[Numpy-discussion] f.p. powers and masked arrays
Michael Fitzgerald
mpfitz at calmail.berkeley.edu
Fri Jun 23 17:15:58 EDT 2006
Ping!
Is anyone else seeing this? It should be easy to test. If so, I
think it's a bug.
Best,
Mike
On Jun 21, 2006, at 9:39 PM, Michael Fitzgerald wrote:
>
> Hello all,
>
> I'm encountering some (relatively new?) behavior with masked arrays
> that
> strikes me as bizarre. Raising zero to a floating-point value is
> triggering
> a mask to be set, even though the result should be well-defined.
> When using
> fixed-point integers for powers, everything works as expected. I'm
> seeing
> this with both numarray and numpy. Take the case of 0**1,
> illustrated below:
>
>>>> import numarray as n1
>>>> import numarray.ma as n1ma
>>>> n1.array(0.)**1
> array(0.0)
>>>> n1.array(0.)**1.
> array(0.0)
>>>> n1ma.array(0.)**1
> array(0.0)
>>>> n1ma.array(0.)**1.
> array(data =
> [1.0000000200408773e+20],
> mask =
> 1,
> fill_value=[ 1.00000002e+20])
>
>>>> import numpy as n2
>>>> import numpy.core.ma as n2ma
>>>> n2.array(0.)**1
> array(0.0)
>>>> n2.array(0.)**1.
> array(0.0)
>>>> n2ma.array(0.)**1
> array(0.0)
>>>> n2ma.array(0.)**1.
> array(data =
> 1e+20,
> mask =
> True,
> fill_value=1e+20)
>
> I've been using python v2.3.5 & v.2.4.3, numarray v1.5.1, and numpy
> v0.9.8,
> and tested this on an x86 Debian box and a PPC OSX box. It may be
> the case
> that this issue has manifested in the past several months, as it's
> causing a
> new problem with some of my older code. Any thoughts?
>
> Thanks in advance,
> Mike
>
>
> All the advantages of Linux Managed Hosting--Without the Cost and
> Risk!
> Fully trained technicians. The highest number of Red Hat
> certifications in
> the hosting industry. Fanatical Support. Click to learn more
> http://sel.as-us.falkag.net/sel?
> cmd=lnk&kid=107521&bid=248729&dat=121642
> _______________________________________________
> Numpy-discussion mailing list
> Numpy-discussion at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/numpy-discussion
More information about the NumPy-Discussion
mailing list