Accession Number : AD0736805

Title :   A Note on the Separation Theorems for Linear Stochastic Systems and Quadratic Costs,

Corporate Author : WASHINGTON UNIV ST LOUIS MO CONTROL SYSTEMS SCIENCE AND ENGINEERING

Personal Author(s) : Rhodes,Ian B.

Report Date : 04 AUG 1971

Pagination or Media Count : 12

Abstract : Direct proofs are presented for the two separation theorems involving linear systems and quadratic cost functionals. The first is for the case when all random vectors are jointly Gaussian and the cost functional is minimized over all control laws whose value at each time is restricted only to being a finite-variance measurable function of the output history of the system up to that time. The second does not require that all random vectors be Gaussian but restricts attention to control laws that at each time are linear functions of the past output history. (Author)

Descriptors :   (*ADAPTIVE CONTROL SYSTEMS, MATHEMATICAL MODELS), STOCHASTIC PROCESSES, LINEAR SYSTEMS, DYNAMIC PROGRAMMING, THEOREMS

Subject Categories : Statistics and Probability

Distribution Statement : APPROVED FOR PUBLIC RELEASE