Toward the history of the St. Petersburg school of probability and statistics. IV: Characterization of distributions and limit theorems in statistics (Q2289243)

The authors present the following: ``This is the fourth article in a series of surveys devoted to the scientific achievements of the Leningrad-St. Petersburg School of Probability and Statistics from 1947 to 2017. It is devoted to studies on the characterization of distributions, limit theorems for kernel density estimators, and asymptotic efficiency of statistical tests. The characterization results are related to the independence and equidistribution of linear forms of sample values, as well as to regression relations, admissibility, and optimality of statistical estimators. When calculating the Bahadur asymptotic efficiency, particular attention is paid to the logarithmic asymptotics of large deviation probabilities of test statistics under the null hypothesis. Constructing new goodness-of-fit and symmetry tests based on characterizations is considered, and their asymptotic behavior is analyzed. Conditions of local asymptotic optimality of various nonparametric statistical tests are studied.'' In the first section, the main attention is given to characterization problems as a branch of analytical statistics. The authors note that results obtained in this field by Yu.V. Linnik's school are presented in this study. The first section consists of the following items: -- Independence of linear forms of independent random variables -- Independence of nonlinear functions of independent random variables -- Constancy of regression -- Equidistribution of linear forms -- Admissibility and optimality of estimators -- Miscellanea. In the second section ``Construction and asymptotic comparison of statistical tests'', the authors note that the problem of a reasonable choice of a statistical test from the several tests at the statistician's disposal, is one of the classical and most well-known problems of mathematical statistics. The authors consider the following items: -- Asymptotic efficiency of tests -- Large deviations of test statistics -- Large deviations of \(U\)-statistics and von Mises functionals -- Calculation of the Bahadur efficiency -- Local asymptotic optimality of tests and characterization of distributions -- Goodness-of-fit and symmetry tests based on characterizations. The third section of this paper is on estimates for the rate of approximation in the central limit theorem for the \(L_1\)-norm of kernel density estimators. A conclusion on the contribution of the Leningrad-St.Petersburg School of Probability to the characterization of distributions by various properties of statistics, is given, and some open problems are noted in sections.