Optimization of Stochastic Systems is an outgrowth of class notes of a graduate level seminar on optimization of stochastic systems. Most of the material in the book was taught for the first time during the 1965 Spring Semester while the author was visiting the Department of Electrical Engineering, University of California, Berkeley. The revised and expanded material was presented at the Department of Engineering, University of California, Los Angeles during the 1965 Fall Semester. The systems discussed in the book are mostly assumed to be of discrete-time type with continuous state variables taking values in some subsets of Euclidean spaces. There is another class of systems in which state variables are assumed to take on at most a denumerable number of values, i.e., these systems are of discrete-time discrete-space type. Although the problems associated with the latter class of systems are many and interesting, andalthough they are amenable to deep analysis on such topics as the limiting behaviors of state variables as time indexes increase to infinity, this class of systems is not included here, partly because there are many excellent books on the subjects and partly because inclusion of these materials would easily double the size of the book.
With the advance of new computing technology, simulation is becoming very popular for designing large, complex and stochastic engineering systems, since closed-form analytical solutions generally do not exist for such problems. However, the added flexibility of simulation often creates models that are computationally intractable. Moreover, to obtain a sound statistical estimate at a specified level of confidence, a large number of simulation runs (or replications) is usually required for each design alternative. If the number of design alternatives is large, the total simulation cost can be very expensive. Stochastic Simulation Optimization addresses the pertinent efficiency issue via smart allocation of computing resource in the simulation experiments for optimization, and aims to provide academic researchers and industrial practitioners with a comprehensive coverage of OCBA approach for stochastic simulation optimization. Starting with an intuitive explanation of computing budget allocation and a discussion of its impact on optimization performance, a series of OCBA approaches developed for various problems are then presented, from the selection of the best design to optimization with multiple objectives. Finally, this book discusses the potential extension of OCBA notion to different applications such as data envelopment analysis, experiments of design and rare-event simulation.
Stochastic programming is the study of procedures for decision making under the presence of uncertainties and risks. Stochastic programming approaches have been successfully used in a number of areas such as energy and production planning, telecommunications, and transportation. Recently, the practical experience gained in stochastic programming has been expanded to a much larger spectrum of applications including financial modeling, risk management, and probabilistic risk analysis. Major topics in this volume include: (1) advances in theory and implementation of stochastic programming algorithms; (2) sensitivity analysis of stochastic systems; (3) stochastic programming applications and other related topics. Audience: Researchers and academies working in optimization, computer modeling, operations research and financial engineering. The book is appropriate as supplementary reading in courses on optimization and financial engineering.
Discrete event systems (DES) have become pervasive in our daily lives. Examples include (but are not restricted to) manufacturing and supply chains, transportation, healthcare, call centers, and financial engineering. However, due to their complexities that often involve millions or even billions of events with many variables and constraints, modeling these stochastic simulations has long been a “hard nut to crack”. The advance in available computer technology, especially of cluster and cloud computing, has paved the way for the realization of a number of stochastic simulation optimization for complex discrete event systems. This book will introduce two important techniques initially proposed and developed by Professor Y C Ho and his team; namely perturbation analysis and ordinal optimization for stochastic simulation optimization, and present the state-of-the-art technology, and their future research directions.
From the Preface The first edition of this book was written mainly for audiences with physical science and engineering backgrounds. Nevertheless, it reached some readers with economic and management science training. Analytical training of graduate students in economics and management sciences had progressed much in the last 20 years, and many new research results and optimization algorithms have also become available. My own interest in the meantime has shifted to the analysis of dynamics and optimization problems of economic and management science origin. With these developments and changes, I decided to rewrite much of the first edition to make it more accessible to graduate students and professionals in social sciences. I have also incorporated some new analytic tools that I deem useful in analyzing the dynamic and stochastic problems which confront these readers. I hope that my efforts successfully bring intertemporal optimization problems closer to economics professionals. New topics introduced into this second edition appear mostly in Chapters 2, 4, 5, 6, and 8. Martingales and martingale differences are introduced early in Chapter 2. Some limit theorems and asymptotic properties of linear state space models driven by martingale differences are presented. Because many excellent books are available on martingales and their limit theorems, derivations and proofs are mostly sketchy, and readers are referred to these sources. The results in Chapteer 2 are applied in Chapters 5, 6, and 8, among other places. The notion of dynamic aggregation and its relation to cointegration and error-correction models are developed in Chapter 4. Some recursive parameter estimation schemes and their statistical properties are included in Chapters 5 and 6. Here again, books devoted entirely to these topics are available in the literature, and much had to be omitted to keep the second edition to a manageable size. In an appendix to Chapter 7, a potentially very powerful tool in proving convergence of adaptive schemes is outlined. Rational expectations models and their solution methods are developed in Chapter 8 because of their wide-spread interest to economists. A very important class of problems in sequential decision problems revolves around questions of approximating nonlinear dynamics or more generally complex situations with a sequence of less complex ones. Chapter 9 does not begin to do justice to this class of problems but is included as being suggestive of works to be done. When I first started contemplating the revision of the first edition, I benefited from a list of excellent suggestions from Rick van der Ploeg, though I did not necessarily incorporate all of his suggestions. Conversations with Thomas Sargent and Victor Solo were useful in organizing the material into the form of the second edition. I also benefited from discussions with Hashem Pesaran and correspondences with L. Broze in finalizing Chapter 8. Some material in this book was used as lecture notes in a graduate course in the Department of Economics, University of California, Los Angeles, the winter quarter of 1987. I thank the participants in the course for many useful comments. Key Features * This major revision of the First Edition addresses optimization problems stated in stochastic difference equations, which often contain uncertain or randomly varying parameters * Presents a set of concepts and techniques useful in analyzing or controlling stochastic dynamic processes, with possible incompletely specified characteristics * It discusses basic system properties such as: * Stability and observability * Dynamic programming formulations of optimal and adaptive control problems * Parameter estimation schemes and their convergence behavior * Solution methods for rational expectations models using martingale differences
Stochastic optimization problems arise in decision-making problems under uncertainty, and find various applications in economics and finance. On the other hand, problems in finance have recently led to new developments in the theory of stochastic control. This volume provides a systematic treatment of stochastic optimization problems applied to finance by presenting the different existing methods: dynamic programming, viscosity solutions, backward stochastic differential equations, and martingale duality methods. The theory is discussed in the context of recent developments in this field, with complete and detailed proofs, and is illustrated by means of concrete examples from the world of finance: portfolio allocation, option hedging, real options, optimal investment, etc. This book is directed towards graduate students and researchers in mathematical finance, and will also benefit applied mathematicians interested in financial applications and practitioners wishing to know more about the use of stochastic optimization methods in finance.
A recent development in SDC-related problems is the establishment of intelligent SDC models and the intensive use of LMI-based convex optimization methods. Within this theoretical framework, control parameter determination can be designed and stability and robustness of closed-loop systems can be analyzed. This book describes the new framework of SDC system design and provides a comprehensive description of the modelling of controller design tools and their real-time implementation. It starts with a review of current research on SDC and moves on to some basic techniques for modelling and controller design of SDC systems. This is followed by a description of controller design for fixed-control-structure SDC systems, PDF control for general input- and output-represented systems, filtering designs, and fault detection and diagnosis (FDD) for SDC systems. Many new LMI techniques being developed for SDC systems are shown to have independent theoretical significance for robust control and FDD problems.
