On Fri, Mar 20, 2015 at 2:47 PM, Rustom Mody <rustompmody at gmail.com> wrote: > IOW: float is a poor approximation to ℝ. Of course it's a poor approximation. That's because real numbers are uncountably infinite, so it's rather tricky to represent and manipulate them efficiently. IEEE 754 gives us practicality rather than purity. ChrisA