Articles | Volume 9, issue 5
https://doi.org/10.5194/gmd-9-1891-2016
https://doi.org/10.5194/gmd-9-1891-2016
Model description paper
 | 
20 May 2016
Model description paper |  | 20 May 2016

An optimized treatment for algorithmic differentiation of an important glaciological fixed-point problem

Daniel N. Goldberg, Sri Hari Krishna Narayanan, Laurent Hascoet, and Jean Utke

Viewed

Total article views: 3,400 (including HTML, PDF, and XML)
HTML PDF XML Total Supplement BibTeX EndNote
2,304 980 116 3,400 491 167 166
  • HTML: 2,304
  • PDF: 980
  • XML: 116
  • Total: 3,400
  • Supplement: 491
  • BibTeX: 167
  • EndNote: 166
Views and downloads (calculated since 03 Feb 2016)
Cumulative views and downloads (calculated since 03 Feb 2016)

Cited

Saved (preprint)

Latest update: 28 Mar 2024
Download
Short summary
Geophysical adjoint models are powerful tools, allowing sensitivity studies that are not possible otherwise, and enabling optimized fit of models to observing data sets. The complexity involved requires the use of algorithmic differentiation (AD) software, but AD adjoint calculation for ice models can be slow, with prohibitive memory requirements. In this paper, we present a method to improve the performance of ice model adjoint generation, in terms of timing, memory load, and accuracy.