Mailman 3 December 2004 - NumPy-Discussion

ufuncs on user-defined types in numarray
by konrad.hinsen＠laposte.net 17 Dec '04

17 Dec '04

I am working on making ScientificPython (http://dirac.cnrs-orleans.fr/ScientificPython/) compatible with numarray. One problem I stumbled across is that one feature of Numeric seems to be absent from numarray: the possibility to call ufuncs with arguments of non-numeric types, leading to a corresponding method call. As an illustration, the following example works fine with Numeric, but crashes with numarray: from Numeric import sqrt #from numarray import sqrt class DummyValue: def __init__(self, string): self.string = string def __str__(self): return self.string def sqrt(self): return DummyValue('sqrt(%s)' % self.string) x = DummyValue('x') print sqrt(x) Is this a bug or a feature? Is there another way to make ufuncs work with user-defined classes and types? Konrad. -- --------------------------------------------------------------------- Konrad Hinsen Laboratoire Léon Brillouin, CEA Saclay, 91191 Gif-sur-Yvette Cedex, France Tel.: +33-1 69 08 79 25 Fax: +33-1 69 08 82 61 E-Mail: hinsen(a)llb.saclay.cea.fr ---------------------------------------------------------------------

2 2

binary thinning
by Bob Klimek 16 Dec '04

16 Dec '04

Hi, I've been playing around with the binary_hit_or_miss function in the nd_image and it seems to work well in general. I understand that binary thinning can be generated by using hit_or_miss, however I can't seem to get it to work right. Anyone have a code snippet that does the job? I follow the procedure at: http://homepages.inf.ed.ac.uk/rbf/HIPR2/thin.htm My implementation is as follows: ## from HIPR2 web site ## [0,0,0] ## [x,1,x] ## [1,1,1] struct1 = numarray.array( [[0, 0, 0], [0, 1, 0], [1, 1, 1]]) struct2 = numarray.array( [[1, 1, 1], [0, 0, 0], [0, 0, 0]]) ## [x,0,0] ## [1,1,0] ## [x,1,x] struct3 = numarray.array( [[0, 0, 0], [1, 1, 0], [0, 1, 0]]) struct4 = numarray.array( [[0, 1, 1], [0, 0, 1], [0, 0, 0]]) b = [] for i in range(4): temp = ND.binary_hit_or_miss(a, struct1, struct2) b.append(temp) temp = ND.binary_hit_or_miss(a, struct3, struct4) b.append(temp) struct1 = ND.rotate(struct1, 90) struct2 = ND.rotate(struct2, 90) struct3 = ND.rotate(struct3, 90) struct4 = ND.rotate(struct4, 90) result = b[0] for i in range(7): result = numarray.logical_or(result, b[i+1]) result = a - result Bob

2 4

Re: [ctypes-users] copying/slicing ctypes arrays, (c_ulong *n)()
by None 15 Dec '04

15 Dec '04

1 0

started work on new version of LAPACK
by Piotr Luszczek 14 Dec '04

14 Dec '04

Hello all, on behalf of the LAPACK/ScaLAPACK team I'd like to inform you that we have just received funding to work on a new version of LAPACK and ScaLAPACK. Right now we are soliciting input from various groups that use the software. Numpy community is an important user and we would like to hear from you. I have subscribed to this list so I will receive any responses and pass them along to the rest the LAPACK/ScaLAPACK team. Here is a list to give you a brief idea of the issues we are trying to address: - programming language of new implementation (we will keep the old code as it is) - support for threading (OpenMP, pthreads, etc.) - 64-bit addressing - new/improved routines - IEEE 794 (floating-point) support Please let us know what you think. Peter (Piotr Luszczek)

1 0

Using new numeric instances in NumArray and RecArray.
by Ivan Vilata i Balaguer 14 Dec '04

14 Dec '04

Hi all! I am trying to define new numeric types to use them in NumArray and RecArray objects. My intention is to define them as equivalent to other types, but also make possible to distinguish them. I tried "MyInt32 = IntegralType(4, 'MyType')", which registered OK, is equivalent to Int32, but distinguishable from it. However, I am not allowe to create a NumArray based on it: "a = array([1,2], MyInt32)" fails with "RuntimeError: _numarray_init: can't get typeno for type". Is it possible to integrate new types seamlessly into NumArray and RecArray objects? Is this the right approach, or am I getting that wrong? Well, thanks a lot for your cooperation. PS: Please CC: me with your replies, since I am not registered in the list. Thanks again. -- Ivan Vilata i Balaguer >qo< http://www.carabos.com/ Cárabos Coop. V. V V Enjoy Data ""

3 6

Re: [Numpy-discussion] Using new numeric instances in NumArray and RecArray.
by Ivan Vilata i Balaguer 13 Dec '04

13 Dec '04

On Mon, Dec 13, 2004 at 12:20:29PM -0500, Colin J. Williams wrote: > [...] > The complexity would depend partly on whether TimeStuff deals only with > data of the MyTime32 type. > > If all other data types (Int8, Float, Complex etc.) are barred, then the > needed overriding would probably be simpler. > [...] I have the feeling that MyInt32-only support should be enough for the NumArray subclass (it's a uniform array). However, the RecArray subclass should be able to handle all existing types in addition to MyInt32. > > I guess that another factor to consider is whether this is a one-off > project or something which will be in production for years to come. If > the former, it might be simpler to do the time related things outside of > numarray. On the other hand, if execution time is a significant factor, > then it could be worthwhile to modify the ufuncs - that is not a thing I > would like to tackle. Hehe, it will be production code with time-critical requirements. :) However, it now seems to me that subclassing NumArray and RecArray will be the right thing (not the easiest, though) to do. Thanks a lot Colin and Todd for your helpful comments. I'll try to do it and will tell you how it went. Bye! -- Ivan Vilata i Balaguer >qo< http://www.carabos.com/ Cárabos Coop. V. V V Enjoy Data ""

1 0

NumArray - take method
by Colin J. Williams 11 Dec '04

11 Dec '04

This undocumented method has puzzling behaviour. [Dbg]>>> a array([[0, 1, 2], [3, 4, 5], [6, 7, 8]]) [Dbg]>>> b= a[1:1] [Dbg]>>> b array([]) [Dbg]>>> b.shape (0, 3) [Dbg]>>> b.rank 2 I had expected a rank of zero and an indeterminate shape. Colin W.

2 2

numarray vs IPython
by Stephen Walton 07 Dec '04

07 Dec '04

When I run numarray.testall.test() from IPython, I get a slew of error messages instead of the successful test output I get when the same version is run from the stock Python interpreter. Anyone else see this?

5 5

segm fault
by René Bastian 04 Dec '04

04 Dec '04

Hi, i use numarray 1.1.1 with python2.3b1 import RandomArray2 as R w=R.uniform(-1., +1., 44100) produces "Segmentation fault" i never meet that. thanks for your help -- René Bastian http://www.musiques-rb.org : Musique en Python

2 2

Numarray 1.1.1 with Python 2.4 #60 crashes on Windows XP Pro
by Vivek Rao 04 Dec '04

04 Dec '04

I have installed numarray 1.1.1 for Python 2.4. When I go to C:\Python24\Lib\site-packages\numarray and type 'python numtest.py' a crash occurs. Debugging it Microsoft Visual Studio, the error message is "unhandled exception in python.exe (Python24.dll): 0xC0000005 Access Violation" The Python version I am using is Python 2.4 (#60, Nov 30 2004, 11:49:19) [MSC v.1310 32 bit (Intel)] on win32 . Is this a known problem? Vivek Rao

2 1