Nonlinear Equality Constrained Minimization Via The Quadratic Loss Penalty Function

Nonlinear Equality Constrained Minimization Via the Quadratic Loss Penalty Function

Computational Solution of Nonlinear Systems of Equations

Nonlinear equations arise in essentially every branch of modern science, engineering, and mathematics. However, in only a very few special cases is it possible to obtain useful solutions to nonlinear equations via analytical calculations. As a result, many scientists resort to computational methods. This book contains the proceedings of the Joint AMS-SIAM Summer Seminar, ``Computational Solution of Nonlinear Systems of Equations,'' held in July 1988 at Colorado State University. The aim of the book is to give a wide-ranging survey of essentially all of the methods which comprise currently active areas of research in the computational solution of systems of nonlinear equations. A number of ``entry-level'' survey papers were solicited, and a series of test problems has been collected in an appendix. Most of the articles are accessible to students who have had a course in numerical analysis.

Computation and Control

The problem of developing a systematic approach to the design of feed back strategies capable of shaping the response of complicated dynamical control systems illustrates the integration of a wide variety of mathemat ical disciplines typical of the modern theory of systems and control. As a concrete example, one may consider the control of fluid flow across an airfoil, for which recent experiments indicate the possibility of delaying the onset of turbulence by controlling viscosity through thermal actuators located on the airfoil. In general, there are two approaches to the con trol of such a complica. ted process, the development of extremely detailed models of the process followed by the derivation of a more "dedicated" feed back law or the development of a more simple model class followed by the derivation of control laws which are more robust to unmodelled dynamics and exogeneous disturbances. In either approach, the two twin themes of approximation and computation play a significant role in the derivation and implementation of resulting control laws. And there is no doubt that the cross-fertilization between these twin themes and control theory will increase unabated throughout the next decade, not just as an important component of design and implementation of control laws but also as a source of new problems in computational mathematics. In this volume, we present a collection of papers which were deliv ered at the first Bozeman Conference on Computation and Control, held at Montana State University on August 1-11, 1988.