Mailman 3 April 2012 - NumPy-Discussion

YouTrack testbed
by Maggie Mari 19 Apr '12

19 Apr '12

Hello, everyone. I work with Travis at Continuum, and he asked me to setup a YouTrack server that everyone is welcome to play around with. There is a test project currently set up, with some fake tickets. Here is the address: http://ec2-107-21-65-210.compute-1.amazonaws.com:8011/issues It's running on an AWS micro instance, so it might be slow at the moment. Any feedback or comments would be welcome. Maggie

9 20

Casting rules - an awkward case
by Matthew Brett 18 Apr '12

18 Apr '12

Hi, I just wanted to point out a situation where the scalar casting rules can be a little confusing: In [113]: a - np.int16(128) Out[113]: array([-256, -1], dtype=int16) In [114]: a + np.int16(-128) Out[114]: array([ 0, -1], dtype=int8) This is predictable from the nice docs here: http://docs.scipy.org/doc/numpy/reference/generated/numpy.result_type.html but I offer it only as a speedbump I hit. On the other hand I didn't find it easy to predict what numpy 1.5.1 was going to do: In [31]: a - np.int16(1) Out[31]: array([127, 126], dtype=int8) In [32]: a + np.int16(-1) Out[32]: array([-129, 126], dtype=int16) As a matter of interest, what was the rule for 1.5.1? See you, Matthew

1 1

All of the PyData videos are now up at the Marakana site
by Fernando Perez 17 Apr '12

17 Apr '12

Hi folks, A number of you expressed interest in attending the PyData workshop last month and unfortunately we had very tight space restrictions. But thanks to the team at Marakana, who pitched in and were willing to film, edit and post videos for many of the talks, you can access them all here: http://marakana.com/s/2012_pydata_workshop,1090/index.html They are in 720p so you can actually read the terminals, though I think you have to click the YouTube link to be able to change the resolution. Enjoy! f

1 0

[Meta] Was: Re: Removing masked arrays for 1.7? (Was 1.7 blockers)
by Fernando Perez 17 Apr '12

17 Apr '12

[ Making a separate thread so the NA one can stay on topic, since I haven't actually followed the discussion well enough to contribute on the technical points ] On Tue, Apr 17, 2012 at 12:43 PM, Matthew Brett <matthew.brett(a)gmail.com> wrote: > The absurd over-statement is the following: > > " I'm afraid I have to disagree: you seem to be proposing an absolute, > 'zero-tolerance'-style policy against any off-list discussion. " Well, that's how I understood what you said; sorry if I mis-interpreted it, but that is indeed how I read your post. > The question - which seems to me to be sensible rational and > important - is how to get better at on-list discussion, and whether > taking this particular discussion mainly off-list is good or bad in > that respect. And on that we completely agree. I just think that in this case, we can try to give people a chance and see if things go better: things have changed since last summer and there's encouraging evidence to suggest that a solid, open discussion will take place this time around. I just propose we give that process a chance first... Best, f

1 0

adding a cut function to numpy
by Skipper Seabold 17 Apr '12

17 Apr '12

Hi, I have a pull request here [1] to add a cut function similar to R's [2]. It seems there are often requests for similar functionality. It's something I'm making use of for my own work and would like to use in statstmodels and in generating instances of pandas' Factor class, but is this generally something people would find useful to warrant its inclusion in numpy? It will be even more useful I think with an enum dtype in numpy. If you aren't familiar with cut, here's a potential use case. Going from a continuous to a categorical variable. Given a continuous variable [~/] [8]: age = np.random.randint(15,70, size=100) [~/] [9]: age [9]: array([58, 32, 20, 25, 34, 69, 52, 27, 20, 23, 51, 61, 39, 54, 39, 44, 27, 17, 29, 18, 66, 25, 44, 21, 54, 32, 50, 60, 25, 41, 68, 25, 42, 69, 50, 69, 24, 69, 69, 48, 30, 20, 18, 15, 50, 48, 44, 27, 57, 52, 40, 27, 58, 45, 44, 32, 54, 19, 36, 32, 55, 17, 55, 15, 19, 29, 22, 25, 36, 44, 29, 53, 37, 31, 51, 39, 21, 66, 25, 26, 20, 17, 41, 50, 27, 23, 62, 69, 65, 34, 38, 61, 39, 34, 38, 35, 18, 36, 29, 26]) Give me a variable where people are in age groups (lower bound is not inclusive) [~/] [10]: groups = [14, 25, 35, 45, 55, 70] [~/] [11]: age_cat = np.cut(age, groups) [~/] [12]: age_cat [12]: array([5, 2, 1, 1, 2, 5, 4, 2, 1, 1, 4, 5, 3, 4, 3, 3, 2, 1, 2, 1, 5, 1, 3, 1, 4, 2, 4, 5, 1, 3, 5, 1, 3, 5, 4, 5, 1, 5, 5, 4, 2, 1, 1, 1, 4, 4, 3, 2, 5, 4, 3, 2, 5, 3, 3, 2, 4, 1, 3, 2, 4, 1, 4, 1, 1, 2, 1, 1, 3, 3, 2, 4, 3, 2, 4, 3, 1, 5, 1, 2, 1, 1, 3, 4, 2, 1, 5, 5, 5, 2, 3, 5, 3, 2, 3, 2, 1, 3, 2, 2]) Skipper [1] https://github.com/numpy/numpy/pull/248 [2] http://stat.ethz.ch/R-manual/R-devel/library/base/html/cut.html

2 4

Segmentation fault during tests with Python 2.7.2 on Debian 6?
by Chris Ball 17 Apr '12

17 Apr '12

Hi, I'm trying out various continuous integration options, so I happen to be testing NumPy on several platforms that I don't normally use. Recently, I've been getting a segmentation fault on Debian 6 (with Python 2.7.2): Linux debian6-amd64 2.6.32-5-amd64 #1 SMP Thu Mar 22 17:26:33 UTC 2012 x86_64 GNU/Linux (Debian GNU/Linux 6.0 \n \l) nosetests --verbose /home/slave/tmp/numpy/numpy/random/__init__.py:91: RuntimeWarning: numpy.ndarray size changed, may indicate binary incompatibility from mtrand import * test_api.test_fastCopyAndTranspose ... ok test_api.test_array_astype ... ok test_api.test_copyto_fromscalar ... ok test_api.test_copyto ... ok test_api.test_copyto_maskna ... ok test_api.test_copy_order ... ok Basic test of array2string. ... ok Test custom format function for each element in array. ... ok This should only apply to 0-D arrays. See #1218. ... ok test_arrayprint.TestArrayRepr.test_nan_inf ... ok test_str (test_arrayprint.TestComplexArray) ... ok test_arrayprint.TestPrintOptions.test_basic ... ok test_arrayprint.TestPrintOptions.test_formatter ... ok test_arrayprint.TestPrintOptions.test_formatter_reset ... ok Ticket 844. ... ok test_blasdot.test_blasdot_used ... SKIP: Skipping test: test_blasdot_used Numpy is not compiled with _dotblas test_blasdot.test_dot_2args ... ok test_blasdot.test_dot_3args ... ok test_blasdot.test_dot_3args_errors ... ok test_creation_overflow (test_datetime.TestDateTime) ... ok test_datetime_add (test_datetime.TestDateTime) ... ok test_datetime_arange (test_datetime.TestDateTime) ... ok test_datetime_array_find_type (test_datetime.TestDateTime) ... ok test_datetime_array_str (test_datetime.TestDateTime) ... ok test_datetime_as_string (test_datetime.TestDateTime) ... ok test_datetime_as_string_timezone (test_datetime.TestDateTime) ... /home/slave/ tmp/numpy/numpy/core/tests/test_datetime.py:1319: UserWarning: pytz not found, pytz compatibility tests skipped warnings.warn("pytz not found, pytz compatibility tests skipped") ok test_datetime_busday_holidays_count (test_datetime.TestDateTime) ... ok test_datetime_busday_holidays_offset (test_datetime.TestDateTime) ... ok test_datetime_busday_offset (test_datetime.TestDateTime) ... ok test_datetime_busdaycalendar (test_datetime.TestDateTime) ... ok test_datetime_casting_rules (test_datetime.TestDateTime) ... ok test_datetime_divide (test_datetime.TestDateTime) ... ok test_datetime_dtype_creation (test_datetime.TestDateTime) ... ok test_datetime_is_busday (test_datetime.TestDateTime) ... ok test_datetime_like (test_datetime.TestDateTime) ... ok test_datetime_maximum_reduce (test_datetime.TestDateTime) ... ok test_datetime_minmax (test_datetime.TestDateTime) ... ok test_datetime_multiply (test_datetime.TestDateTime) ... ok test_datetime_nat_casting (test_datetime.TestDateTime) ... ok test_datetime_scalar_construction (test_datetime.TestDateTime) ... ok test_datetime_string_conversion (test_datetime.TestDateTime) ... ERROR test_datetime_subtract (test_datetime.TestDateTime) ... Segmentation fault With Python 2.6 there doesn't seem to be a problem on the same machine. Unfortunately, I haven't had time to investigate (I don't have Debian 6 to use myself, and I just started a job that doesn't involve any Python...). However, according to the Jenkins instance on ShiningPanda.com, the problem began with these changes: BUG: ticket #1578, Fix python-debug warning for python >= 2.7. STY: Small style fixes. For now, that's all I can say; I haven't manually verified the problem myself (that it exists, or that it truly started after the changes above). I hope to be able to investigate further at the weekend, but I thought I'd post to the list now in case someone else can verify the problem. Chris Segmentation fault is buried in console output of Jenkins: https://jenkins.shiningpanda.com/scipy/job/NumPy/PYTHON=CPython-2.7/6/conso… The previous build was ok: https://jenkins.shiningpanda.com/scipy/job/NumPy/PYTHON=CPython-2.7/5/conso… Changes that Jenkins claims are responsible: https://jenkins.shiningpanda.com/scipy/job/NumPy/PYTHON=CPython-2.7/6/ changes#detail0

3 7

Different behaviour of python built sum and addition on ndarrays
by Christopher Mutel 17 Apr '12

17 Apr '12

So, for both 1.5 and 1.6 (at least), it appears that the builtin sum does not add ndarrays the way "+" (and operator.add) do: a = np.arange(10).reshape((2,5)) b = np.arange(10, 20).reshape((2,5)) sum(a,b) Out[5]: array([[15, 18, 21, 24, 27], [20, 23, 26, 29, 32]]) a + b Out[6]: array([[10, 12, 14, 16, 18], [20, 22, 24, 26, 28]]) Is this expected? I couldn't find a description of why this would occur in the mailing list or in the documentation. I can't figure out what sum does at all, actually, as it doesn't seem to be a case of strange broadcasting or any other tricks I tried. Yours, Chris -- ############################ Chris Mutel Ökologisches Systemdesign - Ecological Systems Design Institut f.Umweltingenieurwissenschaften - Institute for Environmental Engineering ETH Zürich - HIF C 44 - Schafmattstr. 6 8093 Zürich Telefon: +41 44 633 71 45 - Fax: +41 44 633 10 61 ############################

2 1

setting the same object value with a mask?
by Thomas Tanner 16 Apr '12

16 Apr '12

Hi, is there an elegant method for assigning the same value to several indices in a ndarray? (in this case with dtype=object) example: a = empty(4,'O') # object ndarray x = [1,2,'f'] # the value to be set for some indicies - the value is not scalar a[array((True,False,True))] = x # works like put -> not what I want a[array((0,2))] = x # same effect print a # -> [1 None 2 None] a[0],a[2] = x,x # set explicitly - works print a # -> [[1, 2, 'f'] None [1, 2, 'f'] None] thanks for your help! cheers, -- Thomas Tanner ------ email: tanner(a)gmx.de GnuPG: 1024/5924D4DD

1 0

[f2py] f2py ignores 'fortranname' inside F90 modules?
by Irwin Zaid 16 Apr '12

16 Apr '12

Hi all, I've been having an issue with f2py simply ignoring the fortranname option if the Fortran subroutine is inside an F90 module. That option is useful for renaming Fortran subroutines. I don't know if this behaviour is to be expected, or if I am doing something wrong. I would definitely appreciate any help! As an example, here is code that correctly produces a Python module 'test' with a single Fortran subroutine 'my_wrapped_subroutine'. TEST_SUBROUTINE.F90 ------------------- subroutine my_subroutine() write (*,*) 'Hello, world!' end subroutine my_subroutine TEST_SUBROUTINE.PYF ------------------- python module test interface subroutine my_wrapped_subroutine() fortranname my_subroutine end subroutine my_wrapped_subroutine end interface end python module test But, when the Fortran subroutine 'my_subroutine' is placed inside a module, the fortranname option seems to be entirely ignored. The following example fails to compile. The error is "Error: Symbol 'my_wrapped_subroutine' referenced at (1) not found in module 'my_module'". TEST_MODULE.F90 --------------- module my_module contains subroutine my_subroutine() write (*,*) 'Hello, world!' end subroutine my_subroutine end module my_module TEST_MODULE.PYF --------------- python module test interface module my_module contains subroutine my_wrapped_subroutine() fortranname my_subroutine end subroutine my_wrapped_subroutine end module my_module end interface end python module test F2py is a great tool aside from this and a few other minor quibbles. So thanks a lot! Cheers, Irwin

1 0

Keyword argument support for vectorize.
by Michael McNeil Forbes 14 Apr '12

14 Apr '12

Hi, I added a simple enhancement patch to provide vectorize with simple keyword argument support. (I added a new kwvectorize decorator, but suspect this could/should easily be rolled into the existing vectorize.) http://projects.scipy.org/numpy/ticket/2100 This just reorders any kwargs into the correct position (filling in defaults as needed) and then class the "vectorize"d function. If people think this is reasonable, I can improve the patch with more comprehensive testing and error messages. Michael.

3 4