More accuracy at fixed precision

摘要

Several different techniques and software intend to improve the accuracy of results computed in a fixed finite precision. Here we focus on the CENA method that processes an automatic correction of the first-order effect of the rounding errors the computation generates. This method provides a corrected result and a bound of the residual error for a class of algorithms we identify. We present the main features of the CENA method and illustrate its interests and limitations with examples.