[Numpy-discussion] not expected output of fill_diagonal

2012-06-08 Thread Frédéric Bastien
Hi, While reviewing the Theano op that wrap numpy.fill_diagonal, we found an unexpected behavior of it: # as expected for square matrix >>> a=numpy.zeros((5,5)) >>> numpy.fill_diagonal(a, 10) >>> print a # as expected long rectangular matrix >>> a=numpy.zeros((3,5)) >>> numpy.fill_diagonal(a, 10

Re: [Numpy-discussion] Numpy structures

2012-06-08 Thread Nathaniel Smith
On Fri, Jun 8, 2012 at 11:31 AM, Bob Cowdery wrote: > Hi all, > > I am reading a datagram which contains within it a type. The type > dictates the structure of the datagram. I want to put this into a numpy > structure, one of which is: > np.zeros(1,dtype=('2uint8,uint8,uint8,uint32,8uint8,504uint8

Re: [Numpy-discussion] possible enhancement to getitem?

2012-06-08 Thread Mike Ressler
> On 08/06/12 14:14, Neal Becker wrote: >> The fact that this proposed numpy behavior would not match python list >> behavior >> holds little weight for me.  I would still favor this change, unless it added >> significant overhead.  My opinion, of course. As a "Joe User", I think using the [-2:2]

Re: [Numpy-discussion] possible enhancement to getitem?

2012-06-08 Thread Bob Dowling
On 08/06/12 14:14, Neal Becker wrote: > The fact that this proposed numpy behavior would not match python list > behavior > holds little weight for me. I would still favor this change, unless it added > significant overhead. My opinion, of course. It holds enormous weight for me. My opinion i

Re: [Numpy-discussion] possible enhancement to getitem?

2012-06-08 Thread Jonathan T. Niehof
On 06/07/2012 12:55 PM, Neal Becker wrote: > In [3]: u = np.arange(10) > > In [4]: u > Out[4]: array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9]) > > In [5]: u[-2:] > Out[5]: array([8, 9]) > > In [6]: u[-2:2] > Out[6]: array([], dtype=int64) > > I would argue for consistency it would be desirable for this to re

Re: [Numpy-discussion] possible enhancement to getitem?

2012-06-08 Thread Alan G Isaac
On 6/8/2012 9:14 AM, Neal Becker wrote: > The fact that this proposed numpy behavior would not match python list > behavior > holds little weight for me. It is not just Python behavior for lists. It is the semantics for all sequence types. Breaking this would be appalling. Alan Isaac

Re: [Numpy-discussion] possible enhancement to getitem?

2012-06-08 Thread Neal Becker
Robert Kern wrote: > On Thu, Jun 7, 2012 at 7:55 PM, Neal Becker wrote: >> In [3]: u = np.arange(10) >> >> In [4]: u >> Out[4]: array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9]) >> >> In [5]: u[-2:] >> Out[5]: array([8, 9]) >> >> In [6]: u[-2:2] >> Out[6]: array([], dtype=int64) >> >> I would argue for consi

[Numpy-discussion] Numpy structures

2012-06-08 Thread Bob Cowdery
Hi all, I am reading a datagram which contains within it a type. The type dictates the structure of the datagram. I want to put this into a numpy structure, one of which is: np.zeros(1,dtype=('2uint8,uint8,uint8,uint32,8uint8,504uint8,8uint8,504uint8')) As I don't know what I'm getting until I've