Statistics and Computing

, Volume 1, Issue 1, pp 1–12 | Cite as

Positive embedded integration in Bayesian analysis

  • Petros Dellaportas
  • David Wright
Papers

Abstract

This paper describes how embedded sequences of positive interpolatory integration rules (PIIRs) obtained from Gauss-Hermite product rules can be applied in Bayesian analysis. These embedded sequences are very promising for two major reasons. First, they provide a rich class of spatially distributed rules which are particularly useful in high dimensions. Second, they provide a way of producing more efficient integration strategies by enabling approximations to be updated sequentially through the addition of new nodes at each step rather than through changing to a completely new set of nodes. Moreover, as points are added successive rules change naturally from spatially distributed non-product rules to product rules. This feature is particularly attractive when the rules are used for the evaluation of marginal posterior densities. We illustrate the use of embedded sequences of PIIRs in two examples. These illustrate how embedded sequences can be applied to improve the efficiency of the adaptive integration strategy currently in use.

Keywords

Bayesian inference numerical integration Gauss-Hermite formulae embedded sequences positive interpolatory integration rules 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. Berntsen, J. and Espelid, T. (1984) On the use of Gauss-quadrature in adaptive integration schemes.BIT,24, 239–242.Google Scholar
  2. Davis, P. J. and Rabinowitz, P. (1984)Methods of Numerical Integration (2nd edn), Academic Press, Orlando, Florida.Google Scholar
  3. Dellaportas, P. (1990) Imbedded integration rules and their applications in Bayesian analysis. Unpublished PhD thesis, Department of Mathematics and Statistics, Polytechnic South West, Plymouth.Google Scholar
  4. Genz, A. (1986) Fully symmetric interpolatory rules for multiple integrals.SIAM J. Numer. Anal. 23, 6, 1273–1283.Google Scholar
  5. Grieve, A. P. (1987) Applications of Bayesian software: two examples.The Statistician,36, 283–288.Google Scholar
  6. Keast, P. and Lyness, J. N. (1979) On the structure of fully symmetric multidimensional quadrature rules.SIAM J. Numer. Anal. 16, 11–29.Google Scholar
  7. Laurie, D. P. (1985) Practical error estimation in numerical integration.Journal of Computational and Applied Mathematics,12–13, 425–431.Google Scholar
  8. Lyness, J. N. (1965) Symmetric integration rules for hypercubes.Math. Comp.,19, 260–276, 394–407, 625–637.Google Scholar
  9. Mantel, P. and Rabinowitz, P. (1977) The application of integer programming to the computation of fully symmetric integration formulas in two and three dimensions.SIAM J. Numer. Anal. 14, 425–431.Google Scholar
  10. McNamee, J. and Stenger, F. (1967) Construction of fully symmetric numerical integration formulas.Numer. Math. 10, 327–344.Google Scholar
  11. Monegato, G. (1978) Some remarks on the construction of extended Gaussian quadrature rules.Mathematics of Computation,32, 141, 247–252.Google Scholar
  12. Naylor, J. C. (1982) Some numerical aspects of Bayesian inference. Unpublished PhD thesis, University of Nottingham.Google Scholar
  13. Naylor, J. C. (1987) Bayesian alternatives tot-tests.The Statistician,36, 241–246.Google Scholar
  14. Naylor, J. C. and Shaw, J. E. H. (1985) Bayes Four—User guide. Nottingham Statistics Group, Department of Mathematics, University of Nottingham.Google Scholar
  15. Naylor, J. C. and Smith, A. F. M. (1982) Applications of a method for the efficient computation of posterior distributions.Appl. Statist. 31, 3, 214–225.Google Scholar
  16. Naylor, J. C. and Smith, A. F. M. (1983) A contamination model in clinical chemistry.The Statistician,32, 214–225.Google Scholar
  17. Naylor, J. C. and Smith, A. F. M. (1988a) An archaeological inference problem.J. Amer. Statist. Assoc.,83, 403, 588–595.Google Scholar
  18. Naylor, J. C. and Smith, A. F. M. (1988b) Econometric illustrations of novel numerical integration strategies for Bayesian inference.Journal of Economietrics,38, 103–126.Google Scholar
  19. Patterson, T. N. L. (1968a) The optimum addition of points to Quadrature formulae.Math. Comp.,22, 847–856.Google Scholar
  20. Patterson, T. N. L. (1968b) On some Gauss and Lobatto based Quadrature formulae.Math. Comp.,22, 877–881.Google Scholar
  21. Patterson, T. (1973) Algorithm 468: Algorithm for numerical integration over a finite interval.Communications of the ACM,16, 694–699.Google Scholar
  22. Piessens, R. (1973) An algorithm for automatic integration.Angewandte Informatik,9, 399–401.Google Scholar
  23. Piessens, R. and Randers, M. B. (1974) A note on the optimal addition of abscissas to quadrature formulas of Gauss and Lobatto type.Mathematics and Computation,28, 125, 135–139.Google Scholar
  24. Rabinowitz, P., Kautsky, J., Elhay, S. and Butcher, J. C. (1987) On sequences of imbedded integration rules, inNumerical Integration: Recent Developments, P. Keast and G. Fairweather (eds), NATO ASI series, Reidel, Dordrecht, pp. 113–139.Google Scholar
  25. Rabinowitz, P. and Richer, N. (1969) Perfectly symmetric two-dimensional integration formulas with minimal number of points.Math. Comp.,23, 765–780.Google Scholar
  26. Shaw, J. E. H. (1987) Aspects of numerical integration and summarisation. Research report 103, Dept. of Statistics, University of Warwick.Google Scholar
  27. Shaw, J. E. H. (1988). A quasirandom approach to integration in Bayesian statistics.Ann. Statist.,16, 895–914.Google Scholar
  28. Smith, R. L. and Naylor, J. (1987) A comparison of maximum likelihood and Bayesian estimators for the three parameter Weibull distribution.Appl. Statist. 36, 358–369.Google Scholar
  29. Smith, A. F. M., Skene, A. M., Shaw, J. E. H.et al. (1985) The implementation of the Bayesian paradigm.Commun. Statist. Theor. Meth.,14, 1079–1102.Google Scholar
  30. Smith, A. F. M., Skene, A. M., Shaw, J. E. H. and Naylor, J. C. (1987) Progress with numerical and graphical methods for practical Bayesian statistics.The Statistician,36, 75–82.Google Scholar
  31. Stroud, A. H. (1971)Approximate Calculation of Multiple Integrals, Englewood Cliffs, NJ, Prentice Hall.Google Scholar

Copyright information

© Chapman and Hall Ltd 1991

Authors and Affiliations

  • Petros Dellaportas
    • 1
  • David Wright
    • 2
  1. 1.Department of MathematicsUniversity of NottinghamNottingham
  2. 2.Department of Mathematics and StatisticsPolytechnic South WestPlymouth

Personalised recommendations