The motivation to write this paper is the fact that students often blindly trust the results of a computation with the computer and rarely question the results. The general opinion is that the computer is perfect and errors only come into the game because of human errors such as faulty programs or incorrect operation. The aim of this paper is to shake this assumption, at least in terms of arithmetic computation of a computer. We give a short treatise of rounding errors, which is suitable for school lessons or beginner lectures for engineers. By no means we want to discredit computers, they are useful helpers, but their results are definitely to question.

rounding off, b-adic representation, floating-point numbers.

Received: July 4, 2021; Accepted: September 3, 2021; Published: September 28, 2021

How to cite this article: Stefan Ritter, Rounding errors and their effects, Far East Journal of Mathematical Education 21(2) (2021), 155-173. DOI: 10.17654/ME021020155

This Open Access Article is Licensed under Creative Commons Attribution 4.0 International License

References:

[1] Wolfgang Coy, Unsichtbar wird der Fehler, wenn sich alle daran gewoehnt haben, Die Unordnung der Dinge, Eine Wissens- und Mediengeschichte des Unfalls, C. Kassung ed., Bielefeld, 2009, pp. 325-353.
[2] Ulrich Kulisch, Computer arithmetic and validity: theory, implementation, and applications, de Gruyter, Berlin, 2008.
[3] Cleve Moler, Numerical Computing with Matlab, 2nd ed., SIAM, Philadelphia, 2010.
[4] George Szpiro, Kleine Rundungsfehler mit katastrophalen Folgen, Achillesferse der elektronischen Datenverarbeitung, NZZ magazine of 31.10.2013.