Why does numpy.array(a[0],b[0]) have this meaning?

[Rick Giuly]

Hello All,

Case 1
This generates an error, which makes sense because the argument should
be a list of numbers:
numpy.array(10,10)

Case 2
This does not generate an error and the result is an array with a
single element:
a = numpy.array([10])
b = numpy.array([10])
numpy.array(a[0],b[0])

The only different I see here between the numpy.array call in the
cases is that
a[0] is a numpy int32
10 is an int

Why would this minor difference in integer types cause a totally
different result for the two cases - or is something else causing the
difference in results?
-rick

P.S.
I am aware that numpy.array([10,10]) will work, but I'm trying to
understand what is going on syntactically/semantically in the two
cases above.

[Marc 'BlackJack' Rintsch]

On Sun, 02 Nov 2008 23:26:24 -0800, Rick Giuly wrote:

Case 1
This generates an error, which makes sense because the argument should
be a list of numbers:
numpy.array(10,10)

Case 2
This does not generate an error and the result is an array with a single
element:
a = numpy.array([10])
b = numpy.array([10])
numpy.array(a[0],b[0])

The only different I see here between the numpy.array call in the cases
is that
a[0] is a numpy int32
10 is an int

Why would this minor difference in integer types cause a totally
different result for the two cases - or is something else causing the
difference in results?

From the `numpy.array` docstring:

Inputs:
object - an array, any object exposing the array interface, any
object whose __array__ method returns an array, or any
(nested) sequence.

And `numpy.int32` instances have an `__array__()` method:

In [225]: ten = numpy.int32(10)

In [226]: ten.__array__()
Out[226]: array(10)

Ciao,
Marc 'BlackJack' Rintsch

[Robert Kern]

Rick Giuly wrote:

Hello All,

Case 1
This generates an error, which makes sense because the argument should
be a list of numbers:
numpy.array(10,10)

Case 2
This does not generate an error and the result is an array with a
single element:
a = numpy.array([10])
b = numpy.array([10])
numpy.array(a[0],b[0])

The only different I see here between the numpy.array call in the
cases is that
a[0] is a numpy int32
10 is an int

Why would this minor difference in integer types cause a totally
different result for the two cases - or is something else causing the
difference in results?

The second argument is for a dtype. Basically, we'll accept anything there that
can be coerced to a dtype using numpy.dtype(). For some reason, we have an
undocumented feature where dtype(some_array_or_numpy_scalar) will return the
dtype of that value. Plain Python ints and floats don't have a dtype attached to
them, so we raise an exception.

If you have more numpy questions, please join us on the numpy-discussion mailing
list.

http://www.scipy.org/Mailing_Lists

--
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