Level crossing methods in stochastic models (Q5900110): Difference between revisions

From the author's preface: ``This monograph elucidates system point level crossing method (LC) for obtaining probability distributions of state variables in a variety of stochastic models. Most of the analyses are for steady-state distributions. However, some results for transient distributions are also given. The book is intended for research- and applications-oriented workers in operations research, management science, engineering, probability and statistics, actuarial science, mathematics, and the natural sciences. The LC method is very intuitive and leads to exact analytical solutions. It focuses on time rates at which a sample path exits and enters certain measurable state-space sets. Level-crossing theorems equate these transition rates to simple algebraic expressions of the pdf (probability distribution function) and/or cdf (cumulative distribution function) of the state variable. In a steady-state analysis, the algebraic expressions often appear in separate terms of Volterra integral equations of the second type with parameter. Thus, ``physical sample-path transition rates are in one-to-one correspondence with terms of the integral equations. The integral equations themselves are constructed by applying rate conservation laws, e.g., rate balance. The upshot is that we can write down the integral equations ``by inspection'', upon observing the sample-path structure of a model! It is axiomatic that one can reach solutions for mathematical models by applying alternative techniques. My own experience, and that of many other researchers, has demonstrated that LC often leads quickly and easily to solutions. It provides useful intuition about the model dynamics. This is due to the perspective taken: geometric sample-path structure; rate conservation laws; connection to concepts of natural science such as Physics. LC may free the analyst from lengthy derivations of a system of model equations. Chapter 1 outlines the original developmental ideas which led me to the discovery of LC. When combined synergistically, the basic ideas lead to a powerful modeling technique. Chapter 2 defines and discusses basic concepts relevant to the method, such as: state space, sample path, system point (SP), SP jump, state-space level, boundary, downcrossing, upcrossing, tangent, etc. Chapters 3, 4, and 5 analyze steady-state distributions in variants of M/G/1, M/M/c and G/M/c queues, respectively. Chapters 3 and 4 also provide some basic results for transient distributions. Chapter 6 analyzes steady-state distributions in several basic dams, and in two inventory models. It also includes some transient results. Chapter 7 demonstrates a multi-dimensional technique with applications to two 2-dimensional inventory models. Chapter 8 explains the embedded level crossing technique with applications to dams and queues. Chapter 9 gives an introduction to level crossing estimation, which uses simulation of sample paths to obtain solutions. Chapter 10 applies LC to a variety of models including: a replacement model, renewal theory, Markov renewal theory, Markov chains, growth and counter models, a dam with alternating continuous influx and efflux, simple harmonic motion. It also illustrates some transient analyses. I hope that readers will find the monograph interesting, and useful for research.''