Trivial intersection of \(\sigma \)-fields and Gibbs sampling. (English) Zbl 1159.60007

Berti, Patrizia; Pratelli, Luca; Rigo, Pietro

Ann. Probab. 36, No. 6, 2215-2234 (2008).

In this paper the attention focuses on necessary and sufficient conditions for \(\bar\mathcal A\bigcap\bar\mathcal B=\overline{\mathcal A\bigcap\mathcal B}\), where \(\mathcal A, \mathcal B\subset\mathcal F\) are sub \(\sigma\)-fields and \((\Omega,\mathcal F,\mathcal P)\) is a probability space. These conditions are then applied to the (two-component) Gibbs sampler. Suppose \(X\) and \(Y\) are the coordinate projections on \((\Omega, \mathcal F)=(\mathcal X\times\mathcal Y,\mathcal U\otimes\mathcal V)\) where \((\mathcal X,\mathcal U)\) and \((\mathcal Y,\mathcal V)\) are measurable spaces. Let \((X_n, Y_n)_{n\geq0}\) be the Gibbs chain for \(P\). Then, the SLLN holds for \((X_n, Y_n)\) if and only if \(\overline{\sigma(X)}\bigcap\overline{\sigma(Y)}=\mathcal N\), or equivalently if and only if \(P(X\in U)P(Y\in V)=0\) whenever \(U\in\mathcal U, V\in\mathcal V\) and \(P(U\times V)=P(U^c\times V^c)=0\). The latter condition is also equivalent to ergodicity of \((X_n, Y_n)\), on a certain subset \(S_0\subset\Omega\), in case \(\mathcal F=\mathcal U\otimes\mathcal V\) is countably generated and \(P\) absolutely continuous with respect to a product measure.

Reviewer: Nicko G. Gamkrelidze (Moskva)

MSC:

60A05	Axioms; other general questions in probability
60A10	Probabilistic measure theory
60J22	Computational methods in Markov chains
65C05	Monte Carlo methods

Keywords:

ergodicity; Gibbs sampler; iterated conditional expectation; Markov chain; strong law of large numbers

References:

[1]	Berti, P., Pratelli, L. and Rigo, P. (2007). Skorohod representation on a given probability space. Probab. Theory Related Fields 137 277-288. · Zbl 1106.60003
[2]	Burkholder, D. L. and Chow, Y. S. (1961). Iterates of conditional expectation operators. Proc. Amer. Math. Soc. 12 490-495. JSTOR: · Zbl 0106.33201
[3]	Burkholder, D. L. (1961). Sufficiency in the undominated case. Ann. Math. Statist. 32 1191-1200. · Zbl 0221.62003
[4]	Burkholder, D. L. (1962). Successive conditional expectations of an integrable function. Ann. Math. Statist. 33 887-893. · Zbl 0128.12602
[5]	Diaconis, P., Khare, K. and Saloff-Coste, L. (2008). Gibbs sampling, exponential families and orthogonal polynomials. Statist. Sci. 23 151-178. · Zbl 1327.62058
[6]	Diaconis, P., Freedman, D., Khare, K. and Saloff-Coste, L. (2007). Stochastic alternating projections. Preprint, Dept. Statistics, Stanford Univ. Currently available at http://www-stat.stanford.edu/ cgates/PERSI/papers/altproject-2.pdf. · Zbl 1268.60098
[7]	Jones, G. L. and Hobert, J. P. (2001). Honest exploration of intractable probability distributions via Markov chain Monte Carlo. Statist. Sci. 16 312-334. · Zbl 1127.60309
[8]	Liu, J. S. (2001). Monte Carlo Strategies in Scientific Computing. Springer Series in Statistics . Springer, New York. · Zbl 0991.65001
[9]	Meyn, S. P. and Tweedie, R. L. (1996). Markov Chains and Stochastic Stability . Springer, New York. · Zbl 0925.60001
[10]	Tierney, L. (1994). Markov chains for exploring posterior distributions. Ann. Statist. 22 1701-1762. With discussion and a rejoinder by the author. · Zbl 0829.62080