Mailman 3 April 2010 - NumPy-Discussion

Bug in numpy.fix(): broken for scalar arguments
by Eric Firing 18 Apr '10

18 Apr '10

np.fix() no longer works for scalar arguments: In [1]:import numpy as np In [2]:np.version.version Out[2]:'2.0.0.dev8334' In [3]:np.fix(3.14) --------------------------------------------------------------------------- TypeError Traceback (most recent call last) /home/efiring/<ipython console> in <module>() /usr/local/lib/python2.6/dist-packages/numpy/lib/ufunclike.pyc in fix(x, y) 46 if y is None: 47 y = y1 ---> 48 y[...] = nx.where(x >= 0, y1, y2) 49 return y 50 TypeError: 'numpy.float64' object does not support item assignment Eric

3 4

np.mean and np.std performances
by Davide Lasagna 18 Apr '10

18 Apr '10

Hi all, I noticed some performance problems with np.mean and np.std functions. Here is the console output in ipython: # make some test data >>>: a = np.arange(80*64, dtype=np.float64).reshape(80, 64) >>>: c = np.tile( a, [10000, 1, 1]) >>>: timeit np.mean(c, axis=0) 1 loops, best of 3: 2.09 s per loop But using reduce is much faster: def mean_reduce(c): return reduce(lambda som, array: som+array, c) / c.shape[0] >>>:timeit mean_reduce(c) 1 loops, best of 3: 355 ms per loop The same applies to np.std(): # slighlty smaller c matrix (too much memory is used) >>>: c = np.tile( a, [7000, 1, 1]) >>>: timeit np.std(c, axis=0) 1 loops, best of 3: 3.73 s per loop With the reduce version: def std_reduce(c): c -= mean_reduce(c) return np.sqrt( reduce(lambda som, array: som + array**2, c ) / c.shape[0] ) >>>: timeit std_reduce(c) 1 loops, best of 3: 1.18 s per loop For the std function also look at the memory usage during the execution of the function. The functions i gave here can be easily modified to accept an axis option and other stuff needed. Is there any drawback of using them? Why np.mean and np.std are so slow? I'm sure I'm missing something. Cheers Davide

1 0

Fortran ordering
by Paul Northug 17 Apr '10

17 Apr '10

I'd like to use numpy fortran ordering in order to use some external libraries more easily, but it looks like I don't understand how it works and it is causing transposes that are confusing me. When I define an array as: a = np.array([[1.,2.],[3.,4.]], order='F', type=np.float32) how is it stored in memory, as [1, 2, 3, 4] or [1, 3, 2, 4]? a.strides == (4, 8), suggesting [1, 3, 2, 4]. I can also check by stepping through: ap = a.ctypes.data_as(POINTER(c_float)) ap[:4] == [1., 3., 2., 4.] But when I call external libraries through ctypes with a.ctypes.data as argument, the external function gets [1, 2, 3, 4] (I think). Eventually, I plan to use cython's buffer interface and a.data. Will it be [1, 2, 3, 4] there as well? Or is my external code broken. I am glad that numpy hides ordering issues to the extent that it does as it prevents my code from exploding. But I would like to get around this one particular case, so if someone can explain what is going on underneath, I would appreciate it.

3 3

IPythonQt and IPythonZMQ projects
by Gerardo Gutierrez 16 Apr '10

16 Apr '10

Hi all. I'm writting to tell you about some projects we're developing for you and the entire python programmers community. Read some of the documentation of the proposals and please tell us what are your ideas and what do you want or expect to see of this. I'm going to leave you a summary and some links: *IPythonQt* We will implement a Qt-based Graphical User Interface (GUI) to execute Python code with an interpreter that runs in a separate process and the two systems (GUI frontend and interpreter kernel) communicating via the ZeroMQ Messaging library. The bulk of the implementation will be done without dependencies on IPython (only on Zmq). Once the key features are ready, IPython-specific features can be added using the IPython codebase. *links*: http://ipythonqt.blogspot.com/ https://code.launchpad.net/~ipython-contrib/ipython/qt-frontend *IPythonZMQ* IPython's execution in a command-line environment will be ported to a two process model using the ZeroMQ library for inter-process communication. This will: - prevent an interpreter crash from destroying the user session, - allow multiple clients to interact simultaneously with a single interpreter - allow IPython to reuse code for local execution and distributed computing (DC) - give us a path for Python3 support, since ZeroMQ supports Python3 while Twisted (what we use today for DC) does not. *links*: http://ipythonzmq.blogspot.com/ https://code.launchpad.net/~ipython-contrib/+junk/ipython-zmq Best regards. -- Gerardo Gutiérrez Gutiérrez <http://he1.udea.edu.co/gweb> Physics student Universidad de Antioquia Computational physics and astrophysics group (FACom<http://urania.udea.edu.co/sites/sites.php> ) Computational science and development branch(FACom-dev<http://urania.udea.edu.co/sites/facom-dev/> ) Usuario Linux #492295

1 0

rc2 for NumPy 1.4.1 and Scipy 0.7.2
by Ralf Gommers 16 Apr '10

16 Apr '10

Hi, I am pleased to announce the second release candidate of both Scipy 0.7.2 and NumPy 1.4.1, please test them. The issues reported with rc1 should be fixed, and for NumPy there are now Python 2.5 binaries as well. For SciPy there will be no 2.5 binaries - because 0.7.x is built against NumPy 1.2 it does not work on OS X 10.6, and on Windows I see some incomprehensible build failures. Binaries, sources and release notes can be found at http://www.filefactory.com/f/4452c50056df8bba/ (apologies for the flashy ads again). Cheers, Ralf

9 19

Numpy+SWIG with arrays in input and output of a function
by Antoine Delmotte 16 Apr '10

16 Apr '10

Dear Numpy and SWIG users, I am currently trying to use SWIG and Numpy to launch C++ codes from python. My C++ code takes an array as an input (as well as integers, but this will be my next problem...) and returns a different array (with different dimensions). I have managed to make very simple codes taking an array in input OR returning an array in output. However, I have not managed to develop a code with an array in input and returning a different array in output. Here is one of my attempts, which simply copies a vector (I know I could use inplace for this example, but this is not what I will need later since I want to return a different kind of array than what I have as an input): /* matrix.c */ #include <stdlib.h> #include <stdio.h> void vecOut(double* vec,int m,double* vec2,int n){ int i,j; double sum =0.0; for(i=0;i<m;i++){ vec2[i] = 2*vec[i]; } } /* matrix.h */ void vecOut(double* vec,int m,double* vec2,int n); /* matrix.i */ %module matrix %{ #define SWIG_FILE_WITH_INIT #include "matrix.h" %} %include "numpy.i" %init %{ import_array(); %} %apply (double* IN_ARRAY1, int DIM1, int* ARGOUT_ARRAY1, int DIM1) {(double* vec, int m, double* vec2, int n)}; %include "matrix.h" /* matrixSetup.py */ from distutils.core import setup, Extension import numpy setup(name='matrix', version='1.0', ext_modules =[Extension('_matrix', ['matrix.c','matrix.i'], include_dirs = [numpy.get_include(),'.'])]) So, this is not working. The compilation is fine but when I try to launch it, Python asks for 4 arguments. However, I would just like to put one... matrix.vecOut([1,2,3]) Traceback (most recent call last): TypeError: vecOut() takes exactly 4 arguments (1 given) I expected to get [1,2,3]... Is it possible to do this with SWIG? Or should I use something else? Any advice? I thank you in advance for your precious suggestions. Best regards, Antoine _________________________________________________________________ Hotmail : un service de messagerie gratuit, fiable et complet https://signup.live.com/signup.aspx?id=60969

2 2

Very good series of numpy tutorial videos here
by reckoner 16 Apr '10

16 Apr '10

FYI, there is a very good series of numpy tutorial videos here: http://showmedo.com/videotutorials/series?name=i9KuJuNcG It covers a wide range of topics, not just the basics. Cheers!

1 0

Find indices of largest elements
by Nikolaus Rath 16 Apr '10

16 Apr '10

Hello, How do I best find out the indices of the largest x elements in an array? Example: a = [ [1,8,2], [2,1,3] ] magic_function(a, 2) == [ (0,1), (1,2) ] Since the largest 2 elements are at positions (0,1) and (1,2). Best, -Niko -- »Time flies like an arrow, fruit flies like a Banana.« PGP fingerprint: 5B93 61F8 4EA2 E279 ABF6 02CF A9AD B7F8 AE4E 425C

8 18

simple rectangular grid
by Neal Becker 15 Apr '10

15 Apr '10

Is there a simple way to create a rectangular grid, like this loopy code: (This is for a QAM256 constellation, btw): import numpy as np u = np.arange (-7.5, 8) const = np.empty ((16,16), dtype=complex) for ix in range (16): for iy in range (16): const[ix, iy] = u[ix] + 1j*u[iy]

3 2

use of %extend
by Michel Dupront 15 Apr '10

15 Apr '10

Hello, With the following example, given in the documentation: struct Vector { double x,y,z; }; %extend Vector { Vector __add__(Vector *other) { Vector v; v.x = $self->x + other->x; v.y = $self->y + other->y; v.z = $self->z + other->z; return v; } }; one can do in python: >>> import example >>> v = example.Vector(2,3,4) >>> w = example.Vector(10,11,12) >>> print v+w Vector(12,14,16) >>> There is something more that I would like to do but I don't know how. I would like to be able to do: >>>> v1 = v + 2 >>>> v2 = 2 + v where v is a vector and 2 a number that would be added to each component of the vector. I would really appreciate any help on that. Many thanks in advance. _________________________________________________________________ Hotmail arrive sur votre téléphone ! Compatible Iphone, Windows Phone, Blackberry, … http://www.messengersurvotremobile.com/?d=Hotmail

2 2