Mailman 3 May 2008 - NumPy-Discussion

Question for Travis
by Charles R Harris 30 May '08

30 May '08

Travis, What are the fundamental types for ndarrays? We have the c types, 'bBhHiIlLqQfdg', together with the boolean and complex types. Then we have types defined by length, int8, uint8, etc. The long types change length going from 32 to 64 bit machines, so there can be a couple of c-types corresponding to the same precision. So I am wondering which types should be considered "fundamental" and what types are used to identify pickled arrays and arrays stored with the new .nz formats. Chuck

2 2

question about histogram2d
by Darren Dale 30 May '08

30 May '08

I have a question about histogram2d. Say I do something like: import numpy from numpy import random import pylab x=random.rand(1000)-0.5 y=random.rand(1000)*10-5 xbins=numpy.linspace(-10,10,100) ybins=numpy.linspace(-10,10,100) h,x,y=numpy.histogram2d(x,y,bins=[xbins,ybins]) pylab.imshow(h,interpolation='nearest') pylab.show() The output is attached. I think I would have expected the transpose of what numpy histogram2d returned, so the tight x distribution appears along the x axis in the image. Maybe I am thinking about this incorrectly, or there is a convention I am unfamiliar with. If the behavior is correct, could the docstring include a comment explaining the orientation of the histogram array? Thanks, Darren -- Darren S. Dale, Ph.D. Staff Scientist Cornell High Energy Synchrotron Source Cornell University 275 Wilson Lab Rt. 366 & Pine Tree Road Ithaca, NY 14853 darren.dale(a)cornell.edu office: (607) 255-3819 fax: (607) 255-9001 http://www.chess.cornell.edu

7 11

Tutorials at Scipy 2008
by Fernando Perez 30 May '08

30 May '08

[ This is meant as a heads-up here, please keep the discussion on the SciPy user list so we can focus the conversation in one list only. ] Hi all, Travis Oliphant and myself have signed up to coordinate the tutorials sessions at this year's SciPy conference. Our tentative plan is described here: http://scipy.org/SciPy2008/Tutorials but it basically consists of holding in parallel: 1. A 2-day hands-on tutorial for beginners. 2. A set of 2 or 4 hour sessions on special topics. We need input from people on: - Do you like this idea? - If yes for #1, any suggestions/wishes? Eric Jones, Travis O and myself have all taught similar things and could potentially do it again, but none of us is trying to impose it. If someone else wants to do it, by all means mention it. The job could be split across multiple people once an agenda is organized. - For #2, please go to the wiki and fill in ideas for topics and/or presenters. We'll need a list of viable topics with actual presenters before we start narrowing down the schedule into something more concrete. Feel free to either discuss things here or to just put topics on the wiki. I find wikis to be a poor place for conversation but excellent for summarizing items. I'll try to update the wiki with ideas that arise here, but feel free to directly edit the wiki if you just want to suggest a specific topic or brief piece of info, do NOT feel like you have to vet anything on list. Cheers, Travis and Fernando.

2 2

ANN: NumPy 1.1.0
by Jarrod Millman 30 May '08

30 May '08

I'm pleased to announce the release of NumPy 1.1.0. NumPy is the fundamental package needed for scientific computing with Python. It contains: * a powerful N-dimensional array object * sophisticated (broadcasting) functions * basic linear algebra functions * basic Fourier transforms * sophisticated random number capabilities * tools for integrating Fortran code. Besides it's obvious scientific uses, NumPy can also be used as an efficient multi-dimensional container of generic data. Arbitrary data-types can be defined. This allows NumPy to seamlessly and speedily integrate with a wide-variety of databases. This is the first minor release since the 1.0 release in October 2006. There are a few major changes, which introduce some minor API breakage. In addition this release includes tremendous improvements in terms of bug-fixing, testing, and documentation. For information, please see the release notes: http://sourceforge.net/project/shownotes.php?release_id=602575&group_id=1369 Thank you to everybody who contributed to this release. Enjoy, -- Jarrod Millman Computational Infrastructure for Research Labs 10 Giannini Hall, UC Berkeley phone: 510.643.4014 http://cirl.berkeley.edu/

4 5

From float to records
by Pierre GM 29 May '08

29 May '08

All, I have a set of arrays that I want to transform to records. Viewing them as a new dtype is usually sufficient, but fails occasionally. Here's an example: #--------------------------------------- import numpy as np testdtype = [('a',float),('b',float),('c',float)] test = np.random.rand(15).reshape(5,3) # View the (5,3) array as 5 records of 3 fields newrecord = test.view(testdtype) # Create a new array with the wrong shape test = np.random.rand(15).reshape(3,5) #Try to view it try: newrecord = test.T.view(testdtype) except ValueError, msg: print "Error creating new record on transpose: %s" % msg # That failed, but won't with a copy try: newrecord = test.T.copy().view(testdtype) except ValueError, msg: print "Error creating new record on transpose+copy: %s" % msg #--------------------------------------- * Could somebody explain me what goes wrong in the second case (transpose+view) ? Is it because the transpose doesn't own the data ? * Is there a way to transform my (3,5) array into a (5,) recordarray without a copy ? Thanks a lot in advance.

2 2

Any and all NaNs
by Keith Goodman 29 May '08

29 May '08

I'm writing unit tests for a module that contains matrices. I was surprised that these are True: >> import numpy.matlib as mp >> x = mp.matrix([[mp.nan]]) >> x.any() True >> x.all() True My use case is (x == y).all() where x and y are the same matrix except that x contains one NaN. Certianly x and y are not equal. >> x = mp.asmatrix(range(4)).reshape(2,2) >> y = mp.asmatrix(range(4)).reshape(2,2) >> x[0,0] = mp.nan >> (x == y).all() True

4 7

out= corner cases
by Keith Goodman 29 May '08

29 May '08

This looks good: >> import numpy as np >> x = np.random.rand(2,3) >> x.mean(None, out=x) --------------------------------------------------------------------------- ValueError: wrong shape for output But this is strange: >> x.std(None, out=x) 0.28264369725 >> x array([[ 0.54718012, 0.94296181, 0.23668961], [ 0.35561918, 0.80860405, 0.96713833]]) >> >> x.var(None, out=x) 0.0798874595952 >> x array([[ 0.54718012, 0.94296181, 0.23668961], [ 0.35561918, 0.80860405, 0.96713833]]) >> >> x.var(0, out=x) array([ 0.0091739 , 0.004513 , 0.13338883]) >> x array([[ 0.54718012, 0.94296181, 0.23668961], [ 0.35561918, 0.80860405, 0.96713833]]) I'm using numpy 1.0.4 from Debian Lenny.

1 0

C API : slicing?
by Rahul Garg 29 May '08

29 May '08

Hi. Does the C api have some convenience functions for creating slices? For example : if I have a PyArrayObject *A, which represents lets say a 2d ndarray A in Python, is there a C api function to easily do the equivalent of A[a:b:c,d:e:f] ? thanks, rahul

1 0

Python-to-C compiler : very prelim alpha
by Rahul Garg 29 May '08

29 May '08

Hi. I had announced a project called Spyke earlier on this list. The name has been changed to unPython. unPython is a Python to C compiler. The homepage is at www.cs.ualberta.ca/~garg1/unpython/index.php A very prelim release is available for download. You can also download the manual on the download page. Note that this is a very preliminary release. If it compiles anything correctly, I will be very surprised :) The license currently is GPLv3. Please do not judge the compiler by what it is right now but instead please try it out and give me comments/flames/bug reports etc. I have also set up a mailing list at http://groups.google.com/group/unpython-discuss For those of you who do not like to play at the bleeding edge, a slightly better release is scheduled around mid June. thanks, rahul

1 0

List of function-like things with an 'out' parameter
by Alan McIntyre 29 May '08

29 May '08

On Wed, May 28, 2008 at 3:34 PM, Charles R Harris >> I wonder if this is something that ought to be looked at for all >> functions with an "out" parameter? ndarray.compress also had problems >> with array type mismatch (#789); I can't imagine that it's safe to >> assume only these two functions were doing it incorrectly. (Unless of >> course somebody has recently looked at all of them) > > I think that is an excellent idea! A good start would be to list all the > functions with the out parameter and then write some tests. The current > behavior is inconsistent and we not only need to specify the behavior, but > fix all the places that don't follow the rules. Here's a list of things in numpy that have an 'out' argument (and their arguments); I think I eliminated all the duplicated items (that are imported from a subpackage into one of the main packages, for example). There's stuff that's missing, probably; I think the C-implemented functions don't have argument lists programmatically available so I may parse their docstrings or something, but this is a start. numpy.all ['a', 'axis', 'out'] numpy.alltrue ['a', 'axis', 'out'] numpy.amax ['a', 'axis', 'out'] numpy.amin ['a', 'axis', 'out'] numpy.any ['a', 'axis', 'out'] numpy.around ['a', 'decimals', 'out'] numpy.choose ['a', 'choices', 'out', 'mode'] numpy.clip ['a', 'a_min', 'a_max', 'out'] numpy.compress ['condition', 'a', 'axis', 'out'] numpy.core.cumprod ['a', 'axis', 'dtype', 'out'] numpy.core.cumproduct ['a', 'axis', 'dtype', 'out'] numpy.core.cumsum ['a', 'axis', 'dtype', 'out'] numpy.core.defmatrix.matrix.all ['self', 'axis', 'out'] numpy.core.defmatrix.matrix.any ['self', 'axis', 'out'] numpy.core.fromnumeric.mean ['a', 'axis', 'dtype', 'out'] numpy.core.fromnumeric.prod ['a', 'axis', 'dtype', 'out'] numpy.core.fromnumeric.product ['a', 'axis', 'dtype', 'out'] numpy.core.fromnumeric.ptp ['a', 'axis', 'out'] numpy.core.fromnumeric.round_ ['a', 'decimals', 'out'] numpy.core.fromnumeric.sometrue ['a', 'axis', 'out'] numpy.core.fromnumeric.std ['a', 'axis', 'dtype', 'out', 'ddof'] numpy.core.fromnumeric.sum ['a', 'axis', 'dtype', 'out'] numpy.core.fromnumeric.take ['a', 'indices', 'axis', 'out', 'mode'] numpy.core.fromnumeric.trace ['a', 'offset', 'axis1', 'axis2', 'dtype', 'out'] numpy.core.fromnumeric.var ['a', 'axis', 'dtype', 'out', 'ddof'] numpy.lib.function_base.median ['a', 'axis', 'out', 'overwrite_input'] numpy.ma.choose ['indices', 't', 'out', 'mode'] numpy.ma.core.MaskedArray.compress ['self', 'condition', 'axis', 'out'] numpy.ma.core.max ['obj', 'axis', 'out'] numpy.ma.core.min ['array', 'axis', 'out'] numpy.ma.core.round_ ['a', 'decimals', 'out'] numpy.ma.extras.median ['a', 'axis', 'out', 'overwrite_input']

3 3