[FRPythoneers] I must be sitting too close to the monitor.

Fernando Perez Fernando.Perez at Colorado.edu
Thu Jan 13 10:51:16 MST 2005


Ben Harding wrote:
> What we call floating point numbers are a mapping of a continuous number 
> space (real numbers) into a discrete number space (integers.)  There are 
> holes in this map and some real numbers cannot be represented exactly.  
> The internal representation used by the machine or software does it's 
> best, but can't be perfect.
> 
> I bet there are others on this list that really know about this, and I 
> would be interested in learning more about the floating point 
> representation in general and that used by Python.

Python just uses C doubles for floats, so the usual caveats apply.  Here's a 
simple but good discussion:

http://lua-users.org/wiki/FloatingPoint

At the bottom, there are links to in-depth references, including the one in 
the python docs:

http://lua-users.org/wiki/FloatingPoint

Reading reference http://citeseer.ist.psu.edu/goldberg91what.html would be 
most educational, though it's a hefty 78 pages.

Cheers,

f



More information about the FRPythoneers mailing list