Mailman 3 July 2008 - NumPy-Discussion

Change in the representation of complex numbers in NumPy 1.1
by Francesc Alted 04 Jul '08

04 Jul '08

Hi, I've seen that NumPy has changed the representation of complex numbers starting with NumPy 1.1. Before, it was: >>> numpy.__version__ '1.0.3' >>> repr(numpy.complex(0)) # The Python type '0j' >>> repr(numpy.complex128(0)) # The NumPy type '0j' Now, it is: >>> numpy.__version__ '1.2.0.dev5313' >>> repr(numpy.complex(0)) '0j' >>> repr(numpy.complex128(0)) '(0.0+0.0j)' Not that I don't like the new way, but that broke a couple of tests of the PyTables suite, and before fixing it, I'd like to know if the new way would stay. Also, I'm not certain why you have chosen a different representation than the Python type. Thanks, -- Francesc Alted

2 11

Running doctests on buildbots
by Alan McIntyre 04 Jul '08

04 Jul '08

Hi all, It seems to me that the buildbots should be running the doctests in addition to the unit tests, but right now the buildbot test command "numpy.test(level=9999,verbosity=9999)" doesn't do that. We could either add "doctests=True" to the buildbot test command, or make "level" above some threshhold run them (currently the level argument is just ignored anyway). Since the level and verbosity parameters are going away in the release after 1.2, it seems more appropriate to just add the doctests argument. Any thoughts? Thanks, Alan P.S. Thanks to whoever installed nose on the buildbots!

3 6

Should we fix Ticket #709?
by Charles R Harris 04 Jul '08

04 Jul '08

That's Ticket #709 <http://projects.scipy.org/scipy/numpy/ticket/709>: > I'm faily sure that: > numpy.isnan(datetime.datetime.now() > ...should just return False and not raise an exception. > rationale: anything that is not nan should just return False, not raise > TypeErrors? <http://projects.scipy.org/scipy/numpy/wiki/TypeErrors> > <http://projects.scipy.org/scipy/numpy/ticket/709>Chuck

5 6

More doctest questions
by Alan McIntyre 04 Jul '08

04 Jul '08

Just a couple of quick questions for whomever knows: 1. Should we skip the numpy/f2py directory when looking for doctests? 2. Are the functions in numpy/lib/convdtype.py used anywhere? I can fix the doctests so they run, but I can't find anywhere they are used, so I wanted to see if they were intended for removal but forgotten. Thanks, Alan

2 3

Summer Doc Marathon status report and request for more writers
by Joe Harrington 04 Jul '08

04 Jul '08

This is an interim status report on the Summer Documentation Marathon. It is also an invitation and plea for all experienced users to participate! I am cross-posting in an effort to get broader participation. Please hold any discussion on the scipy-dev mailing list. As you know, our immediate goal is to produce first-draft docstrings for the user-visible parts of Numpy in time for a release before Fall classes (about 1 August). The short version is: We are really moving along! But, we need *your* help to make it in time for August. Here's the scoop: 1. We have all our infrastructure, standards, and procedure in place: We have a wiki that makes editing the docs easy and even fun. It communicates directly with the numpy sources. We have PDF and HTML reference guides being generated essentially automatically: http://sd-2116.dedibox.fr/pydocweb http://mentat.za.net/numpy/refguide/NumPy.pdf http://mentat.za.net/numpy/refguide/ The wiki front page contains or points to all you need to get started. The wiki lets you pull down a docstring with a few mouse clicks, edit on your machine, upload it, and see how it will look in its HTML version right away. You can also read everyone else's docstrings, comment on them, see the status of the project, and so on. The formatted versions necessarily lag the docstrings on the wiki because they are made whenever the docstrings are checked into the sources. 2. We have documented about 1/4 of numpy in a fairly professional way, comparable to the reference pages of the major commercial packages. The doc wiki is probably the next place to go if your question isn't answered by the docstring in the current version's help(), since you can look at the new docstrings we've generated, and they're *good*! 3. But, we're only 1/4 of the way there, we're halfway through the summer, and some of the initial enthusiasm is waning. The following page tells the tale: http://sd-2116.dedibox.fr/pydocweb/stats/ As you can see (you did click, right? please click...), there are 2323 numpy objects with docstrings. Of these, 1464 we deemed "unimportant" (for now). These are generally items not seen by regular users. This left 859 objects to document in this first pass. We've done 24% of them at this writing. Now, 24% is really exciting, and I'd like to take a moment to say a public "Hooray!" for the team (no particular order): Stéfan van der Walt Pauli Virtanen Robert Hetland Gael Varoquaux Scott Sinclair Alan Jackson Tim Cera Johann Cohen-Tanugi David Huard Keith Goodman Together these ten have written around 7500 words of documentation on the community's behalf, mainly as volunteers. HOWEVER, we can all do the math. We've spent one of our two months. We are 1/4 of the way there. Progress is slowing, and even if it didn't, we wouldn't make it in time. This is not a sprint, it's a MARATHON. WE NEED YOUR HELP! And we need it now. Are you excited by the idea of having documentation for numpy by the Fall release? Of having docs that answer your questions, that have *good* examples, that really save you time? If so, then please invest just a fraction of the time that documentation will save you in the next year alone by signing up on the wiki and writing some. If each experienced user wrote just a few pages, we'd be done! If you don't think you know enough to write, then do some reviewing. Are the docs readable? Do you understand the examples? Each docstring on the wiki has an easy comment box waiting for your thoughts. You will have a reference guide in the next release of numpy! I hope you will help make it a complete one. Sign up on the doc wiki today: http://sd-2116.dedibox.fr/pydocweb/ Thanks, --jh-- and the SciPy Doc Team Prof. Joseph Harrington Department of Physics MAP 414 4000 Central Florida Blvd. University of Central Florida Orlando, FL 32816-2385 (407) 823-3416 voice (407) 823-5112 fax (407) 823-2325 physics office jh(a)physics.ucf.edu

1 0

New test failure on SPARC buildbots.
by Charles R Harris 04 Jul '08

04 Jul '08

====================================================================== ERROR: test_ValidHTTP (test__datasource.TestDataSourceOpen) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/numpybb/Buildbot/numpy/b13/numpy-install/lib/python2.4/site-packages/numpy/lib/tests/test__datasource.py", line 81, in test_ValidHTTP assert self.ds.open(valid_httpurl()) File "../numpy-install/lib/python2.4/site-packages/numpy/lib/_datasource.py", line 394, in open File "../numpy-install/lib/python2.4/site-packages/numpy/lib/_datasource.py", line 252, in _findfile File "../numpy-install/lib/python2.4/site-packages/numpy/lib/_datasource.py", line 214, in _cache URLError: <urlopen error URL not found: http://www.google.com/index.html> It seems to be from the changes to speed up imports. Chuck

3 3

Numpy Array filling
by Brain Stormer 03 Jul '08

03 Jul '08

I am using numpy to create an array then filling some of the values using a for loop, I was wondering if there is way to easily fill the values without iterating through sort of like "array.fill[start:stop,start:stop]"? The reason for my question is, in some cases, I might have to fill hundreds (within a 10,000x10,000 matrix) of values and I am not sure if iteration is the right way to go. Code: from numpy import * x, y = 5, 5 matrix = zeros((y,x), int) print matrix fillxstart, fillystart = 1,1 fillxstop,fillystop = 4, 4 for i in range(fillystart,fillystop,1): for j in range(fillxstart,fillxstop,1): matrix[i,j] = 1 print matrix Output before filling: [[0 0 0 0 0] [0 0 0 0 0] [0 0 0 0 0] [0 0 0 0 0] [0 0 0 0 0]] Output after filling: [[0 0 0 0 0] [0 1 1 1 0] [0 1 1 1 0] [0 1 1 1 0] [0 0 0 0 0]]

3 2

Record arrays
by Stéfan van der Walt 03 Jul '08

03 Jul '08

Hi all, I am documenting `recarray`, and have a question: Is its use still recommended, or has it been superseded by fancy data-types? Regards Stéfan

11 18

slow import of numpy modules
by Nathan Jensen 03 Jul '08

03 Jul '08

Hi, I was wondering if there was any way to speed up the global import of numpy modules. For a simple import numpy, it takes ~250 ms. In comparison, importing Numeric is only taking 40 ms. It appears that even if you only import a numpy submodule, it loads all the libraries, resulting in the painful performance hit. Are there plans to speed up the importing of numpy, or at least have it not load libraries that aren't requested? Nathan

5 13

FFT's & IFFT's on images
by Mike Sarahan 03 Jul '08

03 Jul '08

Hi, I'm trying to do phase reconstruction on images which involves switching back and forth between Fourier space and real space. I'm trying to test numpy (& scipy, for that matter) just to see if I can go back and forth. After an FFT/iFFT, the resulting image is garbage. I'm using numpy.fft.fftn, but I've also tried fft2, rfftn, rfft2, and the corresponding inverse FFT's. >From looking at the matrices, it appears to be creating complex components that aren't in the matrix prior to any FFT's. Real fft's seem to add some small component to each value (<1). I'm using Image.fromarray to convert arrays to images, and I'm working with 8-bit grayscale images. Any help is much appreciated! Thanks, Mike Sarahan

5 9