Publication: Do you trust derivatives or differences?
Introduction
Applications
Tools
Research Groups
Workshops
Publications
   List Publications
   Advanced Search
   Info
   Add Publications
My Account
About

Do you trust derivatives or differences?

- Article in a journal -
 

Author(s)
Jorge J. Moré , Stefan M. Wild

Published in
Journal of Computational Physics

Year
2014

Abstract
We analyze the relationship between the noise level of a function and the accuracy and reliability of derivatives and difference estimates. We derive and empirically validate measures of quality for both derivatives and difference estimates. Using these measures, we quantify the accuracy of derivatives and differences in terms of the noise level of the function. An interesting observation based on these results is that the derivative of a function is not likely to have working precision accuracy for functions with modest levels of noise.

AD Tools
ADiMat, INTLAB

AD Theory and Techniques
Stability

BibTeX
@ARTICLE{
         More2014Dyt,
       title = "Do you trust derivatives or differences?",
       journal = "Journal of Computational Physics",
       volume = "273",
       pages = "268--277",
       year = "2014",
       issn = "0021-9991",
       doi = "http://dx.doi.org/10.1016/j.jcp.2014.04.056",
       url = "http://www.sciencedirect.com/science/article/pii/S0021999114003325",
       author = "Jorge J. Mor{\'e} and Stefan M. Wild",
       keywords = "Iterative solvers",
       abstract = "We analyze the relationship between the noise level of a function and the accuracy
         and reliability of derivatives and difference estimates. We derive and empirically validate measures
         of quality for both derivatives and difference estimates. Using these measures, we quantify the
         accuracy of derivatives and differences in terms of the noise level of the function. An interesting
         observation based on these results is that the derivative of a function is not likely to have
         working precision accuracy for functions with modest levels of noise.",
       ad_tools = "ADiMat, INTLAB",
       ad_theotech = "Stability"
}


back
  

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