Automated Gradient Calculation.

Abstract

Automatic differentiation is a powerful technique for evaluating derivatives of functions given in the form of a high-level programming language such as Fortran, C, or C++. The program is treated as a sequence of elementary statements to which the chain rule of differential calculus is applied mechanically. A key feature of this technique is its ability to generate accurate derivatives rather than approximations obtained from numerical differentiation like divided differences. We survey automatic differentiation and report on preliminary results that have been obtained applying the automatic differentiation tool ADIFOR to a large-scale computational fluid dynamics solver, TFS, developed at Aerodynamisches Institut, Aachen University of Technology. This solver comprises approximately 24,000 lines of Fortran 77 and is able to resolve steady and unsteady laminar and turbulent flows in two and three dimensions.