Convergence of Griddy Gibbs sampling and other perturbed Markov chains. (English) Zbl 07192007
Summary: The Griddy Gibbs sampling was proposed by C. Ritter and M. A. Tanner [“Facilitating the Gibbs Sampler: the Gibbs Stopper and the Griddy-Gibbs Sampler”, J. Am. Stat. Assoc. 87, No. 419, 861–868 (1992; doi:10.1080/01621459.1992.10475289)] as a computationally efficient approximation of the well-known Gibbs sampling method. The algorithm is simple and effective and has been used successfully to address problems in various fields of applied science. However, the approximate nature of the algorithm has prevented it from being widely used: the Markov chains generated by the Griddy Gibbs sampling method are not reversible in general, so the existence and uniqueness of its invariant measure is not guaranteed. Even when such an invariant measure uniquely exists, there was no estimate of the distance between it and the probability distribution of interest, hence no means to ensure the validity of the algorithm as a means to sample from the true distribution. In this paper, we show, subject to some fairly natural conditions, that the Griddy Gibbs method has a unique, invariant measure. Moreover, we provide \(L^p\) estimates on the distance between this invariant measure and the corresponding measure obtained from Gibbs sampling. These results provide a theoretical foundation for the use of the Griddy Gibbs sampling method. We also address a more general result about the sensitivity of invariant measures under small perturbations on the transition probability. That is, if we replace the transition probability \(P\) of any Monte Carlo Markov chain by another transition probability \(Q\) where \(Q\) is close to \(P\), we can still estimate the distance between the two invariant measures. The distinguishing feature between our approach and previous work on convergence of perturbed Markov chain is that by considering the invariant measures as fixed points of linear operators on function spaces, we do not need to impose any further conditions on the rate of convergence of the Markov chain. For example, the results we derived in this paper can address the case when the considered Monte Carlo Markov chains are not uniformly ergodic.
MSC:
| 60J20 | Applications of Markov chains and discrete-time Markov processes on general state spaces (social mobility, learning theory, industrial processes, etc.) |
| 65C40 | Numerical analysis or methods applied to Markov chains |
| 92B05 | General biology and biomathematics |
| 32A70 | Functional analysis techniques applied to functions of several complex variables |
Keywords:
Griddy Gibbs; nonreversible Markov chain; perturbed Markov kernel; non-uniformly ergodic Markov chainSoftware:
AdMitReferences:
| [1] | Ritter C, Tanner MA. Facilitating the Gibbs Sampler: the Gibbs Stopper and the Griddy-Gibbs Sampler. J Am Stat Assoc. 1992;87(419):861-868. doi: 10.1080/01621459.1992.10475289[Taylor & Francis Online], [Web of Science ®], [Google Scholar] |
| [2] | Ray B, McCulloch R, Tsa R. Bayesian methods for change-point detection in long-range dependent processes. Stat Sin. 1997;7:451-472. [Google Scholar] |
| [3] | Barnard J, McCulloch R, Meng X. Modeling covariance matrices in terms of standard deviations and correlations, with application to shrinkage. Stat Sin. 2000;10:1281-1311. [Web of Science ®], [Google Scholar] · Zbl 0980.62045 |
| [4] | Yang H, Liu F, Ji C, et al. Adaptive sampling for Bayesian geospatial models. Statist Comput. 2014;24(6):1101-1110. doi: 10.1007/s11222-013-9422-4[Crossref], [Web of Science ®], [Google Scholar] · Zbl 1332.62367 |
| [5] | Kundu S, David BD. Bayes variable selection in semiparametric linear models. J Am Stat Assoc. 2014;109(505):437-447. doi: 10.1080/01621459.2014.881153[Taylor & Francis Online], [Web of Science ®], [Google Scholar] · Zbl 1367.62069 |
| [6] | Kundu S, David BD. Latent factor models for density estimation. Biometrika. 2014;101(3):641-654. doi: 10.1093/biomet/asu019[Crossref], [Web of Science ®], [Google Scholar] · Zbl 1334.62051 |
| [7] | Pham TH, Ormerod JT, Wand MP. Mean field variational Bayesian inference for nonparametric regression with measurement error. Comput Stat Data Anal. 2013;68:375-387. doi: 10.1016/j.csda.2013.07.014[Crossref], [Web of Science ®], [Google Scholar] · Zbl 1471.62160 |
| [8] | Mansinghka V, Kulkarni TD, Perov YN, et al. Approximate Bayesian image interpretation using generative probabilistic graphics programs. Adv Neural Inf Process Syst. 2013;26:1520-1528. [Google Scholar] |
| [9] | Ritter C. Statistical analysis of spectra from electron spectroscopy for chemical analysis. J R Stat Soc D - Stat. 1994;43:111-127. [Google Scholar] |
| [10] | Dong W, Tang X, Yu Y, et al. Systems biology of the clock in Neurospora crassa. PLoS ONE. 2008;3(8):e3105. doi: 10.1371/journal.pone.0003105[Crossref], [PubMed], [Web of Science ®], [Google Scholar] |
| [11] | Dinh V, Rundell AE, Buzzard GT. Experimental design for dynamics identification of cellular processes. Bull Math Biol. 2014;76(3):597-626. doi: 10.1007/s11538-014-9935-9[Crossref], [PubMed], [Web of Science ®], [Google Scholar] · Zbl 1329.92016 |
| [12] | Dinh V, Rundell AE, Buzzard GT. Effective sampling schemes for behavior discrimination in nonlinear systems. Int J Uncertain Quantification. 2014;4(6):535-554. doi: 10.1615/Int.J.UncertaintyQuantification.2014010728[Crossref], [Web of Science ®], [Google Scholar] · Zbl 1524.62527 |
| [13] | Li J, Yang X, Shoptaw S. A random-effects Markov transition model for Poisson-distributed repeated measures with non-ignorable missing values. Stat Med. 2007;26(12):2519-2532. doi: 10.1002/sim.2717[Crossref], [PubMed], [Web of Science ®], [Google Scholar] |
| [14] | Ardia D, Hoogerheide L, Dijk H. Adaptive mixture of Student-t distributions as a flexible candidate distribution for efficient simulation: the R Package AdMit. J Stat Softw. 2009;29:1-32. doi: 10.18637/jss.v029.i03[Crossref], [Web of Science ®], [Google Scholar] |
| [15] | Ausna M, Galeano P. Bayesian estimation of the Gaussian mixture GARCH model. Comput Stat Data Anal. 2007;51(5):2636-2652. doi: 10.1016/j.csda.2006.01.006[Crossref], [Web of Science ®], [Google Scholar] · Zbl 1161.62397 |
| [16] | Boatwright P, McCulloch R, Rossi P. Account-level modeling for trade promotion: an application of a constrained parameter hierarchical model. J Am Stat Assoc. 1999;94(448):1063-1073. doi: 10.1080/01621459.1999.10473859[Taylor & Francis Online], [Web of Science ®], [Google Scholar] |
| [17] | Bauwens L, Rombouts J. Bayesian inference for the mixed conditional heteroskedasticity model. Economet J. 2007;10(2):408-425. doi: 10.1111/j.1368-423X.2007.00213.x[Crossref], [Web of Science ®], [Google Scholar] · Zbl 1122.62015 |
| [18] | Walkera D, Prez Barberab F, Marion G. Stochastic modelling of ecological processes using hybrid Gibbs samplers. Ecol Model. 2006;198(1/2):40-52. doi: 10.1016/j.ecolmodel.2006.04.008[Crossref], [Web of Science ®], [Google Scholar] |
| [19] | Bauwens L, Rombouts J. Bayesian clustering of many GARCH models. Economet Rev. 2007;26(2-4):365-386. doi: 10.1080/07474930701220576[Taylor & Francis Online], [Web of Science ®], [Google Scholar] · Zbl 1112.62016 |
| [20] | Michalopoulou Z, Picarelli M. Gibbs sampling for time-delay and amplitude estimation in underwater acoustics. J Acoust Soc Am. 2005;117(2):799-8087. doi: 10.1121/1.1847894[Crossref], [PubMed], [Web of Science ®], [Google Scholar] |
| [21] | Boatwright P, McCulloch R, Rossi P. A multivariate time series model for the analysis and prediction of carbon monoxide atmospheric concentrations. J R Stat Soc C - Appl. 2001;50(2):187-200. doi: 10.1111/1467-9876.00228[Crossref], [Web of Science ®], [Google Scholar] |
| [22] | Ray B, Tsa R. Bayesian methods for change-point detection in long-range dependent processes. J Time Ser Anal. 2002;23(6):687-705. doi: 10.1111/1467-9892.00286[Crossref], [Web of Science ®], [Google Scholar] · Zbl 1114.62092 |
| [23] | Chen C, Wen Y. On goodness of fit for time series regression models. J Stat Comput Simul. 2001;69(3):239-256. doi: 10.1080/00949650108812093[Taylor & Francis Online], [Web of Science ®], [Google Scholar] · Zbl 1023.62087 |
| [24] | Tierney L. Markov chains for exploring posterior distributions. Ann Stat. 1994;22(4):1701-1728. doi: 10.1214/aos/1176325750[Crossref], [Web of Science ®], [Google Scholar] · Zbl 0829.62080 |
| [25] | Liu SJ, Liang F, Wong WH. The multiple-try method and local optimization in metropolis sampling. J Am Stat Assoc. 95(449):121-134. doi: 10.1080/01621459.2000.10473908[Taylor & Francis Online], [Web of Science ®], [Google Scholar] · Zbl 1072.65505 |
| [26] | Gilks WR, Wild P. Adaptive rejection sampling for Gibbs sampling. Appl Stat. 1992;1(2):337-48. doi: 10.2307/2347565[Crossref], [Web of Science ®], [Google Scholar] · Zbl 0825.62407 |
| [27] | Roberts GO, Rosenthal JS, Schwartz PO. Convergence properties of perturbed Markov chains. J Appl Probab. 1998;35(1):1-11. doi: 10.1017/S0021900200014625[Crossref], [Web of Science ®], [Google Scholar] · Zbl 0909.60049 |
| [28] | Mitrophanov AY. Sensitivity and convergence of uniformly ergodic Markov chains. J Appl Probab. 2005;42(04):1003-1014. doi: 10.1017/S0021900200001066[Crossref], [Google Scholar] · Zbl 1092.60027 |
| [29] | Andrieu C, Roberts GO. The pseudo-marginal approach for efficient Monte Carlo computations. Ann Stat. 2009;37(2):697-725. doi: 10.1214/07-AOS574[Crossref], [Web of Science ®], [Google Scholar] · Zbl 1185.60083 |
| [30] | Roberts GO, Rosenthal JS. General state space Markov chains and MCMC algorithms. Probab Surv. 2004;1:20-71. doi: 10.1214/154957804100000024[Crossref], [Google Scholar] · Zbl 1189.60131 |
| [31] | Athreya K, Doss H, Sethuraman J. On the convergence of the Markov Chain simulation method. Ann Stat. 1996;24(1):69-100. doi: 10.1214/aos/1033066200[Crossref], [Web of Science ®], [Google Scholar] · Zbl 0860.60057 |
| [32] | Koralov L, Sinai Y. Theory of probability and random processes. Berlin: Springer; 2007. [Crossref], [Google Scholar] · Zbl 1181.60004 |
| [33] | Pedersen M. Functional analysis in applied mathematics and engineering. Boca Raton (FL): CRC Press; 1999. [Google Scholar] · Zbl 0936.46001 |
| [34] | Lasser R. Introduction to Fourier analysis. New York (NY): CRC Press; 1996. [Google Scholar] · Zbl 0848.42001 |
| [35] | Lillacci G, Khammash G. A distribution-matching method for parameter estimation and model selection in computational biology. Int J Robust Nonlinear Control. 2012;22(10):1065-1081. doi: 10.1002/rnc.2794[Crossref], [Web of Science ®], [Google Scholar] · Zbl 1258.93111 |
| [36] | Lipniacki T, Hat B, Faeder JR, et al. Stochastic effects and bistability in T cell receptor signaling. J Theoret Biol. 2008;254(1):110-122. doi: 10.1016/j.jtbi.2008.05.001[Crossref], [PubMed], [Web of Science ®], [Google Scholar] · Zbl 1400.92173 |
| [37] | Donahue MM, Buzzard GT, Rundell AE. Experiment design through dynamical characterisation of non-linear systems biology models utilising sparse grids. IET Syst Biol. 2010;4(4):249-262. doi: 10.1049/iet-syb.2009.0031[Crossref], [PubMed], [Web of Science ®], [Google Scholar] |
| [38] | Buzzard GT. Global sensitivity analysis using sparse grid interpolation and polynomial chaos. Reliab Eng Syst Saf. 2012;17:82-89. doi: 10.1016/j.ress.2011.07.011[Crossref], [Web of Science ®], [Google Scholar] |
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.
