Automatic Differentiation

• Science: Math: Numerical Analysis: Automatic Differentiation  (15)

AUTO_DERIV - Tool for automatic differentiation of Fortran code. Fortran source is available for download for non-commercial applications.

Automatic Differentiation of Fortran Codes (ADIFOR) - Given code as input, it generates code for derivatives including gradients and hessians.

Compiler-based Automatic Differentiation (CompAD) - The project, developed at the RWTH Aachen University, Germany, aims to put automatic differentiation into the commercial NAGWare Fortran compiler. The AD compiler is available upon request from the author.

GRAD - Given a function coded in Fortran, GRAD produces Fortran code to compute the derivatives with respect to specified variables (i.e. the GRADient). The MS DOS binaries can be downloaded freely.

PCOMP - Modelling language to generate Fortran-code for function and gradient evaluation, developed at the University of Bayreuth, Germany. Site contains a short description and references to publications. The tool is available upon request.

Tapenade - With focus on Parallelization and Differentiation of scientific software, Tapenade is an online Automatic Differentiation Engine for Fortran programs developed by the TROPICS team at INRIA, France. Along with extensive documentation, description and references, the tool can be downloaded or used online for non-commercial applications.

TaylUR - An arbitrary-order auto-differentiation module for Fortran 95. It overloads operators and Fortran intrinsics to act correctly on an object of defined type taylor, which encodes a function along with its first few derivatives with respect to several variables (excluding mixed derivatives). Site has links to descriptions and publications, but no application downloads.