Publication: Combining Source Transformation and Operator Overloading Techniques to Compute Derivatives for MATLAB Programs
Introduction
Applications
Tools
Research Groups
Workshops
Publications
   List Publications
   Advanced Search
   Info
   Add Publications
My Account
About
Impress

Combining Source Transformation and Operator Overloading Techniques to Compute Derivatives for MATLAB Programs

- Part of a collection -
 

Author(s)
Christian H. Bischof , H. Martin Bücker , Bruno Lang , Arno Rasch , Andre Vehreschild

Published in
Proceedings of the Second IEEE International Workshop on Source Code Analysis and Manipulation (SCAM 2002)

Year
2002

Publisher
IEEE Computer Society

Abstract
Derivatives of mathematical functions play a key role in various areas of numerical and technical computing. MATLAB is a popular environmen t for technical computing providing engineers and scientists with capabilities fo r mathematical computing, analysis, visualization, and algorithmic development. For functions written in the MATLAB language, a novel software tool is proposed to automatically transform a given MATLAB program into another MATLA B program capable of computing not only the original function but also user -specified derivatives of that function. That is, a program transformation k nown as automatic differentiation is performed to change the semantics of a program in a fashion based on the chain rule of differential calculus. Th e crucial ingredient of the tool is a combination of source-to-source trans formation and operator overloading. The overall design of the tool is described and numerical experiments are reported demonstrating the efficiency of the resu lting code for a sample problem.

AD Tools
ADiMat

BibTeX
@INPROCEEDINGS{
         Bischof2002CST,
       author = "Christian H.~Bischof and H. Martin B{\"u}cker and Bruno Lang and Arno Rasch
         and Andre Vehreschild",
       title = "Combining Source Transformation and Operator Overloading Techniques to Compute
         Derivatives for {MATLAB} Programs",
       booktitle = "Proceedings of the Second {IEEE} International Workshop on Source Code Analysis
         and Manipulation ({SCAM} 2002)",
       pages = "65--72",
       doi = "10.1109/SCAM.2002.1134106",
       address = "Los Alamitos, CA, USA",
       publisher = "IEEE Computer Society",
       abstract = "Derivatives of mathematical functions play a key role in various areas of numerical
         and technical computing. MATLAB is a popular environmen t for technical computing providing
         engineers and scientists with capabilities fo r mathematical computing, analysis, visualization, and
         algorithmic development. For functions written in the MATLAB language, a novel software tool is
         proposed to automatically transform a given MATLAB program into another MATLA B program capable of
         computing not only the original function but also user -specified derivatives of that function. That
         is, a program transformation k nown as automatic differentiation is performed to change the
         semantics of a program in a fashion based on the chain rule of differential calculus. Th e crucial
         ingredient of the tool is a combination of source-to-source trans formation and operator
         overloading. The overall design of the tool is described and numerical experiments are reported
         demonstrating the efficiency of the resu lting code for a sample problem.",
       year = "2002",
       ad_tools = "ADiMat"
}


back
  

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