Parallel Computing

Teaching

Research

Publications

People

Theses

Jobs

Address

Talks

Previous Highlights

Events

Research Topics in Automatic Differentiation

Automatic differentiation (AD) is a powerful technique to evaluate derivatives accurately and efficiently. In AD, a given computer code for some function is transformed into a new computer program capable of evaluating not only the function, but also its derivatives. See the community portal www.autodiff.org which I founded in 2000 for more details. My work has been concerned with techniques going beyond the black-box application of AD, among others interface contraction and parallel computing.

Interface Contraction

Interface contraction is a technique to reduce time complexity of AD-generated code by making use of local code structure.

[1]	H. M. Bücker and A. Rasch. Efficient Derivative Computations in Neutron Scattering via Interface Contraction. In Proceedings of the ACM Symposium on Applied Computing, Madrid, Spain, March 10-14, 2002, pages 184-188, New York, 2002. ACM Press. Abstract

Parallel Computing

Given a serial code, we have investigated techniques to automatically produce AD-code for shared-memory parallel machines.

[2]	H. M. Bücker, B. Lang, D. an Mey, and C. H. Bischof. Bringing Together Automatic Differentiation and OpenMP. In Proceedings of the 15th ACM International Conference on Supercomputing, Sorrento, Italy, June 17-21, 2001, pages 246-251, New York, 2001. ACM Press. Abstract

Applications

We have brought the power of AD to problems arising from various scientific and engineering applications including biomedical engineering.

[3]

H. M. Bücker, R. Beucker, and C. H. Bischof. Using Automatic Differentiation for the Minimal p-Norm Solution of the Biomagnetic Inverse Problem. In A. W. Heemink, L. Dekker, H. de Swaan Arons, I. Smit, and T. L. van Stijn, editors, Shaping Future with Simulation, Proceedings of the 4th International Eurosim 2001 Congress, Delft, The Netherlands, June 26-29, 2001. Dutch Benelux Simulation Society, 2001.
Abstract

RWTH Aachen

Computer Science

Scientific Computing

Computational Engineering

Center for Computing and Communication

Last modified: 16.05.2009, webmaster@sc.rwth-aachen.de, Impressum, Disclaimer