Automatic Differentiation of Algorithms From Simulation to Optimizat
Automatic Differentiation (AD) is a maturing computational technology and has become a mainstream tool used by practicing scientists and computer engineers. The rapid advance of hardware computing power and AD tools has enabled practitioners to quickly ge
- PDF / 40,581,977 Bytes
- 431 Pages / 439.32 x 666.12 pts Page_size
- 2 Downloads / 204 Views
Springer Science+Business Media, LLC
George Corliss Christele Faure Andreas Griewank Laurent Hascoet UweNaumann Editors
Automatic Differentiation of Algorithms From Simulation to Optimization
With 94 Illustrations
,
Springer
George Corliss Department of Electrical and Computer Engineering Marquc:ne Univasity P.O. Box 1881 1515 W. Wisconsin Ave.
-
... """""
INRlA. prop Tropics Route des Lucioles 06902
Sophia Antipolis, Fnnc:e Uu4rt!N.Hascodlil.wp/li4.illria./r
Milwaukee. W1 53233, USA
Gl!orgt!.CorliSJIiMorqwene.t!.
OuisteIcFaure
U_Naumann
PolySpace Tcchno1ogies 28 Rue Estienne d'Orves 92120
Department of Computer Science University of Her!fordsbire
Monttougc. FIVICC Chrisftit!.Fawf't!fitpolYSJHICt.com
ALIO 9AB Hat:ficld Herts, United Kin(dom U.l.NQWMNt.ClherU.ac.1Ik
Andreas Griewank Institute of Scientific Computina: Tcc:hnical Univenity
Draden, MOIIl!IUCDStr. 13
01062 Drcaden, Germany ,[email protected]""th
Librvy of Conpsa Catalosing-iD-Public:ation DMa Automatic diffemltiatioo of a1JOrl1hms : from ,imulation to optimiz.ation I edilOn Oeorp Corliss ... let al.J.
p.=
..... Selected papers from !be Third iDlemationaI Conference on Automatic Differentiation, .•. June 2000, ... C&c d'Azur." Includes bibliognphical refeteoocs ud index.
Additional matmal to this book can be downloaded from bttp:/lextns.SPrin8er.com. ISBN 978-1-4612-6543-6 1. Differential calculus-Data proccssinJ---Coo.gI'C33eS. I. Corliu. Oecqe F. n. International Confcr-enc:e on Automatic Differentiation (3rd : 2000 : COle d'Azur, France) QA304 .A78 2001 515'.33--dc21 2001042959
Printed on acid-free paper.
Additional material to this book can be dOlmloaded from http://extras.springer,com, ISBN 978-1-4612-6543-6 ISBN 978-1-4613-0075-5 (eBook) DOI 10. 1007/978-1-4613-0075-5 Text C 2002 Springer Science+Business Modia New York
Originally published in Springer-Verlag New York Inc. in 2002 Softcover reprint of the hardcover 1st edition 2002 987654321
SPIN 108]9134
Preface This volume comprises selected papers from the Third International Conference on Automatic Differentiation, which took place in June 2000 near the old harbor of Nice, Cote d' AZUL Like the earlier meetings in Breckenridge, Colorado (1991) and Santa Fe, New Mexico (1996), it was supported by SIAM, but this time the organization was in the most capable hands of INRIA, Sophia Antipolis.
Goals As demonstrated by many case studies in this volume, a growing number of scientific and engineering computing applications require derivatives. This need for accurate sensitivity information arises when parameters of nonlinear models need to be adjusted to fit data or when one wishes to optimize performance characteristics by varying design and control variables. Differencing is still a very popular way to approximate derivatives. When exact (truncation-error free) sensitivity values are required, computer algebra, hand-coding and automatic differentiation (AD, also sometimes called computational or algorithmic differentiation) are the only choices. As many studie
