Mailman 3 January 2009 - NumPy-Discussion

error handling with f2py?
by Sturla Molden Jan. 16, 2009

Jan. 16, 2009

Is it possible to make f2py raise an exception if a fortran routine signals an error? If I e.g. have subroutine foobar(a, ierr) Can I get an exception automatically raised if ierr != 0? Sturla Molden

2 2

remove need for site.cfg on default
by Jarrod Millman Jan. 16, 2009

Jan. 16, 2009

Due to the fact that I was tired of adding site.cfg to scipy and numpy when building on Fedora and Ubuntu systems as well as a scipy ticket (http://scipy.org/scipy/numpy/ticket/985), I decided to try and add default system paths to numpy.distutils. You can find out more details on this ticket: http://projects.scipy.org/scipy/scipy/ticket/817 I would like to have as many people test this as possible. So I would like everyone, who has had to build numpy/scipy with a site.cfg despite having installed all the dependencies in the system default locations, to test this. If you would please update to the numpy trunk and build numpy and scipy without your old site.cfg. Regardless of whether it works or not, I would appreciate it if you could let me know. Thanks, Jarrod

3 3

Long integer 'in' arange()
by Rodney Stephenson Jan. 16, 2009

Jan. 16, 2009

There was some recent discussion about the (inefficiency) of 'in' when applied to a large xrange. I was wondering how numpy handled this and found the following >>> from numpy import * >>> u = arange(2**26) >>> 2**24 in u True >>> 2**31 in u False >>> 2**32 in u True Hmmmm... >>> u = arange(10, 2**26) >>> 1 in u False >>> 2**37 in u False >>> 0 in u False >>> u[0]=0 >>> 2**37 in u True >>> So I presume that in casting the long to int for the test is done by just taking the last 32 bits! Bug or feature? Python silently turns ints to longs when they overflow, so one should be careful...

1 0

limit function
by Neal Becker Jan. 15, 2009

Jan. 15, 2009

Is there a function to apply a limit to an array? I want to (efficiently) do: y = x if x < limit, otherwise limit

3 2

any news concerning numpy for python-2.6 on windows?
by Darren Dale Jan. 15, 2009

Jan. 15, 2009

Hello, I've been developing a numpy based package that I'd like to share with this list, but I need to test it on windows first. Is there any news concerning numpy installers for python-2.6 on windows, and is there any chance of installing on Vista? Thank you, Darren

4 4

Build errors
by Gael Varoquaux Jan. 15, 2009

Jan. 15, 2009

OK, here we go for the stupid questions showing that I really don't understaind building well: I am building numpy on a Mandriva x86_64. I built an optimized ATLAS, and added the relevant lines to the site.cfg so that numpy does find the libraries. But now I get the following error at build: /usr/bin/ld: /volatile/varoquau/usr/lib//libptcblas.a(cblas_dptgemm.o): relocation R_X86_64_32 against `a local symbol' can not be used when making a shared object; recompile with -fPIC /volatile/varoquau/usr/lib//libptcblas.a: could not read symbols: Bad value collect2: ld returned 1 exit status I must confess I really have no clue what this means, and how I solve this. Cheers, Gaël

2 2

How to handle fields less than one byte in a recarray
by Kim Hansen Jan. 15, 2009

Jan. 15, 2009

Hi Numpy forum Let me start out with a generic example: In [3]: test_byte_str = "".join(["\x12\x03", "\x23\x05", "\x35\x08"]) In [4]: desc = dtype({'names' : ["HIGHlow", "HIGH + low"], 'formats': [uint8, ui nt8]}) In [5]: r = rec.fromstring(test_byte_str, dtype=desc) In [6]: r[0] Out[6]: (18, 3) In [7]: r['HIGHlow'] Out[7]: array([18, 35, 53], dtype=uint8) In [8]: high = r['HIGHlow'] >> 4 In [9]: high Out[9]: array([1, 2, 3], dtype=uint8) In [10]: low = r['HIGHlow'] & 0x0F In [11]: low Out[11]: array([2, 3, 5], dtype=uint8) In [12]: high_plus_low = high + low In [13]: high_plus_low Out[13]: array([3, 5, 8], dtype=uint8) In [14]: r["HIGH + low"] Out[14]: array([3, 5, 8], dtype=uint8) So I have here an array, where the first byte actually represents two fields, HIGH and low, where HIGH are the four most significant bits and low are the four least significant bits. As you can see in the example I can figure out how to get an array of HIGH and low by postprocessing the recordarray, but I would really like to encapsulate this logic somehow inside the recordarray (or in an instance of derived class), such that I could access the arrays directly as r["HIGH"] and r["low"] or perhaps even better wrap it all up inside a class RWrapper, such that r = RWrapper(test_byte_str) r.low would work but not r.HIGHlow How do I implment that in a Pythonic way? My real application is considerebaly more complicated, but I think this describes the main problem.

1 0

Re: [Numpy-discussion] Complete LAPACK needed (Frank Lagor)
by Frank Lagor Jan. 14, 2009

Jan. 14, 2009

Thanks so much for your help, David. I'm sorry I did not receive your posts previously -- I have the digest mode on and there is a bit of a delay. I'll try to change my options next time I post a request. Thanks so much again, Frank

2 1

Trying to implement the array interface
by Mark Asbach Jan. 14, 2009

Jan. 14, 2009

Hi there, I'm currently extending the Python wrapper for the Open Computer Vision Library (opencv) with the goal to interface numerical libraries as seemless as possible. Unfortunately, it doesn't seem to be that easy ;-) What I've done so far: - Added an __array_interface__ property to the Python side of OpenCV data structures (matrices and images) that uses version 3 of the interface definition and supplies the keys 'version', 'shape', 'typestr', 'data' and in some cases 'strides' when we have non- continuos memory layout. I think, I'm compatible to http://numpy.scipy.org/array_interface.shtml . - Added parsing the __array_interface__ of Python objects passed to OpenCV methods. I'm a bit unsure of how to use the C/C++ side (array struct) and if I can expect it to be there (for example: I don't provide one with OpenCV). Since I intend to keep OpenCV independent of numpy, calling functions from numpy.h is not an option, as far as I can see. The stuff described above is in the head revision of OpenCV, accessible via "svn co https://opencvlibrary.svn.sourceforge.net/svnroot/opencvlibrary/trunk/opencv ". I've tried using the following packages with OpenCV this way: - numpy (1.0.4): everything works as expected. This is the most important library for OpenCV users, so this is a good sign. - pylab/matplotlib (0.91.2): seems to use numpy / scipy-core. Everything okay. - PIL (1.1.6): the array interface (Python side) doesn't adhere to the definition -> no 'version' key, 'data' is string, not a tuple holding the pointer. What to do with this? - Numeric (24.2): I can create arrays from OpenCV datatypes and print them. Converting to other types however always yields 'Cannot convert scalar to float' or 'a float is required'. Strange :-/ Numeric.array instances also don't carry an __array_interface__. I can however convert by using numpy.arrays as intermediate step. - Gnuplot (1.7): uses Numeric, so doesn't work as well - pymat: didn't check. Seems to use Numeric, test results cover Numeric 23 and Matlab 6.5 only, so this package might be dead? - numarray: didn't check. Is there still any relevance of this package? Best, Mark -- Mark Asbach Institut für Nachrichtentechnik, RWTH Aachen University http://www.ient.rwth-aachen.de/cms/team/m_asbach

5 4

PyExc_IOError
by Igor Sylvester Jan. 13, 2009

Jan. 13, 2009

Has anyone had problems with dynamically loading PyExc_IOError when embedding a cython module? While calling init below: *in main.c: *void init { Py_Initialize(); initmymodule(); do_somethind(); Py_Finalize(); } *in module.c:* cdef public void do_something(): import exceptions print exceptions.IOError # this prints <type 'exceptions.IOError'>, so I don't know what to make of the error below import numpy # exception is thrown here, and numpy library is in pythonpath Exception: Exception exceptions.ImportError: '/usr/lib/python2.5/lib-dynload/time.so: undefined symbol: PyExc_IOError' in 'mymodule.myfunction' ignored I'm not yet sure which library in numpy causes the exception but this example is easy to reproduce. I built it against *Python 2.5.1. compile:* gcc -c -I/usr/lib/python2.5/site-packages/numpy-1.0.4.0002-py2.5-linux-x86_64.egg/numpy/core/include -I/usr/bin/python/2.5.1/include/python2.5 -fPIC -fno-omit-frame-pointer -pthread -fexceptions -ansi -g main.c compile line for the other c-file is identical to main.c's *link:* gcc -g -pthread -shared main.o module.o -L/usr/bin/python/2.5.1/lib * -lpython2.5* -ldl -lm -lpthread -lutil -lstdc++ I can import the module from python without changing any environment variables. Thanks.

1 0