Epsilon

August 24, 2010 by dberenstein

Today’s message is rather simple.

$1.0 \times 10^{-16}=0$

This is what I get numerically when I try to see if something is really zero on my computer.

The correct test is

$1.0\times 10^{-16}<\epsilon$

where $\epsilon$ is determined by the problem.

Posted in computers, Physics | 2 Comments

on August 25, 2010 at 5:42 am Luboš Motl

Your computer must be pretty bad for the calculation of the vacuum energy.
on August 27, 2010 at 11:04 pm Just Learning

What’s really bad is when you push your computer to the limit and get all sorts of cool looking patterns in the plots of your data, only to realize later that it is purely an artifact forced upon you by some effectively arbitrary limit on significant digits that’s been hard coded into the software.

Comments are closed.

Shores of the Dirac Sea