LANS Publications

"Estimating Computational Noise"

J. J. More and S. M. Wild

SIAM J. Sci. Comput., vol. 33, no. 3, , pp. 1291-1314. Also Preprint ANL/MCS-P1721-0210

Preprint Version: [pdf]

Computational noise in deterministic simulations is as ill-defined a concept as can be found in scientifi c computing. When coupled with adaptive strategies, the eff ects of finite precision destroy smoothness of the simulation output and complicate subsequent analysis. Following the work of Hamming on roundoff errors, we present a new algorithm, ECnoise, for quantifying the noise level of a computed function. Our theoretical framework is based on stochastic noise but does not assume a speci fic distribution for the noise. For the deterministic simulations considered, ECnoise produces reliable results in few function evaluations and o ffers new insights into building blocks of large scale simulations.