Stochastic Approximation Methods for Constrained and Unconstrained Systems

by H.J. Kushner, D.S. Clark

Overview

The book deals with a powerful and convenient approach to a great variety of types of problems of the recursive monte-carlo or shastic approximation type. Such recu- sive algorithms occur frequently in shastic and adaptive control and optimization theory and in statistical esti- tion theory. Typically, a sequence {X } of estimates of a n parameter is obtained by means of some recursive statistical th st procedure. The n estimate is some function of the n_l estimate and of some new observational data, and the aim is to study the convergence, rate of convergence, and the pa- metric dependence and other qualitative properties of the - gorithms. In this sense, the theory is a statistical version of recursive numerical analysis. The approach taken involves the use of relatively simple compactness methods. Most standard results for Kiefer-Wolfowitz and Robbins-Monro like methods are extended considerably. Constrained and unconstrained problems are treated, as is the rate of convergence problem. While the basic method is rather simple, it can be elaborated to allow a broad and deep coverage of shastic approximation like problems. The approach, relating algorithm behavior to qualitative properties of deterministic or shastic differential equations, has advantages in algorithm conceptualiza­ tion and design. It is often possible to obtain an intuitive understanding of algorithm behavior or qualitative dependence upon parameters, etc., without getting involved in a great deal of deta~l.

Product Details

ISBN-13:	9780387903415
Publisher:	Springer New York
Publication date:	08/03/1978
Series:	Applied Mathematical Sciences , #26
Edition description:	Softcover reprint of the original 1st ed. 1978
Pages:	263
Product dimensions:	6.10(w) x 9.25(h) x 0.02(d)

Table of Contents

I. Introduction.- 1.1. General Remarks.- 1.2. The Robbins-Monro Process.- 1.3. A “Continuous” Process Version of Section 2.- 1.4. Regulation of a Dynamical System; a simple example.- 1.5. Function Minimization: The Kiefer-Wolfowitz Procedure.- 1.6. Constrained Problems.- 1.7. An Economics Example.- II. Convergence w.p.1 for Unconstrained Systems.- 2.1. Preliminaries and Motivation.- 2.2. The Robbins-Monro and Kiefer-Wolfowitz Algorithms: Conditions and Discussion.- 2.3. Convergence Proofs for RM and KW-like Procedures.- 2.4. A General Robbins-Monro Process: “Exogenous Noise”.- 2.5. A General RM Process; State Dependent Noise.- 2.5.1. Extensions and Localizations of Theorem 2.5.2.- 2.6. Some Applications.- 2.7. Mensov-Rademacher Estimates.- III. Weak Convergence of Probability Measures.- IV. Weak Convergence for Unconstrained Systems.- 4.1. Conditions and General Discussion.- 4.2. The Robbins-Monro and Kiefer-Wolfowitz Procedures.- 4.3. A General Robbins-Monro Process: Exogenous Noise.- 4.4. A General RM Process: State Dependent Noise.- 4.5. The Identification Problem.- 4.6. A Counter-Example to Tightness.- 4.7. Boundedness of {Xn} and Tightness of {Xn(•)}.- V. Convergence w.p.1 For Constrained Systems.- 5.1. A Penalty-Multiplier Algorithm for Equality Constraints.- 5.2. A Lagrangian Method for Inequality Constraints.- 5.3. A Projection Algorithm.- 5.4. A Penalty-Multiplier Method for Inequality Constraints.- VI. Weak Convergence: Constrained Systems.- 6.1. A Multiplier Type Algorithm for Equality Constraints.- 6.2. The Lagrangian Method.- 6.3. A Projection Algorithm.- 6.4. A Penalty-Multiplier Algorithm for Inequality Constraints.- VII. Rates of Convergence.- 7.1. The Problem Formulation.- 7.2. Conditions and Discussions.- 7.3. Rates of Convergence for Case 1,the KW Algorithm.- 7.4. Discussion of Rates of Convergence for Two KW Algorithms.