Voir la notice de l'article provenant de la source Numdam
The high-quality floating-point implementation of elementary functions is a complex process. A variety of techniques ranging from numerical polynomial approximation algorithms to rigorous computation for bounding the approximation error are used. Firstly, we present a brief overview of this process. Then we show how rigorous computation methods based on multiprecision interval arithmetic and Taylor models are used for certified computation of tight bounds for supremum norms of approximation errors.
Joldes, Mioara 1
@article{RFM_2010__9__43_0, author = {Joldes, Mioara}, title = {When a logarithm is a misspelled algorithm}, journal = {Femmes & math}, pages = {43--47}, publisher = {Association femmes et math\'ematiques}, volume = {9}, year = {2010}, language = {en}, url = {http://geodesic.mathdoc.fr/item/RFM_2010__9__43_0/} }
Joldes, Mioara. When a logarithm is a misspelled algorithm. Femmes & math, Forum 9 des Jeunes Mathématiciennes, Tome 9 (2010), pp. 43-47. http://geodesic.mathdoc.fr/item/RFM_2010__9__43_0/