Mailman 3 September 2014 - NumPy-Discussion

Is this a bug?
by Charles R Harris Sept. 17, 2014

Sept. 17, 2014

Hi All, It turns out that gufuncs will broadcast the last dimension if it is one. For instance, inner1d has signature `(n), (n) -> ()`, yet In [27]: inner1d([1,1,1], [1]) Out[27]: 3 In [28]: inner1d([1,1,1], [1,1]) --------------------------------------------------------------------------- ValueError Traceback (most recent call last) <ipython-input-28-e53e62e35349> in <module>() ----> 1 inner1d([1,1,1], [1,1]) ValueError: inner1d: Operand 1 has a mismatch in its core dimension 0, with gufunc signature (i),(i)->() (size 2 is different from 3) I'd think this is a bug, as the dimensions should match. Note that scalar 1 will be promoted to [1] in this case. Thoughts? Chuck

6 24

Linear algebra functions on empty arrays
by Sebastian Berg Sept. 16, 2014

Sept. 16, 2014

Hey all, for https://github.com/numpy/numpy/pull/3861/files I would like to allow 0-sized dimensions for generalized ufuncs, meaning that the gufunc has to be able to handle this, but also that it *can* handle it at all. However lapack does not support this, so it needs some explicit fixing. Also some of the linalg functions currently explicitly allow and others explicitly disallow empty arrays. For example the QR and eigvals does not allow it, but on the other hand solve explicitly does (most probably never did, simply because lapack does not). So I am wondering if there is some convention for this, or what convention we should implement. Regards, Sebastian

3 4

Question about broadcasting vs for loop performance
by Ryan Nelson Sept. 15, 2014

Sept. 15, 2014

Hello all, I have a question about the performance of broadcasting versus Python for loops. I have the following sample code that approximates some simulation I'd like to do: ## Test Code ## import numpy as np def lorentz(x, pos, inten, hwhm): return inten*( hwhm**2 / ( (x - pos)**2 + hwhm**2 ) ) poss = np.random.rand(100) intens = np.random.rand(100) xs = np.linspace(0,10,10000) def first_try(): sim_inten = np.zeros(xs.shape) for freq, inten in zip(poss, intens): sim_inten += lorentz(xs, freq, inten, 5.0) return sim_inten def second_try(): sim_inten2 = lorentz(xs.reshape((-1,1)), poss, intens, 5.0) sim_inten2 = sim_inten2.sum(axis=1) return sim_inten2 print np.array_equal(first_try(), second_try()) ## End Test ## Running this script prints "True" for the final equality test. However, IPython's %timeit magic, gives ~10 ms for first_try and ~30 ms for second_try. I tried this on Windows 7 (Anaconda Python) and on a Linux machine both with Python 2.7 and Numpy 1.8.2. I understand in principle why broadcasting should be faster than Python loops, but I'm wondering why I'm getting worse results with the pure Numpy function. Is there some general rules for when broadcasting might give worse performance than a Python loop? Thanks Ryan

2 2

Tracking and inspecting numpy objects
by Mads Ipsen Sept. 15, 2014

Sept. 15, 2014

Hi, I am trying to inspect the reference count of numpy arrays generated by my application. Initially, I thought I could inspect the tracked objects using gc.get_objects(), but, with respect to numpy objects, the returned container is empty. For example: import numpy import gc data = numpy.ones(1024).reshape((32,32)) objs = [o for o in gc.get_objects() if isinstance(o, numpy.ndarray)] print objs # Prints empty list print gc.is_tracked(data) # Print False Why is this? Also, is there some other technique I can use to inspect all numpy generated objects? Thanks in advance. Best regards, Mads -- +---------------------------------------------------------+ | Mads Ipsen | +----------------------+----------------------------------+ | Gåsebæksvej 7, 4. tv | phone: +45-29716388 | | DK-2500 Valby | email: mads.ipsen(a)gmail.com | | Denmark | map : www.tinyurl.com/ns52fpa | +----------------------+----------------------------------+

4 7

Broadcasting with np.logical_and.reduce
by Antony Lee Sept. 12, 2014

Sept. 12, 2014

Hi, I thought that ufunc.reduce performs broadcasting, but it seems a bit confused by boolean arrays: <ipython with pylab mode on> In [1]: add.reduce([array([1, 2]), array([1])]) Out[1]: array([2, 3]) In [2]: logical_and.reduce([array([True, False], dtype=bool), array([True], dtype=bool)]) --------------------------------------------------------------------------- ValueError Traceback (most recent call last) <ipython-input-2-bedbab4c13e1> in <module>() ----> 1 logical_and.reduce([array([True, False], dtype=bool), array([True], dtype=bool)]) ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all() Am I missing something here? Thanks, Antony

3 7

Remove numpy/core/src/multiarray/testcalcs.py file
by Charles R Harris Sept. 12, 2014

Sept. 12, 2014

Hi All, There is an old file, numpy/core/src/multiarray/testcalcs.py, that looks like a forgotten leftover from the original 2009 datetime work. Is there any reason that this file should not be removed? Chuck

1 0

numpy.vectorize docstrings not shown by help command
by John Haiducek Sept. 12, 2014

Sept. 12, 2014

When I apply numpy.vectorize() to a function, documentation tools behave inconsistently with regard to the new, vectorized function. The function's __doc__ attribute does contain the docstring of the original function as expected, but the built-in help() command displays the documentation of the numpy.vectorize class, and sphinx-autodoc fails to display the function at all. Is there a way to get the docstring of the original function display everywhere as expected? For instance: >>> import numpy as np >>> def myfunc(x): ... "Square x" ... return x**2 ... >>> myfunc=np.vectorize(myfunc) >>> print myfunc.__doc__ Square x >>> help(myfunc) (displays documentation of np.vectorize)

2 2

@ operator
by Charles R Harris Sept. 12, 2014

Sept. 12, 2014

Hi All, I'm in the midst of implementing the '@' operator (PEP 465), and there are some behaviors that are unspecified by the PEP. 1. Should the operator accept array_like for one of the arguments? 2. Does it need to handle __numpy_ufunc__, or will __array_priority__ serve? 3. Do we want PyArray_Matmul in the numpy API? 4. Should a matmul function be supplied by the multiarray module? If 3 and 4 are wanted, should they use the __numpy_ufunc__ machinery, or will __array_priority__ serve? Note that the type number operators, __add__ and such, currently use __numpy_ufunc__ in combination with __array_priority__, this in addition to the fact that they are by default using ufuncs that do the same. I'd rather that the __*__ operators simply rely on __array_priority__. Thoughts? Chuck

6 16

why does u.resize return None?
by Neal Becker Sept. 11, 2014

Sept. 11, 2014

It would be useful if u.resize returned the new array, so it could be used for chaining operations -- -- Those who don't understand recursion are doomed to repeat it

3 3

Re: [Numpy-discussion] SFMT (faster mersenne twister)
by James A. Bednar Sept. 10, 2014

Sept. 10, 2014

| Date: Fri, 05 Sep 2014 13:19:57 -0400 | From: Neal Becker <ndbecker2(a)gmail.com> | | I think it's somewhat debatable whether generating a different | sequence of random numbers counts as breaking backward | compatibility. Please don't ever, ever break the sequence of numpy's random numbers! Please! We have put a lot of effort into being able to reproduce our published work exactly, and all of that would be in vain if the sequence changes. See e.g.: http://journal.frontiersin.org/Journal/10.3389/fninf.2013.00044/full We'd be very happy to see additional number generators appear *alongside* the existing ones, though, particularly if there are faster or otherwise better ones! Jim Bednar -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.

8 11