On the numerical stability of algorithmic differentiation

DOI10.1007/s00607-011-0162-zzbMath1238.65013OpenAlexW1999333806WikidataQ57389613 ScholiaQ57389613MaRDI QIDQ411440

Andreas Griewank, Kshitij Kulshreshtha, Andrea Walther

Publication date: 4 April 2012

Published in: Computing (Search for Journal in Brave)

Full work available at URL: https://doi.org/10.1007/s00607-011-0162-z

zbMATH Keywords

numerical experimentsautomatic differentiationforward error analysisforward-backward stabilityIEEE arithmeticrounding error estimatesultimate rounding

Mathematics Subject Classification ID

Roundoff error (65G50) Numerical differentiation (65D25) Numerical algorithms for computer arithmetic, etc. (65Y04)

Related Items

Unnamed Item, Efficient computation of derivatives for solving optimization problems in R and Python using SWIG-generated interfaces to ADOL-C, Do you trust derivatives or differences?, Algorithm 946

Uses Software

• Matlab
• mctoolbox
• MAPM
• XBLAS

Cites Work

• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Automatic computation of partial derivatives and rounding error estimates with applications to large-scale systems of nonlinear equations
• Evaluating Derivatives
• Handbook of Floating-Point Arithmetic
• Taylor expansion of the accumulated rounding error
• Fast evaluation of elementary mathematical functions with correctly rounded last bit
• Accuracy and Stability of Numerical Algorithms
• Evaluating higher derivative tensors by forward propagation of univariate Taylor series
• Design, implementation and testing of extended and mixed precision BLAS
• Modern Error Analysis
• Accurate Sum and Dot Product
• Rigorous and portable standard functions