Abstract
In the Bayesian Statistics, the use of conjugate prior distributions allows to obtain exact posterior quantities, avoiding the use of approximate methods. However, there are only a few conjugate structures available, which limits the use in practical problems. In this work we propose a general procedure to obtain the posterior distribution in an exact form in non-conjugate location parameter models with known (possibly, heteroskedastic) variance. The theory is based on special functions, specifically the H-function, which embraces most of the probability distributions. We express explicitly in a computable form all the Bayesian tools for posterior inferences, that is the posterior and the posterior predictive distributions and their moments, as well as the cumulative posterior distribution. As an illustration we consider a problem in astronomy of estimating the distance to the Large Magellanic Cloud, the largest galaxy now orbiting the Milky Way Galaxy.
Similar content being viewed by others
References
Andrade JAA, Rathie PN (2015) On exact posterior distributions using H-functions. J Comput Appl Math 290:459–475
Clementini G, Gratton R, Bragaglia A, Carretta E, Fabrizio LD, Maio M (2003) Distance to the Large Magellanic Cloud: the RR Lyrae stars. Astron J 125:1309–1329
Goyal SP (1975) The H-function of two variables. Kyungpook Math J 15:117–131
Goyal SP, Garg RS (1981) Certain results for multivariate H-function involving Jacobi polynomials. Kyungpook Math J 21:123–134
Mathai AM, Saxena RK, Haubold HJ (2010) The H-function: theory and applications. Springer, New York
Mittal PK, Gupta KC (1972) An integral involving generalized function of two variables. Proc Indian Acad Sci A 78:117–123
Raiffa H, Schlaifer R (1961) Applied statistical decision theory. Division of Research, Graduate School of Business Administration, Harvard University, Cambridge
Srivastava HM, Gupta KC, Goyal SP (1982) The H-function of one and two variables with applications. South Asian Publishers, New Delhi
Weatherill GB (1961) Bayesian sequential analysis. Biometrika 48:281–292
Acknowledgements
The second author is thankful to CAPES (Brazil) for supporting his Senior National Visiting Professorship.
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by Josselin Garnier.
Rights and permissions
About this article
Cite this article
Andrade, J.A.A., Rathie, P.N. & Farias, R.B.A. Exact Bayesian computation using H-functions. Comp. Appl. Math. 37, 2277–2293 (2018). https://doi.org/10.1007/s40314-017-0451-z
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s40314-017-0451-z
Keywords
- Non-conjugate prior distributions
- Location parameter
- Special functions
- Heteroskedasticity
- Exact posterior distribution