[Numpy-discussion] 0/0 is 0 or nan?

Robert Kern robert.kern@gmail....
Tue Mar 13 23:30:17 CDT 2007


Christopher Ball wrote:

> In Python, a/0 gives a divide-by-zero error for any a, but in numpy, I 
> can ignore divide-by-zero errors. If I do this, why can't I still have 
> 0/0 be nan?
> 
> seterr(divide='ignore')
> a=array([0],dtype=float_)
> b=array([0],dtype=float_)
> divide(a,b)
> 
> gives nan, so why doesn't
> 
> seterr(divide='ignore')
> a=array([0],dtype=int_)
> b=array([0],dtype=int_)
> divide(a,b)
> 
> also give nan, rather than 0 as it gives now? Is this because inf is not 
> something to which an int can be set (as Christopher Barker wrote), and 
> it matches Python's behavior?

Pretty much. ints operated with ints will give ints, not floats. nan and inf are
floating point concepts.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
 that is made terrible by our own mad attempt to interpret it as though it had
 an underlying truth."
  -- Umberto Eco


More information about the Numpy-discussion mailing list