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 generate derivative-enhanced versions of their code for a broad range of applications in applied research and development. "Automatic Differentiation of Algorithms" provides a comprehensive and authoritative survey of all recent developments, new techniques, and tools for AD use. The book covers all aspects of the subject: mathematics, scientific programming ( i.e., use of adjoints in optimization) and implementation (i.e., memory management problems). A strong theme of the book is the relationships between AD tools and other software tools, such as compilers and parallelizers. A rich variety of significant applications are presented as well, including optimum-shape design problems, for which AD offers more efficient tools and techniques. Topics and features: * helpful introductory AD survey chapter for brief overview of the field *extensive applications chapters, i.e., for circuit simulation, optimization and optimal-control shape design, structural mechanics, and multibody dynamical systems modeling *comprehensive bibliography for all current literature and results for the field *performance issues *optimal control sensitivity analysis *AD use with object oriented software tool kits The book is an ideal and accessible survey of recent developments and applications of AD tools and techniques for a broad scientific computing and computer engineering readership. Practitioners, professionals, and advanced graduates working in AD development will find the book a useful reference and essential resource for their work.

A survey book focusing on the key relationships and synergies between automatic differentiation (AD) tools and other software tools, such as compilers and parallelizers, as well as their applications. The key objective is to survey the field and present the recent developments. In doing so the topics covered shed light on a variety of perspectives. They reflect the mathematical aspects, such as the differentiation of iterative processes, and the analysis of nonsmooth code. They cover the scientific programming aspects, such as the use of adjoints in optimization and the propagation of rounding errors. They also cover "implementation" problems.

The Fourth International Conference on Automatic Di?erentiation was held July20-23inChicago,Illinois.Theconferenceincludedaonedayshortcourse, 42 presentations, and a workshop for tool developers. This gathering of au- matic di?erentiation researchers extended a sequence that began in Breck- ridge, Colorado, in 1991 and continued in Santa Fe, New Mexico, in 1996 and Nice, France, in 2000. We invited conference participants and the general - tomatic di?erentiation community to submit papers to this special collection. The28acceptedpapersre?ectthestateoftheartinautomaticdi?erentiation. The number of automatic di?erentiation tools based on compiler techn- ogy continues to expand. The papers in this volume discuss the implem- tation and application of several compiler-based tools for Fortran, including the venerable ADIFOR, an extended NAGWare compiler, TAF, and TAPE- NADE. While great progress has been made toward robust, compiler-based tools for C/C++, most notably in the form of the ADIC and TAC++ tools, for now operator-overloading tools such as ADOL-C remain the undisputed champions for reverse-mode automatic di?erentiation of C++. Tools for - tomatic di?erentiation of high level languages, including COSY and ADiMat, continue to grow in importance as the productivity gains o? ered by high-level programming are recognized.