Skip to main content
SpringerLink
Log in

On the flexibility of the design of multiple try Metropolis schemes

  • Original Paper
  • Published:
Computational Statistics Aims and scope Submit manuscript

Abstract

The multiple try Metropolis (MTM) method is a generalization of the classical Metropolis–Hastings algorithm in which the next state of the chain is chosen among a set of samples, according to normalized weights. In the literature, several extensions have been proposed. In this work, we show and remark upon the flexibility of the design of MTM-type methods, fulfilling the detailed balance condition. We discuss several possibilities, show different numerical simulations and discuss the implications of the results.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2
Fig. 3
Fig. 4

Similar content being viewed by others

Notes

  1. Recall that \(y_i\) are drawn from \(\pi _i(\cdot |x)\) whereas \(x_i^*\) are drawn from \(\pi _i(\cdot |y),\,i=1,\ldots ,k-1,k+1,\ldots N\) and \(x_k^*=x_{t}=x\).

  2. Note that the balance condition is a sufficient but not necessary condition. Namely, the detailed balance ensures invariance. The converse is not true. Markov chains that satisfy the detailed balance condition are called reversible.

  3. However, it is important to remark that high acceptance rates are not a suitable indicator of good performance since, in general, the best acceptance rate is different from 1 (Roberts et al. 1997).

  4. Another simple MTM scheme is the “orientational bias Monte Carlo” (Frenkel and Smit 1996, Chapter 13). In this case, the proposal pdf must be symmetric, i.e., \(\pi (y|x)=\pi (x|y)\), and the weights must be proportional to the target, i.e., \(\omega (y_i)=p(y_i),\,i=1,\ldots ,N\).

  5. Note that, in this work, we have mainly considered scalar variables in order to simplify the treatment and the notation. All the considerations and algorithms contained in this work are also valid for multi-dimensional variables (see, for instance, the last numerical example in Sect. 6.6).

  6. We do not provide the estimated linear correlation because of the moments (as the mean, for instance) of the target do not exist, and it makes difficult a right estimation of the correlation.

References

  • Barker AA (1965) Monte Carlo calculations of the radial distribution functions for a proton–electron plasma. Aust J Phys 18:119–133

    Article  Google Scholar 

  • Bédard M, Douc R, Mouline E (2012) Scaling analysis of multiple-try MCMC methods. Stoch Process Appl 122:758–786

    Article  MATH  Google Scholar 

  • Brooks SP (1998) Markov Chain Monte Carlo method and its application. J R Stat Soc Ser D (The Statistician) 47(1):69–100

    Article  Google Scholar 

  • Casarin R, Craiu R, Leisen F (2013) Interacting multiple try algorithms with different proposal distributions. Stat Comput 23(2):185–200. doi:10.1007/s11222-011-9301-9

    Google Scholar 

  • Craiu RV, Lemieux C (2007) Acceleration of the multiple-try Metropolis algorithm using antithetic and stratified sampling. Stat Comput 17(2):109–120

    Article  MathSciNet  Google Scholar 

  • Devroye L (1986) Non-uniform random variate generation. Springer, Berlin

    Book  MATH  Google Scholar 

  • Fitzgerald WJ (2001) Markov Chain Monte Carlo methods with applications to signal processing. Signal Process 81(1):3–18

    Article  MATH  Google Scholar 

  • Frenkel D, Smit B (1996) Understanding molecular simulation: from algorithms to applications. Academic Press, San Diego

    MATH  Google Scholar 

  • Gamerman D, Lopes HF (2006) Markov Chain Monte Carlo: stochastic simulation for Bayesian inference. Chapman and Hall/CRC, Boca Raton

    Google Scholar 

  • Gilks WR, Richardson S, Spiegelhalter D (1995) Markov Chain Monte Carlo in practice: interdisciplinary statistics. Taylor & Francis, Inc., UK

    Google Scholar 

  • Haario H, Saksman E, Tamminen J (1999) Adaptive proposal distribution for random walk Metropolis algorithm. Comput Stat 14:375–395

    Article  MATH  Google Scholar 

  • Haario H, Saksman E, Tamminen J (2001) An adaptive Metropolis algorithm. Bernoulli 7(2):223–242

    Article  MathSciNet  MATH  Google Scholar 

  • Hastings WK (1970) Monte Carlo sampling methods using Markov chains and their applications. Biometrika 57(1):97–109

    Article  MATH  Google Scholar 

  • Lan S, Stathopoulosy V, Shahbaba B, Girolami M (2012) Langrangian dynamical Monte Carlo. arXiv:1211.3759v1

  • Liang F, Liu C, Caroll R (2010) Advanced Markov Chain Monte Carlo methods: learning from past samples. Wiley Series in Computational Statistics, England

  • Liu JS, Liang F, Wong WH (2000) The multiple-try method and local optimization in Metropolis sampling. J Am Stat Assoc 95(449):121–134

    Article  MathSciNet  MATH  Google Scholar 

  • Liu JS (2004) Monte Carlo strategies in scientific computing. Springer, Berlin

    Book  Google Scholar 

  • Martino L, Del Olmo VP, Read J (2012a) A multi-point Metropolis scheme with generic weight functions. Stat Probab Lett 82(7):1445–1453

    Article  MATH  Google Scholar 

  • Martino L, Read J, Luengo D (2012b) Improved adaptive rejection Metropolis sampling algorithms. arXiv:1205.5494v4

  • Metropolis N, Rosenbluth A, Rosenbluth M, Teller A, Teller E (1953) Equations of state calculations by fast computing machines. J Chem Phys 21:1087–1091

    Article  Google Scholar 

  • Mira A (2001) On Metropolis–Hastings algorithms with delayed rejection. Metron 59:231–241

    MathSciNet  MATH  Google Scholar 

  • Pandolfi S, Bartolucci F, Friel N (2010) A generalization of the multiple-try Metropolis algorithm for Bayesian estimation and model selection. J Mach Learn Res (Workshop and conference proceedings volume 9: AISTATS 2010) 9:581–588

    Google Scholar 

  • Peskun PH (1973) Optimum Monte-Carlo sampling using Markov Chains. Biometrika 60(3):607–612

    Article  MathSciNet  MATH  Google Scholar 

  • Qin ZS, Liu JS (2001) Multi-point Metropolis method with application to hybrid Monte Carlo. J Comput Phys 172:827–840

    Article  MathSciNet  MATH  Google Scholar 

  • Robert CP (2012) “Xi’ An’s Og, an attempt at bloggin... ” Blog (by Christian P. Robert). http://xianblog.wordpress.com/2012/01/23/multiple-trypoint-metropolis-algorithm/

  • Robert CP, Casella G (2004) Monte Carlo statistical methods. Springer, Berlin

    Book  MATH  Google Scholar 

  • Roberts GO, Gelman A, Gilks WR (1997) Weak convergence and optimal scaling of random walk Metropolis algorithms. Ann Appl Probab 7:110–120

    Article  MathSciNet  MATH  Google Scholar 

  • Storvik G (2011) On the flexibility of Metropolis–Hastings acceptance probabilities in auxiliary variable proposal generation. Scand J Stat 38(2):342–358

    Article  MathSciNet  MATH  Google Scholar 

  • Tierney L (1994) Markov chains for exploring posterior distributions. Ann Stat 33:1701–1728

    Article  MathSciNet  Google Scholar 

  • Tierney L, Mira A (1999) Some adaptive Monte Carlo methods for Bayesian inference. Stat Med 18:2507–2515

    Article  Google Scholar 

  • Zhang Y, Zhang W (2012) Improved generic acceptance function for multi-point Metropolis algorithm. In: 2nd International conference on electronic and mechanical engineering and information technology (EMEIT-2012)

Download references

Acknowledgments

We would like to thank the Reviewers for their comments which have helped us to improve the first version of manuscript. Moreover, this work has been partially supported by Ministerio de Ciencia e Innovacin and by the Ministerio de Economía of Spain (project MONIN, ref. TEC-2006-13514-C02- 01/TCM, project COMONSENS, id. CSD2008-00010, project DEIPRO ref. TEC2009-14504-C02-01, project ALCIT, id. TEC2012-38800-C03-01 and project COMPREHENSION, id. TEC2012-38883-C02-01) and Comunidad Autonoma de Madrid (project PROMULTIDIS-CM, ref. S-0505/TIC/0233).

Author information

Authors and Affiliations

  1. Department of Signal Theory and Communications, Universidad Carlos III de Madrid, Avenida de la Universidad 30, 28911 , Leganés, Madrid, Spain

    Luca Martino & Jesse Read

Authors
  1. Luca Martino
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Jesse Read
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Luca Martino.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Martino, L., Read, J. On the flexibility of the design of multiple try Metropolis schemes. Comput Stat 28, 2797–2823 (2013). https://doi.org/10.1007/s00180-013-0429-2

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00180-013-0429-2

Keywords

Search

Navigation