FP-ANR: A representation format to handle floating-point cancellation at run-time - LIRMM - Laboratoire d’Informatique, de Robotique et de Microélectronique de Montpellier Access content directly
Preprints, Working Papers, ... Year : 2017

FP-ANR: A representation format to handle floating-point cancellation at run-time

Abstract

When dealing with floating-point numbers there are several sources of error which can drastically reduce the numerical quality of computed results. Among those errors, the loss of significance, or cancellation, occurs during for example the subtraction of two nearly equal numbers. In this article, we propose a representation format named Floating-Point Adaptive Noise Reduction (FP-ANR). This format embeds cancellation information directly into the floating-point representation format thanks to a dedicated pattern. With this format, unsignificant trailing bit lost during cancellation are removed from every manipulated floating-point number. The immediate consequence is that it increases the numerical confidence of computed values. The proposed representation format corresponds to a simple and efficient implementation of significance arithmetic based and compatible with the IEEE-754 standard.
Fichier principal
Vignette du fichier
cancellation.pdf (254.82 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

lirmm-01549601 , version 1 (28-06-2017)
lirmm-01549601 , version 2 (30-10-2017)
lirmm-01549601 , version 3 (22-04-2018)

Identifiers

  • HAL Id : lirmm-01549601 , version 3

Cite

David Defour. FP-ANR: A representation format to handle floating-point cancellation at run-time. 2017. ⟨lirmm-01549601v3⟩
398 View
942 Download

Share

Gmail Facebook X LinkedIn More