DESCRIPTION OF THE PROGRAM ESOD-3 FOR CALCULATING CONSTRAINED MAXIMUM LIKELIHOOD ESTIMATES OF N STOCHASTICALLY ORDERED DISTRIBUTIONS,

Abstract

A detailed description is given of the FORTRAN IV computer program for calculating constrained maximum likelihood estimates of N stochastically ordered distributions. It was shown in AD-672 954 that the reformulation of the statistical problem as a concave program, and the reduction of the resulting constrained problem to a sequence of essentially unconstrained subproblems, yields satisfying numerical results. The ESOD-3 program of this reduction strategy, which incorporates an already existing nonlinear programming routine, reduces the execution time and improves the accuracy of the problem by orders of magnitude. Since formulas are already known for the step function maximum likelihood estimates for the case where N equals 2, the program is concerned exclusively with the case where N is equal to or greater than 3. (Author)

Document Details

Document Type
Technical Report
Publication Date
Jul 01, 1968
Accession Number
AD0672956

Entities

People

  • S. P. Azen

Organizations

  • RAND Corporation

Tags

DTIC Thesaurus Topics

  • Accuracy
  • Application Software
  • Computer Programming
  • Computer Programs
  • Computers
  • Digital Information
  • Mathematics
  • Nonlinear Programming
  • Sequences
  • Step Functions

Fields of Study

  • Mathematics

Readers

  • Computer Science.
  • Operations Research
  • Regression Analysis.