Mailman 3 May 2013 - NumPy-Discussion

somewhat less stupid problem with 0-d arrays
by Neal Becker 11 May '13

11 May '13

It would be convenient if in arithmetic 0-d arrays were just ignored - it would seem to me to be convenient in generic code where a degenerate array is treated as "nothing" np.zeros ((0,0)) + np.ones ((2,2)) --------------------------------------------------------------------------- ValueError Traceback (most recent call last) <ipython-input-17-27af0e0bbc6f> in <module>() ----> 1 np.zeros ((0,0)) + np.ones ((2,2)) ValueError: operands could not be broadcast together with shapes (0,0) (2,2)

3 4

printing array in tabular form
by Sudheer Joseph 11 May '13

11 May '13

Dear experts, I need to print few arrays in a tabular form for example below array IL has 25 elements, is there an easy way to print this as 5x5 comma separated table? in python IL=[] for i in np.arange(1,bno+1): IL.append(i) print(IL) %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% in fortran I could do it as below %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% integer matrix(5,5) in=0 do, k=1,5 do, l=1,5 in=in+1 matrix(k,l)=in enddo enddo m=5 n=5 do, i=1,m write(*,"(5i5)") ( matrix(i,j), j=1,n ) enddo end *************************************************************** Sudheer Joseph Indian National Centre for Ocean Information Services Ministry of Earth Sciences, Govt. of India POST BOX NO: 21, IDA Jeedeemetla P.O. Via Pragathi Nagar,Kukatpally, Hyderabad; Pin:5000 55 Tel:+91-40-23886047(O),Fax:+91-40-23895011(O), Tel:+91-40-23044600(R),Tel:+91-40-9440832534(Mobile) E-mail:sjo.India@gmail.com;sudheer.joseph@yahoo.com Web- http://oppamthadathil.tripod.com ***************************************************************

7 19

0-dim arrays inconsistency
by Neal Becker 10 May '13

10 May '13

np.array ((0,0)) Out[10]: array([0, 0]) <<< ok, it's 2 dimensional In [11]: np.array ((0,0)).shape Out[11]: (2,) <<< except, it isn't

2 2

tests not running
by KACVINSKY Tom 10 May '13

10 May '13

Here is my set up: Mac OS 10.7.5 Xcode 4.5.1 Intel Fortran 12.1 Python 2.7.3 built from source Numpy 1.6.2 built from source, using MKL 11.0 nose 0.11.4 installed I run the numpy tests as documented (python -c 'import numpy; numpy.test()'), but get this output: tkacvins@macomsim> python -c 'import numpy; numpy.test()' Running unit tests for numpy NumPy version 1.6.2 NumPy is installed in /rd/gen/tky/do_not_delete/Python/macpython27/lib/python2.7/site-packages/numpy Python version 2.7.3 (default, Oct 10 2012, 14:47:52) [GCC 4.2.1 (Based on Apple Inc. build 5658) (LLVM build 2336.9.00)] nose version 0.11.4 ---------------------------------------------------------------------- Ran 0 tests in 0.025s OK Any diagnostics or options I can pass to the tests to see what is going on? This is rather odd, I thought the tests would run and possibly fail. I didn't expect 0 tests to run. Thanks, Tom This email and any attachments are intended solely for the use of the individual or entity to whom it is addressed and may be confidential and/or privileged. If you are not one of the named recipients or have received this email in error, (i) you should not read, disclose, or copy it, (ii) please notify sender of your receipt by reply email and delete this email and all attachments, (iii) Dassault Systemes does not accept or assume any liability or responsibility for any use of or reliance on this email. For other languages, go to http://www.3ds.com/terms/email-disclaimer

2 1

Name change of the ptp() function
by Eli Bressert 10 May '13

10 May '13

The peak-to-peak function, ptp(), is a simple and handy function to use, but it confuses the PEP20 suggestions a bit. See this discussion for reference: http://d.pr/i/snXG Based on the discussion, it may be a good idea to make the name more relevant? Some suggestions include span() or valuerange(). Among few others who have discussed this on Github (https://github.com/numpy/numpy/issues/3292#issuecomment-17353454) valuerange() appears to the best most favored one. -Eli

6 7

Scalar output from sub-classed Numpy array
by Thomas Robitaille 10 May '13

10 May '13

Hi everyone, I am currently trying to write a sub-class of Numpy ndarray, but am running into issues for functions that return scalar results rather than array results. For example, in the following case: import numpy as np class TestClass(np.ndarray): def __new__(cls, input_array, unit=None): return np.asarray(input_array).view(cls) def __array_finalize__(self, obj): if obj is None: return def __array_wrap__(self, out_arr, context=None): return np.ndarray.__array_wrap__(self, out_arr, context) I get: In [4]: a = TestClass([1,2,3]) In [5]: print type(np.dot(a,a)) <type 'numpy.int64'> In [6]: a = TestClass([[1,2],[1,2]]) In [7]: print type(np.dot(a,a)) <class '__main__.TestClass'> that is, in the case where the output is a scalar, it doesn't get wrapped, while in the case where the output is an array, it does. Could anyone explain this behavior to me, and most importantly, is there a way around this and have the above example return a wrapped 0-d TestClass array instead of a Numpy int64? Thanks, Tom

1 0

www.scipy.org down?
by KACVINSKY Tom 09 May '13

09 May '13

I am looking for a release of numpy for Mac OS, but I can't reach the scipy web server. IS it down for maintenance? Thanks, Tom This email and any attachments are intended solely for the use of the individual or entity to whom it is addressed and may be confidential and/or privileged. If you are not one of the named recipients or have received this email in error, (i) you should not read, disclose, or copy it, (ii) please notify sender of your receipt by reply email and delete this email and all attachments, (iii) Dassault Systemes does not accept or assume any liability or responsibility for any use of or reliance on this email. For other languages, go to http://www.3ds.com/terms/email-disclaimer

3 6

Integer type casting and OverflowError
by Robert Kern 09 May '13

09 May '13

With master numpy (and back to 1.6.1, at least): [~] |1> np.int32(3054212286) -1240755010 It seems like at one time, this used to raise an OverflowError. We can see this in at least one place in scipy: https://github.com/scipy/scipy/blob/master/scipy/interpolate/fitpack.py#L912 Is this a regression in numpy, or should the code in scipy be fixed to use a different test? Even if it is technically a regression, it might be one of those "it's been broken so long, it's a feature now" regressions. -- Robert Kern

3 5

Performance degradation when installing Numpy and Python not from packaging system?
by Bin Huang 08 May '13

08 May '13

Hi list, I have observed slowdown if Numpy and Python are installed manually (i.e. not from packaging system). The slowdown can be significant when problem size is large. I wish someone in this mailing list can give me some hints on this issue. I also post performance results and my source code here. My platform is a 12-core dual-socket SMP machine running Ubuntu (Linux kernel 3.2.0). On this platform, I have two sets of installations. One is pre-installed Python 2.7.3 and Numpy 1.6.1 installed with 'apt-get'. The other is Python 2.7.3 and Numpy 1.6.1 both compiled freshly from official tarball release by myself. In order to compare performance, I wrote a piece of code that simply performs matrix-matrix multiplication using one of Numpy's functions. The source code is listed as follows: *if __name__ == "__main__": import numpy, random, time import sys size = int(sys.argv[1]) # For simplicity, we only test square matrix matrix_a = numpy.matrix(numpy.random.randn(size,size)) matrix_b = numpy.matrix(numpy.random.randn(size,size)) start_time = time.time() result = numpy.dot(matrix_a, matrix_b) print '%0.3f ms' % ((time.time() - start_time)*1000.0)* I ran this code with both pre-installed Python and compiled Python. Here is performance numbers I got: Problem size Pre-installed Python Compiled Python/Numpy Slowdown 256x256 14.165 ms 44.538 ms ~3X 512x512 109.330 ms 347.143 ms ~3X 1024x1024 867.936 ms 8975.016 ms ~10X 2048x2048 7755.084 ms 84713.997 ms ~10X Here are commands I used to compile Python and Numpy: a) ~/python2.7.3$ ./configure --prefix=MY_LOCAL_DIR b) ~/python2.7.3$ make c) ~/python2.7.3$ make install e) ~/numpy-1.6.1$ MY_LOCAL_DIR/bin/python setup.py build f) ~/numpy-1.6.1$ MY_LOCAL_DIR/bin/python setup.py install Any ideas? Thanks in advance. Bin Huang

2 1

References to Numerical Recipes in Numpy code
by MACKEITH Andrew 08 May '13

08 May '13

The numpy libraries (numpy/core/machar.py, numpy/fft/info.py, numpy/lib/function_base.py) contain references to "Numerical Recipes: The Art of Scientific Computing". The book contains a licensing section which states "... Unlicensed transfer of Numerical Recipes programs to any other format is strictly prohibited". From this we understand that use of the book as a reference is permitted, but copying of code for distribution, whether non-commercially or otherwise is not permitted. We are wondering if anyone has done an analysis of this situation around numpy, and if so what conclusion they might have reached? Andrew MacKeith Python and Scripting Development Dassault Systemes Simulia Corp. This email and any attachments are intended solely for the use of the individual or entity to whom it is addressed and may be confidential and/or privileged. If you are not one of the named recipients or have received this email in error, (i) you should not read, disclose, or copy it, (ii) please notify sender of your receipt by reply email and delete this email and all attachments, (iii) Dassault Systemes does not accept or assume any liability or responsibility for any use of or reliance on this email. For other languages, go to http://www.3ds.com/terms/email-disclaimer

2 1