The Annals of Statistics

Laplace approximations for hypergeometric functions with matrix argument

Roland W. Butler and Andrew T. A. Wood

Full-text: Open access

Abstract

In this paper we present Laplace approximations for two functions of matrix argument: the Type I confluent hypergeometric function and the Gauss hypergeometric function. Both of these functions play an important role in distribution theory in multivariate analysis, but from a practical point of view they have proved challenging, and they have acquired a reputation for being difficult to approximate. Appealing features of the approximations we present are: (i) they are fully explicit (and simple to evaluate in practice); and (ii) typically, they have excellent numerical accuracy. The excellent numerical accuracy is demonstrated in the calculation of noncentral moments of Wilks' $\Lambda$ and the likelihood ratio statistic for testing block independence, and in the calculation of the CDF of the noncentral distribution of Wilks' $\Lambda$ via a sequential saddlepoint approximation. Relative error properties of these approximations are also studied, and it is noted that the approximations have uniformly bounded relative errors in important cases.

Article information

Source
Ann. Statist., Volume 30, Number 4 (2002), 1155-1177.

Dates
First available in Project Euclid: 10 September 2002

Permanent link to this document
https://projecteuclid.org/euclid.aos/1031689021

Digital Object Identifier
doi:10.1214/aos/1031689021

Mathematical Reviews number (MathSciNet)
MR1926172

Zentralblatt MATH identifier
1029.62047

Subjects
Primary: 62H10: Distribution of statistics
Secondary: 62E17: Approximations to distributions (nonasymptotic)

Keywords
Confluent hypergeometric function Gauss hypergeometric function Laplace approximation likelihood ratio test matrix-argument hypergeometric function saddlepoint approximation sequential saddlepoint approximation Wilks' $\Lambda$

Citation

Butler, Roland W.; Wood, Andrew T. A. Laplace approximations for hypergeometric functions with matrix argument. Ann. Statist. 30 (2002), no. 4, 1155--1177. doi:10.1214/aos/1031689021. https://projecteuclid.org/euclid.aos/1031689021


Export citation

References

  • ABRAMOWITZ, M. and STEGUN, I. A. (1972). Handbook of Mathematical Functions, 9th ed. Dover, New York.
  • BAGCHI, P. and KADANE, J. B. (1991). Laplace approximations to posterior moments and marginal distributions on circles, spheres, and cy linders. Canad. J. Statist. 19 67-77.
  • BARNDORFF-NIELSEN, O. E. and WOOD, A. T. A. (1998). On large deviations and choice of ancillary for p and r. Bernoulli 4 35-63.
  • BINGHAM, C., CHANG, T. and RICHARDS, D. (1992). Approximating the matrix Fisher and Bingham distributions: applications to spherical regression and Procrustes analysis. J. Multivariate Anal. 41 314-337.
  • BLEISTEIN, N. and HANDELSMAN, R. A. (1975). Asy mptotic Expansions of Integrals. Dover, New York.
  • BOOTH, J. G., BUTLER, R. W., HUZURBAZAR, S. and WOOD, A. T. A. (1995). Saddlepoint approximations for p-values of some tests of covariance matrices. J. Statist. Comput. Simulation 53 165-180.
  • BUTLER, R. W., HUZURBAZAR, S. and BOOTH, J. G. (1992). Saddlepoint approximations for the Bartlett-Nanda-Pillai trace statistic in multivariate analysis. Biometrika 79 705-715.
  • BUTLER, R. W. and WOOD, A. T. A. (2000). Laplace approximations for hy pergeometric functions of matrix argument. Technical Report 00-05, Div. Statistics, Univ. Nottingham. Available at www.maths.nottingham.ac.uk/statsdiv/research/reports.html.
  • CHIKUSE, Y. (1998). Density estimation on the Stiefel manifold. J. Multivariate Anal. 66 188-206.
  • CONSTANTINE, A. G. (1963). Some noncentral distribution problems in multivariate analysis. Ann. Math. Statist. 34 1270-1285.
  • CONSTANTINE, A. G. and MUIRHEAD, R. J. (1976). Asy mptotic expansions for distributions of latent roots in multivariate analysis. J. Multivariate Anal. 6 369-391.
  • EFSTATHIOU, M., GUTIÉRREZ-PEÑA, E. and SMITH, A. F. M. (1998). Laplace approximations for natural exponential families with cuts. Scand. J. Statist. 25 77-92.
  • FISHER, R. A. (1990). Statistical Methods, Experimental Design, and Scientific Inference. Oxford Univ. Press.
  • FRASER, D. A. S., REID, N. and WONG, A. (1991). Exponential linear models: a two-pass procedure for saddlepoint approximation. J. Roy. Statist. Soc. Ser. B 53 483-492.
  • GLy NN, W. J. and MUIRHEAD, R. J. (1978). Inference in canonical correlation analysis. J. Multivariate Anal. 8 468-478.
  • GUPTA, R. D. and RICHARDS, D. ST. P. (1985). Hy pergeometric functions of scalar matrix argument are expressible in terms of classical hy pergeometric functions. SIAM J. Math. Anal. 16 852-858.
  • HERZ, C. S. (1955). Bessel functions of matrix argument. Ann. Math. 61 474-523.
  • JAMES, A. T. (1964). Distributions of matrix variates and latent roots derived from normal samples. Ann. Math. Statist. 35 475-501.
  • JENSEN, J. L. (1995). Saddlepoint Approximations. Oxford Univ. Press.
  • JOHNSON, N. L. and KOTZ, S. (1972). Distributions in Statistics: Continuous Multivariate Distributions. Wiley, New York.
  • LUGANNANI, R. and RICE, S. O. (1980). Saddlepoint approximations for the distribution of the sum of independent random variables. Adv. Appl. Probab. 12 475-490.
  • MAGNUS, J. R. and NEUDECKER, H. (1988). Matrix Differential Calculus with Applications in Statistics and Econometrics. Wiley, New York.
  • MATHAI, A. M. (1993). A Handbook of Generalized Special Functions for Statistical and physical Sciences. Clarendon Press, Oxford.
  • MUIRHEAD, R. J. (1975). Expressions for some hy pergeometric functions of matrix argument with applications. J. Multivariate Anal. 5 283-293.
  • MUIRHEAD, R. J. (1978). Latent roots and matrix variates: a review of some asy mptotic results. Ann. Statist. 6 5-33.
  • MUIRHEAD, R. J. (1982). Aspects of Multivariate Statistical Theory. Wiley, New York.
  • OLVER, F. W. J. (1991). Uniform, exponentially improved, asy mptotic expansions for the confluent hy pergeometric function and other integral transforms. SIAM J. Math. Anal. 22 1475- 1489.
  • OLVER, F. W. J. (1993). Exponentially-improved asy mptotic solutions of ordinary differential equations. I. The confluent hy pergeometric function. SIAM J. Math. Anal. 24 756-767.
  • PARIS, R. B. (1992). Smoothing of the Stokes phenomenon using Mellin-Barnes integrals. Asy mptotic methods in analysis and combinatorics. J. Comput. Appl. Math. 41 117-133.
  • PHILLIPS, P. C. B. (1983). Exact small sample theory in the simultaneous equations model. In Handbook of Econometrics 1 (Z. Griliches, and M. D. Intriligator, eds.) Chapter 8. North-Holland, Amsterdam.
  • ROY, S. N. (1939). p-statistics, or some generalizations in analysis of variance appropriate to multivariate problems. Sankhy¯a 4 381-396.
  • SRIVASTAVA, M. S. and CARTER, E. M. (1980). Asy mptotic expansions for hy pergeometric functions. In Multivariate Analy sis V (P. Krishnaiah, ed.) 337-347. North-Holland, Amsterdam.
  • SUGIURA, N. (1969). Asy mptotic non-null distributions of the likelihood ratio criteria for covariance matrix under local alternatives. Mimeo Ser. 609, Inst. Statistics, Univ. North Carolina, Chapel Hill.
  • SUGIURA, N. (1973). Further asy mptotic formulas for the non-null distributions of three statistics for multivariate linear hy pothesis. Ann. Inst. Statist. Math. 25 153-163.
  • SUGIURA, N. and FUJIKOSHI, Y. (1969). Asy mptotic expansions of the non-null distributions of the likelihood ratio criteria for multivariate linear hy pothesis and independence. Ann. Math. Statist. 40 942-952.
  • SUGIy AMA, T. (1967). On the distribution of the largest latent root of the covariance matrix. Ann. Math. Statist. 38 1148-1151.
  • FORT COLLINS, COLORADO 80523 E-MAIL: walrus@stat.colostate.edu SCHOOL OF MATHEMATICAL SCIENCES UNIVERSITY OF NOTTINGHAM NOTTINGHAM NG7 2RD UNITED KINGDOM E-MAIL: atw@maths.nott.ac.uk