., Schauer, M., and Sommer, S.: Diffusion bridges for stochastic Hamiltonian systems with applications to shape analysis. Available from http://arxiv.org/abs/2002.00885(2020)
[6] Bérard, J.; Del Moral, P.; Doucet, A., A lognormal central limit theorem for particle approximations of normalizing constants, Electron. J. Prob., 19, 1-28 (2014) · Zbl 1308.65014
[7] Botha, I.: Bayesian inference for stochastic differential equation mixed effects models. Mphil thesis, Queensland University of Technology(2020)
[8] Chen, N.; Giannakis, D.; Herbei, R.; Majda, AJ, An MCMC algorithm for parameter estimation in signals with hidden intermittent instability, SIAM/ASA J. Uncertain. Quantification, 2, 647-669 (2014) · Zbl 1350.60071 · doi:10.1137/130944977
[9] Choppala, P., Gunawan, D., Chen, J., Tran, M.-N., and Kohn, R.:Bayesian inference for state space models using block and correlated pseudo marginal methods. Available from http://arxiv.org/abs/1311.3606(2016)
[10] Dahlin, J., Lindsten, F., Kronander, J., and Schön, T. B. Accelerating pseudo-marginal Metropolis-Hastings by correlating auxiliary variables. Available from https://arxiv.1511.05483v1(2015)
[11] Del Moral, P., Feynman-Kac Formulae: Genealogical and Interacting Particle Systems with Applications (2004), New York: Springer, New York · Zbl 1130.60003 · doi:10.1007/978-1-4684-9393-1
[12] Deligiannidis, G.; Doucet, A.; Pitt, MK, The correlated pseudo-marginal method, J. R. Soc. Series B (Statistic. Methodol.), 80, 5, 839-870 (2018) · Zbl 1407.62074 · doi:10.1111/rssb.12280
[13] Doucet, A.; Pitt, MK; Deligiannidis, G.; Kohn, R., Efficient implementation of Markov chain Monte Carlo when using an unbiased likelihood estimator, Biometrika, 102, 2, 295-313 (2015) · Zbl 1452.62055 · doi:10.1093/biomet/asu075
[14] Duane, S.; Kennedy, AD; Pendleton, BJ; Roweth, D., Hybrid Monte Carlo, Phys. Lett. B, 195, 216-222 (1987) · doi:10.1016/0370-2693(87)91197-X
[15] Durham, GB; Gallant, RA, Numerical techniques for maximum likelihood estimation of continuous time diffusion processes, J. Bus. Econ. Stat., 20, 279-316 (2002) · doi:10.1198/073500102288618397
[16] Fearnhead, P.; Giagos, V.; Sherlock, C., Inference for reaction networks using the linear noise approximation, Biometrics, 70, 456-457 (2014) · Zbl 1419.62346 · doi:10.1111/biom.12152
[17] Fearnhead, P.; Meligkotsidou, L., Augmentation schemes for particle MCMC, Statistics and Computing, 26, 1293-1306 (2016) · Zbl 1356.65025 · doi:10.1007/s11222-015-9603-4
[18] Feller, W., The parabolic differential equations and the associated semi-groups of transformations, Annal. Math., 55, 468-519 (1952) · Zbl 0047.09303 · doi:10.2307/1969644
[19] Fuchs, C., Inference for Diffusion Processes with Applications in Life Sciences (2013), Heidelberg: Springer, Heidelberg · Zbl 1276.62051 · doi:10.1007/978-3-642-25969-2
[20] Golightly, A.; Bradley, E.; Lowe, T.; Gillespie, CS, Correlated pseudo-marginal schemes for time-discretised stochastic kinetic models, CSDA, 136, 92-107 (2019) · Zbl 1507.62064
[21] Golightly, A.; Wilkinson, DJ, Bayesian inference for stochastic kinetic models using a diffusion approximation, Biometrics, 61, 3, 781-788 (2005) · Zbl 1079.62110 · doi:10.1111/j.1541-0420.2005.00345.x
[22] Golightly, A.; Wilkinson, DJ, Bayesian inference for nonlinear multivariate diffusion models observed with error, Comput. Statistics Data Anal., 52, 3, 1674-1693 (2008) · Zbl 1452.62603 · doi:10.1016/j.csda.2007.05.019
[23] Golightly, A., Wilkinson, D.J.: Markov chain Monte Carlo algorithms for SDE parameter estimation. In: Lawrence, N.D., Girolami, M., Rattray, M., Sanguinetti, G. (eds.) Learning and Inference in Computational Systems Biology. MIT Press (2010)
[24] Golightly, A.; Wilkinson, DJ, Bayesian parameter inference for stochastic biochemical network models using particle Markov chain Monte Carlo, Interface Focus, 1, 6, 807-820 (2011) · doi:10.1098/rsfs.2011.0047
[25] Kalogeropoulos, K.; Roberts, G.; Dellaportas, P., Inference for stochastic volatility models using time change transformations, Annal. Statistics, 38, 784-807 (2010) · Zbl 1189.91220 · doi:10.1214/09-AOS702
[26] Majda, AJ; Franzke, C.; Crommelin, D., Normal forms for reduced stochastic climate models, PNAS, 106, 3649-3653 (2009) · Zbl 1202.86011 · doi:10.1073/pnas.0900173106
[27] Nemeth, C.; Fearnhead, P.; Mihaylova, L., Particle approximations of the score and observed information matrix for parameter estimation in state space models with linear computational cost, J. Comput. Gr. Statistics, 25, 1138-1157 (2016) · doi:10.1080/10618600.2015.1093492
[28] Owen, J.; Wilkinson, DJ; Gillespie, CS, Scalable inference for Markov processes with intractable likelihoods, Statistics Comput., 25, 145-156 (2015) · Zbl 1331.62065 · doi:10.1007/s11222-014-9524-7
[29] Papaspiliopoulos, O.; Roberts, GO; Stramer, O., Data augmentation for diffusions, J. Comput. Gr. Statistics, 22, 665-688 (2013) · doi:10.1080/10618600.2013.783484
[30] Picchini, U.; Forman, JL, Bayesian inference for stochastic differential equation mixed effects models of a tumour xenography study, J. R. Statistical Soc. Series C, 68, 887-913 (2019) · doi:10.1111/rssc.12347
[31] Pitt, MK; dos Santos Silva, R.; Giordani, P.; Kohn, R., On some properties of Markov chain Monte Carlo simulation methods based on the particle filter, J. Econometrics, 171, 2, 134-151 (2012) · Zbl 1443.62499 · doi:10.1016/j.jeconom.2012.06.004
[32] Plummer, M.; Best, N.; Cowles, K.; Vines, K., CODA: convergence diagnosis and output analysis for MCMC, R News, 6, 1, 7-11 (2006)
[33] Poyiadjis, G.; Doucet, A.; Singh, SS, Particle approximations of the score and observed information matrix in state space models with application to parameter estimation, Biometrika, 98, 65-80 (2011) · Zbl 1214.62093 · doi:10.1093/biomet/asq062
[34] Roberts, GO; Stramer, O., On inference for non-linear diffusion models using Metropolis-Hastings algorithms, Biometrika, 88, 3, 603-621 (2001) · Zbl 0985.62066 · doi:10.1093/biomet/88.3.603
[35] Schauer, M.; van der Meulen, F.; van Zanten, H., Guided proposals for simulating multi-dimensional diffusion bridges, Bernoulli, 23, 2917-2950 (2017) · Zbl 1415.65022 · doi:10.3150/16-BEJ833
[36] Sherlock, C.; Thiery, A.; Roberts, GO; Rosenthal, JS, On the effciency of pseudo-marginal random walk Metropolis algorithms, Annal. Statistics, 43, 1, 238-275 (2015) · Zbl 1326.65015 · doi:10.1214/14-AOS1278
[37] Stathopoulos, V.; Girolami, MA, Markov chain Monte Carlo inference for Markov jump processes via the linear noise approximation, Philosophic. Transactions R. Soc. A, 371, 20110541 (2013) · Zbl 1353.60075 · doi:10.1098/rsta.2011.0541
[38] Stramer, O.; Bognar, M., Bayesian inference for irreducible diffusion processes using the pseudo-marginal approach, Bayesian Anal., 6, 231-258 (2011) · Zbl 1330.60092 · doi:10.1214/11-BA608
[39] Stramer, O.; Shen, X.; Bognar, M., Bayesian inference for Heston-STAR models, Statistics Comput., 27, 331-348 (2017) · Zbl 1505.62388 · doi:10.1007/s11222-015-9625-y
[40] Tran, M.-N., Kohn, R., Quiroz, M., and Villani, M.:Block-wise pseudo-marginal Metropolis-Hastings. Available from http://arxiv.org/abs/1603.02485(2016)
[41] van der Meulen, F.; Schauer, M., Bayesian estimation of discretely observed multi-dimensional diffusion processes using guided proposals, Electron. J. Statistics, 11, 2358-2396 (2017) · Zbl 1378.62050
[42] Whitaker, GA; Golightly, A.; Boys, RJ; Sherlock, C., Improved bridge constructs for stochastic differential equations, Statistics Comput., 27, 885-900 (2017) · Zbl 1384.65011 · doi:10.1007/s11222-016-9660-3
[43] Wilkinson, DJ, Stochastic Modelling for Systems Biology (2018), Boca Raton, Florida: Chapman & Hall/CRC Press, Boca Raton, Florida · Zbl 1403.92003
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.