B-Splines and Linear Combinations of Uniform Order Statistics.

Abstract

It is shown in this document that the density of the sum of n independent random variables uniformly distributed on unequal intervals is given by a linear combination of n! B-splines with constant coefficients. Another useful representation of the same density is given using de Boor's definition of the discrete B-spline. It is also shown that the B-spline is the density of a linear combination of order statistics from the uniform distribution on (0,1). This interpretation of the B-spline allows one to establish easily validity of recurrence relations for densities and moments of linear combinations of order statistics as well as to state limit theorems both for B-splines and linear combinations. Examples given illustrate how asymptotic results for B-splines may be applied to linear combinations of order statistics. Using limit theorems of probability theory two examples of Curry and Schoenberg (1966) for B-splines are derived even in somewhat greater generality.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
May 01, 1985
Accession Number
ADA158151

Entities

People

  • A. G. Ignatov
  • V. K. Kaishev

Organizations

  • University of Wisconsin–Madison

Tags

DTIC Thesaurus Topics

  • Abstracts
  • Asymptotic Normality
  • Classification
  • Coefficients
  • Contracts
  • Distribution Functions
  • Mathematics
  • New York
  • Numbers
  • Order Statistics
  • Probability
  • Random Variables
  • Statistics
  • Stochastic Processes
  • United States
  • Universities
  • Wisconsin

Fields of Study

  • Mathematics

Readers

  • Approximation Theory.
  • Mathematical Modeling and Probability Theory.