Direct Methods of Parameter Optimization for Dynamical Systems

摘要

A mathematical basis is established to facilitate solution of problems in the general area of parameter optimization for physical systems that are described by ordinary differential equations. Also included is a survey of direct methods of solution of parameter optimization problems. Steepest descent, conjugate gradient, and Davidon algorithms are specifically considered and a penalty function approach to the imposition of constraints (both equality and inequality) is taken. A problem of particular interest that is specifically treated involves those systems for which the optimum parameters can be defined without knowledge of the solution of the differential equations. (Author)