[Numpy-discussion] dtype and array creation

[josef.pktd at gmail.com]

On Thu, May 20, 2010 at 4:28 PM, Keith Goodman <kwgoodman at gmail.com> wrote:
> On Thu, May 20, 2010 at 1:19 PM,  <josef.pktd at gmail.com> wrote:
>> On Thu, May 20, 2010 at 4:04 PM, Keith Goodman <kwgoodman at gmail.com> wrote:
>>> Why do the follow expressions give different dtype?
>>>
>>>>> np.array([1, 2, 3], dtype=str)
>>> array(['1', '2', '3'],
>>>      dtype='|S1')
>>>>> np.array(np.array([1, 2, 3]), dtype=str)
>>> array(['1', '2', '3'],
>>>      dtype='|S8')
>>
>> you're on a 64bit machine?
>>
>> S8 is the same size as the float
>>
>>
>>>>> np.array([8]).itemsize
>> 4
>>>>> np.array(np.array([1, 2, 3]), dtype=str)
>> array(['1', '2', '3'],
>>      dtype='|S4')
>>>>> np.array([8]).view(dtype='S4')
>> array(['\x08'],
>>      dtype='|S4')
>>>>> np.array([8]).view(dtype='S1')
>> array(['\x08', '', '', ''],
>>      dtype='|S1')
>>
>> But I don't know whether this is a desired feature, numpy might reuse
>> the existing buffer (?)
>
> Yes, I'm on a 64-bit machine.
>
> That's what I thought so I tried this:
>
>>> a = np.array([1, 2, 3])
>>> type(a[0])
>   <type 'numpy.int64'>
>>> np.array([a[0], a[1], a[2]], dtype=str)
> array(['1', '2', '3'],
>      dtype='|S1')
>
> But it gives '|S1' too. I guess I'm lost.

for sure it doesn't look very consistent, special treatment of 0-dim ?

>>> np.array(a[0], dtype=str)
array('1',
      dtype='|S1')
>>> np.array(a[:1], dtype=str)
array(['1'],
      dtype='|S4')
>>> a[:1].shape
(1,)
>>> a[0].shape
()

Josef

> _______________________________________________
> NumPy-Discussion mailing list
> NumPy-Discussion at scipy.org
> http://mail.scipy.org/mailman/listinfo/numpy-discussion
>