Mailman 3 December 2009 - NumPy-Discussion

Can someone backport r7870 to numpy 1.4 branch?
by Travis Oliphant Dec. 4, 2009

Dec. 4, 2009

I just checked in another fix (for ticket #1254), but I don't have time to backport the fix to the 1.4 branch. It would be great if someone could do that. Thanks, -Travis

2 1

Re: [Numpy-discussion] Find the N maximum values and corresponding indexes in an array
by Howard Chong Dec. 3, 2009

Dec. 3, 2009

Thanks for all the help and suggestions. I think the partial sort is exactly what I need. I thought of doing it as a full sort with argsort(), but that would be much slower if I just need a small number (maybe 7) from a large array, potentially thousands or a million repeated many times. In case you are wondering, I am doing this to find the 7 nearest neighbors in a GIS. For a list of zipcodes in America, find the 7 nearest weather stations. > Message: 8 > Date: Wed, 02 Dec 2009 22:26:29 -0500 > From: Neal Becker <ndbecker2(a)gmail.com> > Subject: Re: [Numpy-discussion] Find the N maximum values and > corresponding indexes in an array > To: numpy-discussion(a)scipy.org > Message-ID: <hf7b56$jam$1(a)ger.gmane.org> > Content-Type: text/plain; charset="ISO-8859-1" > > Neal Becker wrote: > > > Keith Goodman wrote: > > ... > >> Oh, I thought he meant there was a numpy function for partial > sorting. > >> > Try this one: > > template<typename in_t> > inline void partial_sort (in_t in, int n_el) { > std::partial_sort (boost::begin (in), boost::begin(in) + n_el, > boost::end (in)); > } > ... > def ("partial_sort", > &partial_sort<pyublas::numpy_ > strided_vector<int32_t> >); > def ("partial_sort", > &partial_sort<pyublas::numpy_strided_vector<int64_t> >); > def ("partial_sort", > &partial_sort<pyublas::numpy_strided_vector<double> >); > --------- > import pyublas > import numpy as np > u = np.arange (20)[::-1] > from numpy_fncs import partial_sort > partial_sort (u, 4) > In [2]: u > Out[2]: > array([ 0, 1, 2, 3, 19, 18, 17, 16, 15, 14, 13, 12, 11, 10, 9, 8, > 7, > 6, 5, 4]) > > > -- > Howard Chong > Dept. of Agricultural and Resource Economics and Energy Institute @ Haas > Business School > UC Berkeley > hgchong(a)berkeley.edu > Cell: 510-333-0539 >

2 1

Python 3K merge
by Charles R Harris Dec. 3, 2009

Dec. 3, 2009

Hi Pauli, It looks like you doing great stuff with the py3k transition. Do you and David have any sort of merge schedule in mind? Chuck

6 7

Find the N maximum values and corresponding indexes in an array
by Howard Chong Dec. 3, 2009

Dec. 3, 2009

I will need to find the N largest numbers and corresponding indexes in an 1-D array. If N==1, I can easily do: def myFindMaxC(myList): """implement finding maximum value with using numpy.array()""" myA=np.array(myList) maxIndex=myA.argmax() maxVal=myA[maxIndex] return [maxIndex, maxVal] For me, I'm likely going to be running this with N==7. So, I think I have to iterate over the array. Doing it with non-numpy procedures is **quite slow**. Here, I run it with N==1 without using any numpy procedures. def myFindMaxA(myList): """implement finding maximum value with for loop iteration""" maxIndex=0 maxVal=myList[0] for index, item in enumerate(myList): if item[0]>maxVal: maxVal=item[0] maxIndex=index return [maxIndex, maxVal] My question is: how can I make the latter version run faster? I think the answer is that I have to do the iteration in C. If that's the case, can anyone point me to where np.array.argmax() is implemented so I can write np.array.argmaxN() extend it to the N largest values? Thanks! -- Howard Chong Dept. of Agricultural and Resource Economics and Energy Institute @ Haas Business School UC Berkeley hgchong(a)berkeley.edu Cell: 510-333-0539

8 21

SPARC Buildbot bus errors after recent commits
by Charles R Harris Dec. 3, 2009

Dec. 3, 2009

Hi Travis, I think this is yours ;)

1 0

ANN: qimage2ndarray - converting between QImages and numpy.ndarrays
by Hans Meine Dec. 2, 2009

Dec. 2, 2009

Hi, I have just uploaded a first release of qimage2ndarray, a tiny python extension for quickly converting between QImages and numpy.ndarrays (in both directions). These are very common tasks when programming e.g. scientific visualizations in Python using PyQt4 as the GUI library. Similar code was found in Qwt and floating around on mailing lists, but qimage2ndarray has the following unique feature set: * Supports conversion of scalar and RGB data, with arbitrary dtypes and memory layout, with and without alpha channels, into QImages (e.g. for display or saving using Qt). * Using a tiny C++ extension, qimage2ndarray makes it possible to create ndarrays that are *views* into a given QImage's memory. This allows for very efficient data handling and makes it possible to modify Qt image data in-place (e.g. for brightness/gamma or alpha mask modifications). * qimage2ndarray is stable and unit-tested: * proper reference counting even with views (ndarray.base points to the underlying QImage) * handles non-standard widths and respects QImage's 32-bit row alignment * Masked arrays are also supported and are converted into QImages with transparent pixels. * Supports value scaling / normalization to 0..255 for convenient display of arbitrary NumPy arrays. The extension is open source, BSD-licensed, and available via PyPI or here: http://kogs-www.informatik.uni-hamburg.de/~meine/software/qimage2ndarray/ I hope this is useful to many of you and look forward to your feedback, Hans PS: Now that I am announcing this, I suddenly have the feeling that I should have talked with some lawyer (or Phil) about possible license issues because of PyQt. I really hope there will not turn out to be problems with this.

3 2

Numpy 1.4.0 rc1 released
by David Cournapeau Dec. 2, 2009

Dec. 2, 2009

Hi, The first release candidate for 1.4.0 has been released. The sources, as well as mac and windows installers may be found here: https://sourceforge.net/projects/numpy/files/ The main improvements compared to 1.3.0 are: * Faster import time * Extended array wrapping mechanism for ufuncs * New Neighborhood iterator (C-level only) * C99-like complex functions in npymath As well as more than 50 bug fixes. The detailed list of changes may be found on trac: http://projects.scipy.org/numpy/roadmap cheers, David

11 16

a simple examplr showing numpy and matplotlib failing
by Yeates, Mathew C (388D) Dec. 2, 2009

Dec. 2, 2009

Click on "Hello World" twice and get a memory error. Comment out the ax.plot call and get no error. import numpy import sys import gtk from matplotlib.figure import Figure from matplotlib.backends.backend_gtkagg import FigureCanvasGTKAgg as FigureCanvas ax=None fig=None canvas=None def doplot(widget,box1): global ax,fig,canvas data=numpy.zeros(shape=(3508,125,129)) plot_data=data[0,0:,0] if canvas: box1.remove(canvas) canvas=None if ax: ax.cla() ax=None if fig: fig=None fig = Figure(figsize=(5,5), dpi=100) ax = fig.add_subplot(111) mif=numpy.arange(plot_data.shape[0]) #if the next line is commented out, all is good ax.plot(plot_data,mif) canvas = FigureCanvas(fig) box1.pack_start(canvas, True, True, 0) canvas.show() def delete_event(widget, event, data=None): return False window = gtk.Window(gtk.WINDOW_TOPLEVEL) window.connect("destroy", lambda x: gtk.main_quit()) box1 = gtk.HBox(False, 0) window.add(box1) button = gtk.Button("Hello World") box1.pack_start(button, True, True, 0) #window.add(box1) button.show() button.connect("clicked", doplot, box1) box1.show() window.set_default_size(500,400) window.show() gtk.main()

4 5

numpy 1.3.0 eggs with python2.6 seem broken on osx, and linux
by René Dudfield Dec. 2, 2009

Dec. 2, 2009

Hi, the other day I tried installing numpy with easy_install, and did not have much luck with python2.6. To reproduce, try installing with easy_install-2.6, or with buildout. The work around is to just install it in another way (eg, dmg or with setup.py install). cheers,

2 4

convert strides/shape/offset into nd index?
by Zachary Pincus Dec. 1, 2009

Dec. 1, 2009

Hi all, I'm curious as to what the most straightforward way is to convert an offset into a memory buffer representing an arbitrarily strided array into the nd index into that array. (Let's assume for simplicity that each element is one byte...) Does sorting the strides from largest to smallest and then using integer division and mod (in the obvious way) always work? (I can't seem to find a counterexample, but I am not used to thinking too deeply about bizarrely-strided configurations.) Is there a method that doesn't involve sorting? Thanks, Zach

4 5