Strong law of large numbers for a class of U-statistics /

This dissertation studies the strong law of large numbers for U-statistics with kernel h : Rm-- R, h(xl,... , xm) = x1.....xm, and normalizing sequence - yn. satisfying some regularity conditions, in several instances, namely, when the indices of summation are restricted to a square, to a strip or t...

Full description

Bibliographic Details
Main Author: Gadidov, Anda Hedwiga, 1958-
Format: Thesis Book
Language:English
Published: [Place of publication not identified] : [publisher not identified] ; 1996.
Subjects:
Online Access:http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=739363491&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD
Description
Summary:This dissertation studies the strong law of large numbers for U-statistics with kernel h : Rm-- R, h(xl,... , xm) = x1.....xm, and normalizing sequence - yn. satisfying some regularity conditions, in several instances, namely, when the indices of summation are restricted to a square, to a strip or to a sector in Zm+. For m = 2, results analogous to the ones in Cuzick, Gine and Zinn (1995) are obtained for 2-sample U- statistics, and under some regularity assumptions on the distributions of the random variables, these results are generalized to m > 2. Convergence in a strip of Z2+ is characterized for both one-sample and 2-sample U- statistics, and it is shown to be equivalent to convergence in the square. Convergence in a sector of Zm of the normalized maxima of products is proved to be equivalent to convergence in the square. Similar results are obtained for one-sample and 2-sample U- statistics in a sector Z2+. Finally, a complete characterization of those powers of order statistics for which there exists a constant c e R+, such that whenever Xi, i = 1, . . . , k are independent and identically distributed non-negative random variables, and X(1) <- (2) <... < Xlk) are the order statistics, we have EXr1......Xrk<=cEX91) is given.
Item Description:Vita.
"Major Subject: Mathematics".
Physical Description:v,57 leaves ; 28 cm.
Issued also on microfiche from University Microfilms Inc.
Bibliography:Includes bibliographical references: pages 55-56.