1097-0363/asset/FLD_left.gif?v=1&s=acdf92c67291aadb6aee12c67965913ef3672990)
Research Article
On the performance of discrete adjoint CFD codes using automatic differentiation
Article first published online: 27 JAN 2005
DOI: 10.1002/fld.885
Copyright © 2005 John Wiley & Sons, Ltd.
Issue
1097-0363/asset/cover.gif?v=1&s=4f4c032b9f06da7e1e8ac4e112013c5bf916f2e7 "International Journal for Numerical Methods in Fluids")
International Journal for Numerical Methods in Fluids
Special Issue: 8th ICFD Conference on Numerical Methods for Fluid Dynamics
Volume 47, Issue 8-9, pages 939–945, 20 - 30 March 2005
Additional Information
How to Cite
Müller, J.-D. and Cusdin, P. (2005), On the performance of discrete adjoint CFD codes using automatic differentiation. International Journal for Numerical Methods in Fluids, 47: 939–945. doi: 10.1002/fld.885
Publication History
- Issue published online: 4 MAR 2005
- Article first published online: 27 JAN 2005
- Manuscript Accepted: 25 NOV 2004
- Manuscript Revised: 24 NOV 2004
- Manuscript Received: 27 APR 2004
- Abstract
- References
- Cited By
Keywords:
- discrete adjoint;
- CFD;
- automatic differentiation;
- adifor;
- TAF;
- TAMC;
- Tapenade
Abstract
Adjoint methods are a computationally inexpensive way of deriving sensitivity information where there are fewer dependent (cost) variables than there are independent (input) variables. Automatic differentiation (AD) software makes it possible to create discrete adjoint codes with minimal human effort, an issue that had previously restricted acceptance of adjoint CFD codes. In terms of computational performance, automatic code is often assumed to be inferior to hand code. The structure of the underlying code is critical to the performance of the transformed code. This paper reviews the implementation of AD on Fortran CFD codes and gives details of how small rearrangements can be used to produce competitive tangent and adjoint code using source transformation AD. Copyright © 2005 John Wiley & Sons, Ltd.