Mailman 3 November 2011 - SciPy-User

Numpy / Scipy build / test errors on Mac OS X Lion with Macports
by Christoph Deil 23 Nov '11

23 Nov '11

Hi, I would like to use numpy / scipy git master on Mac OS X Lion with XCode 4.2. The recommendation at http://www.scipy.org/Installing_SciPy/Mac_OS_X is to use the official Python distribution, but I already have a ton of other libraries and python packages (e.g. ipython 0.11 with working qtconsole) installed with the Macports python, and I guess it is not possible to mix packages installed with the Macports / official python? Would it be helpful if I file tickets with the build and test logs for numpy and scipy using the XCode and Macports compilers, or are these problems well-known and simply too hard to fix? I frequently had build problems with numpy / scipy also on other machines and would like to understand the issues better. E.g. does numpy need a Fortran compiler or is it all C/C++? Why are there often problems with gfortran versions/builds other than http://r.research.att.com/gfortran-lion-5666-3.pkg? Is the problem that the fortran compilers or the numpy / scipy libraries are non-standard-complient? Does anyone have a reference that explains building C/C++/Fortran Python extensions in general or specifically for numpy / scipy? Thanks! Christoph

4 5

Problems installing scipy 0.10.0 using a local installation of numpy 1.6.1
by David Pitchford 23 Nov '11

23 Nov '11

I am trying to install scipy on a lab computer for research at my university. I do not have root access to these machines, so when I need to install new python modules I usually install them locally to a folder I have control over (using python setup.py build), then add it to my PYTHONPATH variable. I got numpy 1.6.1 working this way, but when I try to install scipy 0.10.0 with my PYTHONPATH variable pointing to build/lib.linux-x86_64-2.6/ (relative to the top-level directory numpy is in, numpy-1.6.1), I get some errors that appear to involve files and directories that were expected and not found. Here is part of the output from my build command: umfpack_info: libraries umfpack not found in /usr/local/lib libraries umfpack not found in /usr/lib64 libraries umfpack not found in /usr/lib libraries umfpack not found in /opt/local/lib /export/scratch/pitch/trunk/pyserver/numpy-1.6.1/build/lib.linux-x86_64-2.6/numpy/distutils/system_info.py:460: UserWarning: UMFPACK sparse solver (http://www.cise.ufl.edu/research/sparse/umfpack/) not found. Directories to search for the libraries can be specified in the numpy/distutils/site.cfg file (section [umfpack]) or by setting the UMFPACK environment variable. warnings.warn(self.notfounderror.__doc__) NOT AVAILABLE non-existing path in 'scipy/spatial': '/export/scratch/pitch/trunk/pyserver/numpy-1.6.1/build/lib.linux-x86_64-2.6/numpy/core/include' non-existing path in 'scipy/spatial': '/export/scratch/pitch/trunk/pyserver/numpy-1.6.1/build/lib.linux-x86_64-2.6/numpy/core/include' non-existing path in 'scipy/special': '/export/scratch/pitch/trunk/pyserver/numpy-1.6.1/build/lib.linux-x86_64-2.6/numpy/core/include' non-existing path in 'scipy/special': '/export/scratch/pitch/trunk/pyserver/numpy-1.6.1/build/lib.linux-x86_64-2.6/numpy/core/include' Traceback (most recent call last): [Most of long stack trace omitted] File "/export/scratch/pitch/trunk/pyserver/numpy-1.6.1/build/lib.linux-x86_64-2.6/numpy/distutils/npy_pkg_config.py", line 309, in _read_config meta, vars, sections, reqs = parse_config(f, dirs) File "/export/scratch/pitch/trunk/pyserver/numpy-1.6.1/build/lib.linux-x86_64-2.6/numpy/distutils/npy_pkg_config.py", line 281, in parse_config raise PkgNotFound("Could not find file(s) %s" % str(filenames)) numpy.distutils.npy_pkg_config.PkgNotFound: Could not find file(s) ['/export/scratch/pitch/trunk/pyserver/numpy-1.6.1/build/lib.linux-x86_64-2.6/numpy/core/lib/npy-pkg-config/npymath.ini'] It seems like these files are supposed to be there, but they aren't, and I have no idea why. Is it possible to install scipy locally this way? If so, do I need to change something in my installation of numpy or PYTHONPATH so it finds the files? If not, I can ask my administrator to install the packages for me. (Which is a last resort, as I would expect it to take at least a week) -- -David Pitchford

2 1

importing griddata
by Bala subramanian 23 Nov '11

23 Nov '11

Friends, I have scipy version 0.8.0 installed in my system running with ubuntu. When i give 'import scipy' or 'import scipy.interpolate' i dnt get any problem. When i give from scipy.interpolate import griddata, i get an import error message. Kindly write why this happens. Python 2.7.1+ (r271:86832, Apr 11 2011, 18:05:24) [GCC 4.5.2] on linux2 Type "help", "copyright", "credits" or "license" for more information. >>> *import scipy* >>> *from scipy.interpolate import griddata* Traceback (most recent call last): File "<stdin>", line 1, in <module> ImportError: cannot import name griddata >>> *import scipy.interpolate* However in the scipy cookbook ( http://www.scipy.org/Cookbook/Matplotlib/Gridding_irregularly_spaced_data) importing the griddata is possible. Thanks, Bala

2 1

fmin_cg fmin_bfgs "Desired error not necessarily achieveddue to precision loss"
by Dan Stowell 23 Nov '11

23 Nov '11

Hi, I'm translating a fairly straightforward optimisation code example from octave. (Attached - it does a quadratic regression, with a tweaked regularisation function.) Both fmin_cg and fmin_bfgs give me poor convergence and this warning: "Desired error not necessarily achieveddue to precision loss" This is with various regularisation strengths, with normalised data, and with high-precision data (float128). Is there something I can do to enable these to converge properly? Thanks Dan (Using ubuntu 11.04, python 2.7.1, scipy 0.8) -- Dan Stowell Postdoctoral Research Assistant Centre for Digital Music Queen Mary, University of London Mile End Road, London E1 4NS http://www.elec.qmul.ac.uk/digitalmusic/people/dans.htm http://www.mcld.co.uk/

2 3

Basic bandpass filtering/fourier transform of live audio?
by Matthew Arceri 22 Nov '11

22 Nov '11

I'm a highschool senior working on an engineering project where I'm trying to make a 3D positioning system using microphones. The problem I'm having is that I can't find anyway to filter specific frequencies from a live microphone input. I've found a number of python libraries that seem able to achieve just that but they any examples I find are for far more complicated things. SciPi is the only one with a decent following that could possibly help me. So, to reiterate, I wanted to know how I could choose an audio port and look for a specific frequency being picked up on that port. (I'm using Ubuntu, Python 2.7.1, and the latest version of Numpy + Scipy)

2 1

Optimization, Matlab/Octave and Duplication
by Felipe Schneider 21 Nov '11

21 Nov '11

Hi all, I'm new to SciPy and even scientific computing in general, but I have some basic experience with Octave. Reading about Python and how it can be amazing but still very fast and all, I started to get some interest on SciPy/NumPy. I would like to ask some questions, hoping it's the right place to make them: 1. It seems Octave is really slow compared to the SciPy approach. I would like to know if it's due to some low level coding, Cython, etc. How's the procedure when someone wants to, say, create a new routine for SciPy? 2. Is there anything similar to Matlab's Toolboxes or Octave's Octave-Forge? Or is it all a huge pack? 3. I was searching for a LP solver and it seems SciPy doesn't have it! But there's cvxopt, am I wrong? So, there's no future plans on this area, I presume, i. e., no LP solver for SciPy? I would like to have a general answer (i. e., "when and when not should SciPy has this/that funtionality?"), which leads to my third question... 4. It seems that there are way more than one implemention in the Python world for a lot of things, am I wrong? How come? Why reinventing the wheel? Where does SciPy stand on this matter? Really thanks, Felipe.

7 7

interp1d results vs. MatLab interp1
by Lynn Oliver 21 Nov '11

21 Nov '11

I'm converting a MatLab program to Python, and I'm having problems understanding why scipy.interpolate.interp1d is giving different results than MatLab interp1. In MatLab the usage is slightly different: yi = interp1(x,Y,xi,'cubic') While in SciPy it's like this: f = interp1d(x,Y,kind='cubic') yi = f(xi) For a trivial example the results are the same: MatLab: interp1([0 1 2 3 4], [0 1 2 3 4],[1.5 2.5 3.5],'cubic') 1.5000 2.5000 3.5000 Python: interp1d([1,2,3,4],[1,2,3,4],kind='cubic')([1.5,2.5,3.5]) array([ 1.5, 2.5, 3.5]) But for a real-world example they are not the same: x = 0.0000e+000 2.1333e+001 3.2000e+001 1.6000e+004 2.1333e+004 2.3994e+004 Y = -6 -6 20 20 -6 -6 xi = 0.00000 11.72161 23.44322 35.16484 Matlab: -6.0000 -12.3303 -3.7384 22.7127 Python: -6. -15.63041012 -2.04908267 30.43054192 Any thoughts as to how I can get results that are consistent with MatLab? Thanks- Lynn

3 4

[job] two STScI positions
by Perry Greenfield 21 Nov '11

21 Nov '11

STScI has posted two positions in the Science Software Branch: astronomical applications developer: https://rn11.ultipro.com/SPA1004/jobboard/JobDetails.aspx?__ID=*B12AC582943… software distribution and installation support: https://rn11.ultipro.com/SPA1004/jobboard/JobDetails.aspx?__ID=*61BE073DB16…

1 0

mask an array using another array
by questions anon 20 Nov '11

20 Nov '11

I am trying to mask one array using another array. I have created a masked array using mask=MA.masked_equal(myarray,0), that looks something like: [1 - - 1, 1 1 - 1, 1 1 1 1, - 1 - 1] I have an array of values that I want to mask whereever my mask has a nan. how do I do this? I have looked at http://www.cawcr.gov.au/bmrc/climdyn/staff/lih/pubs/docs/masks.pdf but the command: d = array(a, mask=c.mask() does not work. I basically want to do exactly what that article does in that equation. Any feedback will be greatly appreciated.

1 0

test issues with 0.10
by Jean-Baptiste Marquette 19 Nov '11

19 Nov '11

Le 14 nov. 2011 à 20:54, Ralf Gommers a écrit : > I've run nosetests on my Mac (64-bit 10.7.2 build on EPD) which fails on the following test: > > test_definition (test_basic.TestDoubleIFFT) ... FAIL > test_definition_real (test_basic.TestDoubleIFFT) ... ok > test_djbfft (test_basic.TestDoubleIFFT) ... python(60968) malloc: *** error for object 0x105435b58: incorrect checksum for freed object - object was probably modified after being freed. > *** set a breakpoint in malloc_error_break to debug > Abort trap: 6 > > The exact same issue was just reported on scipy-user, thread "test issues with 0.10". Can you please move the follow-up over there? > > What compilers did you use? And do you know if EPD was built with the same compilers? Hi Ralf, My compiler is that coming with the latest version of Xcode: macprojb:workdir marquett$ gcc -v Using built-in specs. Target: i686-apple-darwin11 Configured with: /private/var/tmp/llvmgcc42/llvmgcc42-2336.1~1/src/configure --disable-checking --enable-werror --prefix=/Developer/usr/llvm-gcc-4.2 --mandir=/share/man --enable-languages=c,objc,c++,obj-c++ --program-prefix=llvm- --program-transform-name=/^[cg][^.-]*$/s/$/-4.2/ --with-slibdir=/usr/lib --build=i686-apple-darwin11 --enable-llvm=/private/var/tmp/llvmgcc42/llvmgcc42-2336.1~1/dst-llvmCore/Developer/usr/local --program-prefix=i686-apple-darwin11- --host=x86_64-apple-darwin11 --target=i686-apple-darwin11 --with-gxx-include-dir=/usr/include/c++/4.2.1 Thread model: posix gcc version 4.2.1 (Based on Apple Inc. build 5658) (LLVM build 2336.1.00) gfortran comes from the HPC site: macprojb:workdir marquett$ gfortran -v Utilisation des specs internes. COLLECT_GCC=gfortran COLLECT_LTO_WRAPPER=/usr/local/libexec/gcc/x86_64-apple-darwin10.7.0/4.6.0/lto-wrapper Target: x86_64-apple-darwin10.7.0 Configuré avec: ../gcc-4.6.0/configure --enable-languages=fortran Modèle de thread: posix gcc version 4.6.0 (GCC) Unfortunately I don't know very much about those used for EPD. I just got the 64-bit binary from the Enthought academic link. HTH, Cheers Jean-Baptiste

2 1