Publication: The Adjoint Data-Flow Analyses: Formalization, Properties, and Applications
Introduction
Applications
Tools
Research Groups
Workshops
Publications
   List Publications
   Advanced Search
   Info
   Add Publications
My Account
About

The Adjoint Data-Flow Analyses: Formalization, Properties, and Applications

- incollection -
 

Area
General

Author(s)
Laurent HascoŽt , Mauricio Araya-Polo

Published in
Automatic Differentiation: Applications, Theory, and Implementations

Editor(s)
H. M. Bücker, G. Corliss, P. Hovland, U. Naumann, B. Norris

Year
2005

Publisher
Springer

Abstract
Automatic Differentiation (ad) is a program transformation that yields derivatives. Building efficient derivative programs requires complex and specific static analysis algorithms to reduce run time and memory usage. Focusing on the reverse mode of ad, which computes adjoint programs, we specify jointly the central static analyses that are required to generate an efficient adjoint code. We use a set-based formalization from classical data-flow analysis to specify Adjoint Liveness, Adjoint Write, and To Be Recorded analyses, and their mutual influences, taking into account the specific structure of adjoint programs. We give illustrations on examples taken from real numerical programs, that we differentiate with our ad tool TAPENADE, which implements these analyses.

Cross-References
Bucker2005ADA

AD Tools
TAPENADE

AD Theory and Techniques
Data Flow Analysis, Adjoint

BibTeX
@INCOLLECTION{
         Hascoet2005TAD,
       title = "The Adjoint Data-Flow Analyses: {F}ormalization, Properties, and Applications",
       editor = "H. M. B{\"u}cker and G. Corliss and P. Hovland and U. Naumann and B.
         Norris",
       booktitle = "Automatic Differentiation: {A}pplications, Theory, and Implementations",
       series = "Lecture Notes in Computational Science and Engineering",
       publisher = "Springer",
       ad_area = "General",
       ad_tools = "TAPENADE",
       ad_theotech = "Data Flow Analysis, Adjoint",
       year = "2005",
       author = "Laurent Hasco{\"e}t and Mauricio Araya-Polo",
       abstract = "Automatic Differentiation (AD) is a program transformation that yields derivatives.
         Building efficient derivative programs requires complex and specific static analysis algorithms to
         reduce run time and memory usage. Focusing on the reverse mode of AD, which computes adjoint
         programs, we specify jointly the central static analyses that are required to generate an efficient
         adjoint code. We use a set-based formalization from classical data-flow analysis to specify Adjoint
         Liveness, Adjoint Write, and To Be Recorded analyses, and their mutual influences, taking into
         account the specific structure of adjoint programs. We give illustrations on examples taken from
         real numerical programs, that we differentiate with our AD tool TAPENADE, which implements these
         analyses.",
       crossref = "Bucker2005ADA",
       pages = "135--146",
       doi = "10.1007/3-540-28438-9_12"
}


back
  

Contact:
autodiff.org
Username:
Password:
(lost password)