numpy-discussion

[Numpy-discussion] flatiter and inequality comparison

[Alan G I. <ai...@am...>]

I do not understand how to think about this:

    >>> x=arange(3).flat
    >>> x
    <numpy.flatiter object at 0x01BD0C58>
    >>> x>2
    True
    >>> x>10
    True

Why?  (I realize this behaves like xrange,
so this may not be a numpy question,
but I do not understand that behavior either.)

What I expected:
that a flatiter object would iterate through
its values and return either
- a flatiter of the resulting comparisons, or
- an array of the resulting comparisons

Thank you,
Alan Isaac

Re: [Numpy-discussion] flatiter and inequality comparison

[Travis O. <oli...@ee...>]

Alan G Isaac wrote:

>I do not understand how to think about this:
>
>    >>> x=arange(3).flat
>    >>> x
>    <numpy.flatiter object at 0x01BD0C58>
>    >>> x>2
>    True
>    >>> x>10
>    True
>
>Why?  (I realize this behaves like xrange,
>so this may not be a numpy question,
>but I do not understand that behavior either.)
>  
>
The flatiter object didn't have comparisons implemented so I guess it 
was using some default implementation.  This is quite confusing and 
option 2 does make sense (an array of resulting comparisions is returned).

Thus now:

 >> x=arange(3).flat
 >>> x>2
array([False, False, False], dtype=bool)
 >>> x>1
array([False, False, True], dtype=bool)
 >>> x>0
array([False, True, True], dtype=bool)


-Travis

Re: [Numpy-discussion] flatiter and inequality comparison

[Alan G I. <ai...@am...>]

On Fri, 23 Jun 2006, Travis Oliphant apparently wrote: 
> option 2 does make sense (an array of resulting comparisions is returned).  
> Thus now: 
>  >> x=arange(3).flat 
>  >>> x>2 
> array([False, False, False], dtype=bool) 

Thanks!!
Alan