Kybernetika 50 no. 6, 883-895, 2014

An efficient estimator for Gibbs random fields

Martin JanžuraDOI: 10.14736/kyb-2014-6-0883

Abstract:

An efficient estimator for the expectation $\int f \d P$ is constructed, where $P$ is a Gibbs random field, and $f$ is a local statistic, i. e. a functional depending on a finite number of coordinates. The estimator coincides with the empirical estimator under the conditions stated in Greenwood and Wefelmeyer \cite{greenwood_wefelmeyer_1999}, and covers the known special cases, namely the von Mises statistic for the i.i.d. underlying fields and the case of one-dimensional Markov chains.

Keywords:

Gibbs random field, efficient estimator, empirical estimator

Classification:

62F12, 62M40

References:

  1. P. J. Bickel, C. A. J. Klaassen, Y. Ritov and J. A. Wellner: Efficient and Adaptive Estimation for Semiparametric Models. Johns Hopkins University Press, Baltimore 1993.   CrossRef
  2. R. L. Dobrushin: Prescribing a system of random variables by conditional distributions. Theor. Probab. Appl. 15 (1970), 458-486.   CrossRef
  3. R. L. Dobrushin and B. S. Nahapetian: Strong convexity of the pressure for lattice systems of classical physics (in Russian). Teoret. Mat. Fiz. 20 (1974), 223-234.   CrossRef
  4. H. O. Georgii: Gibbs Measures and Phase Transitions. De Gruyter Studies in Mathematics 9, De Gruyter, Berlin 1988.   CrossRef
  5. P. E. Greenwood and W. Wefelmeyer: Efficiency of empirical estimators for Markov Chains. Ann. Statist. 23 (1995), 132-143.   CrossRef
  6. P. E. Greenwood and W. Wefelmeyer: Characterizing efficient empirical estimators for local interaction Gibbs fields. Stat. Inference Stoch. Process. 2 (1999), 119-134.   CrossRef
  7. K. Gross: Absence of second-order phase transitions in the Dobrushin uniqueness region. J. Statist. Phys. 25 (1981), 57-72.   CrossRef
  8. J. Hájek: A characterization of limiting distributions of regular estimates. Wahrsch. Verw. Gebiete 14 (1970), 323-330.   CrossRef
  9. M. Janžura: Statistical analysis of Gibbs random fields. In: Trans. 10th Prague Conference on Inform. Theory, Stat. Dec. Functions, Random Processes, Prague 1984, pp. 429-438.   CrossRef
  10. M. Janžura: Local asymptotic normality for Gibbs random fields. In: Proc. Fourth Prague Symposium on Asymptotic Statistics (P. Mandl and M. Hušková, eds.), Charles University, Prague 1989, pp. 275-284.   CrossRef
  11. M. Janžura: Asymptotic behaviour of the error probabilities in the pseudo-likelihood ratio test for Gibbs-Markov distributions. In: Prof. Asymptotic Statistics (P. Mandl and M. Hušková, eds.), Physica-Verlag, Heidelberg 1994, pp. 285-296.   CrossRef
  12. M. Janžura: Asymptotic results in parameter estimation for Gibbs random fields. Kybernetika 33 (1997), 2, 133-159.   CrossRef
  13. M. Janžura: On the concept of the asymptotic Rényi distances for random fields. Kybernetika 35 (1999), 3, 353-366.   CrossRef
  14. H. Künsch: Decay of correlations under Dobrushin's uniqueness condition and its applications. Comm. Math. Phys. 84 (1982), 207-222.   CrossRef
  15. L. Younes: Parameter inference for imperfectly observed Gibbsian fields. Probab. Theory Rel. Fields 82 (1989), 625-645.   CrossRef