Publication: Automatically Differentiating a Two-dimensional Finite-Difference Time-Domain Program
Introduction
Applications
Tools
Research Groups
Workshops
Publications
   List Publications
   Advanced Search
   Info
   Add Publications
My Account
About

Automatically Differentiating a Two-dimensional Finite-Difference Time-Domain Program

- Part of a collection -
 

Area
Electrical Engineering

Author(s)
H. M. Bücker , A. Vehreschild

Published in
Proceedings of the 16th Conference on the Computation of Electromagnetic Fields COMPUMAG 2007, Aachen, Germany, June 24--28, 2007

Year
2007

Abstract
Numerical simulations of electromagnetic phenomena are increasingly becoming the starting point of further scientific investigations. Examples include assessing the robustness of a given simulation code, fitting parameters of a computer model to experimental data, or designing an appropriate computer model of a technical system. Derivatives play an important role in these analyzes. Rather than using numerical differentiation based on divided differencing, we sketch the transformation of a simulation code written in MATLAB into a new MATLAB program capable of evaluating the derivatives without truncation error.

AD Tools
ADiMat

BibTeX
@INPROCEEDINGS{
         Bucker2007ADa,
       author = "H. M. B{\"u}cker and A. Vehreschild",
       title = "Automatically Differentiating a Two-dimensional Finite-Difference Time-Domain
         Program",
       booktitle = "Proceedings of the 16th Conference on the Computation of Electromagnetic Fields
         COMPUMAG~2007, Aachen, Germany, June~24--28, 2007",
       pages = "83--84",
       abstract = "Numerical simulations of electromagnetic phenomena are increasingly becoming the
         starting point of further scientific investigations. Examples include assessing the robustness of a
         given simulation code, fitting parameters of a computer model to experimental data, or designing an
         appropriate computer model of a technical system. Derivatives play an important role in these
         analyzes. Rather than using numerical differentiation based on divided differencing, we sketch the
         transformation of a simulation code written in MATLAB into a new MATLAB program capable of
         evaluating the derivatives without truncation error.",
       year = "2007",
       ad_area = "Electrical Engineering",
       ad_tools = "ADiMat"
}


back
  

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