Summary
We establish a connection between admissible simultaneous estimation and recurrence of reversible Markov chains on ℤ p+ . Specifically, to each generalized Bayes estimator of the mean of a vector of p independent Poisson variables for a weighted quadratic loss, we associate a variational problem and a reversible birth and death chain on ℤ p+ . The variational problem is closely related to the Dirichlet principle for reversible chains studied recently by Griffeath, Liggett and Lyons. Under side conditions, admissibility of the estimator is equivalent to zero infimal energy in the variational problem and to recurrence of the Markov chain. This yields analytic and probabilistic criteria for inadmissibility which are applied to establish a broad class of results and previous conjectures.
Article PDF
Similar content being viewed by others
References
Abramowitz, M., Stegun, I.A.: Handbook of Mathematical Functions. New York: Dover 1965
Berger, J.: Improving on inadmissible estimators in continuous exponential families with applications to simultaneous estimation of gamma scale parameters. Ann. Statist. 8, 545–571 (1980)
Berger, J.: Multivariate Estimation — A Synthesis of Bayesian and Frequentist Approaches. Forthcoming SIAM monograph (1985)
Birnbaum, A.: Characterizations of complete classes of tests of some multiparametric hypotheses, with applications to likelihood ratio tests. Ann. Math. Statist. 26, 21–26 (1955)
Brown, L.D.: Admissible estimators, recurrent diffusions and insoluble boundary value problems. Ann. Math. Statist. 42, 855–903 (1971). Correction, Ann. Statist. 1, 594–596 (1973)
Brown, L.D.: Counterexample — An inadmissible estimator which is generalized Bayes for a prior with “light” tails. J. Multivariate Anal. 9, 332–336 (1979a)
Brown, L.D.: A heuristic method for determining admissibility of estimators — with applications. Ann. Statist. 7, 960–994 (1979b)
Brown, L.D., Farrell, R.H.: Complete class theorems for estimation of multivariate Poisson means and related problems. Ann. Statist. 13, 706–726 (1985a)
Brown, L.D., Farrell, R.H.: All admissible linear estimators of the multivariate Poisson mean vector. Ann. Statist. 13, 282–294 (1985b)
Brown, L.D., Farrell, R.H.: Proof of a necessary and sufficient condition for admissibility in discrete multivariate problems. J. Multivarite Anal. To appear (1985c)
Brown, L.D., Hwang, J.T.: A unified admissibility proof. Proc. 3rd Purdue Symposium Statistical Decision Theory 1, 205–231 (1982)
Cohen, A., Sackrowitz, H.: Estimation of the last mean of a monotone sequence. Ann. Math. Statist. 41, 2021–2034 (1970)
Clevenson, M., Zidek, J.: Simultaneous estimation of the mean of independent Poisson laws. J. Am. Stat. Assoc. 70, 698–705 (1975)
Efron, B., Morris, C.: Stein's estimation rule and its competitors — An empirical Bayes approach. J. Am. Stat. Assoc. 68, 117–130 (1973)
Farrell, R.H.: Towards a theory of generalized Bayes tests. Ann. Math. Statist. 39, 1–22 (1968)
Freedman, D.: Markov Chains. San Francisco: Holden-Day 1971
Fukushima, M.: Dirichlet forms and Markov processes. Amsterdam-New York: North-Holland 1980
Ghosh, M., Hwang, J.T., Tsui, K.W.: Construction of improved estimators in multiparameter estimation for discrete exponential families (with discussion). Ann. Statist. 11, 351–367 (1983)
Ghosh, M., Parsian, A.: Bayes minimax estimation of multiple Poisson parameters. J. Multivarite Anal. 11, 280–288 (1981)
Griffeath, D., Liggett, T.M.: Critical phenomena for Spitzer's reversible nearest particle systems. Ann. Probab. 10, 881–895 (1982)
Guttman, S.: Stein's paradox is impossible in a finite sample space. Ann. Statist. 10, 1017–1020 (1982)
Guttman, S.: Decisions immune to Stein's effect. Sankhya, To appear (1983)
Hwang, J.T.: Improving upon standard estimators in discrete exponential families with applications to Poisson and negative binomial cases. Ann. Statist. 10, 857–867 (1982)
Hwang, J.T.: Semi tail upper bounds on the class of admissible estimators with applications to Poisson and negative binomial families. Ann. Statist. 10, 1137–1147 (1982)
Ichihara, K.: Some global properties of symmetric diffusion processes. Publ. Res. Inst. Math. Sci., Kyoto Univ., 14, 441–486 (1978)
James, W., Stein, C.: Estimation with quadratic loss. Proc. Fourth Berkeley Sympos. Univ. Calif. Math. Statist. Probab. 1, 361–379 (1960)
Johnstone, I.M.: Admissibility, difference equations and recurrence in estimating a Poisson mean. Ann. Statist. 12, 1173–1198 (1984)
Johnstone, I.M.: Admissibility from recurrence via Poincaré inequalities in the Gaussian case. Manuscript notes (1983)
Johnstone, I.M., Lalley, S.P.: On independent statistical decision problems and products of diffusions. Zeitschrift für Wahrscheinlichkeistheor. Verw. Geb. 68, 29–47 (1984)
Johnstone, I.M., Shahshahani, M.: A Poincaré-type inequality for solutions of linear elliptic equations with statistical applications. M.S.R.I. technical report (1983)
Karlin, S.: Admissibility for estimation with quadratic loss. Ann. Math. Statist. 29, 406–436 (1958)
Lebedev, N.N.: Special Functions and Their Applications. New York: Dover 1972
LeCam, L.: An extension of Wald's theory of statistical decision functions. Ann. Math. Statist. 26, 69–81 (1955)
Lyons, T.: A simple criterion for transience of a reversible Markov chain. Ann. Probab. 11, 393–402 (1983)
Morris, C.N.: Parametric empirical Bayes inference: Theory and Applications (with discussion). J. Am. Statist. Assoc., 47–65 (1983)
Peng, J.C.M.: Simultaneous estimation of the parameters of independent Poisson distributions. Technical Report 78, Dept. of Statistics, Stanford University (1975)
Srinivasan, C.: Admissible generalised Bayes estimates and exterior boundary value problems. Sankhya 43, Ser. A, 1–25 (1981)
Stein, C.: A necessary and sufficient condition for admissibility. Ann. Math. Statist. 26, 518–522 (1955)
Stein, C.: The admissibility of Hotelling's T 2-test. Ann. Math. Statist. 27, 616–623 (1956a)
Stein, C.: Inadmissibility of the usual estimator for the mean of a multivariate normal distribution. Proc. Third Berkeley Sympos. Univ. Calif. Math. Statist. Probability 1, 197–206 (1956b)
Stein, C.: Approximation of improper prior measures by prior probability measures. In Bernouilli, Bayes, Laplace Anniversary Volume, pp. 217–241. Berlin Heidelberg New York: Springer 1965
Author information
Authors and Affiliations
Additional information
Research supported by an Australian National University Scholarship and A.D. White Fellowship at Cornell University and by NSF at Mathematical Sciences Research Institute, Berkeley and at Stanford
Rights and permissions
About this article
Cite this article
Johnstone, I. Admissible estimation, Dirichlet principles and recurrence of birth-death chains on ℤ p+ . Probab. Th. Rel. Fields 71, 231–269 (1986). https://doi.org/10.1007/BF00332311
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF00332311