[Numpy-discussion] incorrect behavior when complex number with zero imaginary part is multiplied by inf

Pauli Virtanen pav at iki.fi
Fri Jan 28 10:23:09 EST 2011


Fri, 28 Jan 2011 11:49:34 +0100, Mark Bakker wrote:
[clip]
> Yet when the imaginary part is zero (and it really is a real number),
> the imaginary part is nan:
> 
> In [19]: inf * (1+0j)
> Out[19]: (inf+nan*j)
> 
> That is not correct. It should really given (inf+0*j). (I know where it
> comes from, inf*0 is not defined, but in this case it is, as 1+0j is
> really a real number and inf is by definition real as well).
> 
> If there is consensus I can file a ticket.

Both behaviors are accepted by the C99 standard: all combinations where 
one entry is `+-inf` are equivalent to the complex infinity. gcc itself 
returns `inf-1j*nan`. gfortran returns `inf+1j*nan`.

A good rationale for the present behavior is that there is no way to know 
that 1+0j is supposed to be real; it could as well be a number too small 
to represent (eg. result from an underflow in the imaginary part), in 
which case `nan` is indeed the correct result.

-- 
Pauli Virtanen




More information about the NumPy-Discussion mailing list