Recent Developments in Computational Differentiation
Christian Bischof, RWTH Aachen University, Germany
Date: Wednesday, July 27, 2005
Time: 10:00 - 11:00
Location: 1131 Kemper Hall
Abstract:
The computation of derivatives is a frequent task in nonlinear numerical
methods, sensitivity analysis, parameter identification, or optimization
and experimental design. Automatic differentiation (AD) algorithms and tools
provide a sound basis for reliably and efficiently augmenting computer programs
with statements for the computation of derivatives. Computational differentiation
(CD) approaches then employ AD-tools in suitably chosen computational harnesses
to capitalize on high-level knowledge about numerical paradigms or problem
structure. In this talk, we give an overview of recent developments and
application successes in AD and CD.
Dr. Bischof is the director of the Institute for Scientific Computing at RWTH Aachen University, Germany