Mailman 3 December 2005 - SciPy-User

Bug in scipy.io.fread?
by Paul Ray Dec. 29, 2005

Dec. 29, 2005

Hi, I've compiled scipy_core and scipy from the current (yesterday, Rev 1502) svn versions. I'm running Mac OS X 10.4.3, with a fink-installed python2.4. The new scipy builds and installs just fine, and all the unit tests I tried worked fine. However, I get a Bus Error from the following simple program: =================================================== import scipy import scipy.io print 'Opening test.dat for writing' outfile = open('test.dat', mode='wb') a = scipy.arange(1024.0,dtype='f') print 'Writing array' scipy.io.fwrite(outfile, 1024, a, 'f') print 'Closing file' outfile.close() print 'Opening test.dat for reading' infile = open('test.dat', mode='rb') print 'Reading chunk' tmpchunk = scipy.io.fread(infile, 1024, 'f') print 'Done :',tmpchunk[0:10] =================================================== Here is the run: > mcxr2 : 192>python2.4 freadtest.py > Opening test.dat for writing > Writing array > Closing file > Opening test.dat for reading > Reading chunk > Bus error The written file looks fine... > mcxr2 : 193>od -f test.dat |head > 0000000 0.000000e+00 1.000000e+00 2.000000e+00 > 3.000000e+00 > 0000020 4.000000e+00 5.000000e+00 6.000000e+00 > 7.000000e+00 > 0000040 8.000000e+00 9.000000e+00 1.000000e+01 > 1.100000e+01 > 0000060 1.200000e+01 1.300000e+01 1.400000e+01 > 1.500000e+01 > 0000100 1.600000e+01 1.700000e+01 1.800000e+01 > 1.900000e+01 > 0000120 2.000000e+01 2.100000e+01 2.200000e+01 > 2.300000e+01 > 0000140 2.400000e+01 2.500000e+01 2.600000e+01 > 2.700000e+01 > 0000160 2.800000e+01 2.900000e+01 3.000000e+01 > 3.100000e+01 > 0000200 3.200000e+01 3.300000e+01 3.400000e+01 > 3.500000e+01 > 0000220 3.600000e+01 3.700000e+01 3.800000e+01 > 3.900000e+01 Cheers, -- Paul -- Dr. Paul S. Ray E-mail: Paul.Ray(a)nrl.navy.mil Naval Research Laboratory WWW : http://xweb.nrl.navy.mil/ personnel/paulr/ Code 7655 Phone : (202) 404-1619 Washington, DC 20375 AIM : NRLPSR

3 3

Where is arrayobject.h supposed to be?
by Paul Ray Dec. 29, 2005

Dec. 29, 2005

Hi, I've compiled scipy_core and scipy from the current (yesterday, Rev 1502) svn versions. I'm running Mac OS X 10.4.3, with a fink-installed python2.4. I'd like to compile a C extension (ppgplot in particular) against the new scipy_core, and I'm wondering where arrayobject.h is supposed to get installed? Given an earlier comment on the list, it looks like #include "scipy/ arrayobject.h" should get it, which would imply to me that it would get put into /sw/include/python2.4/scipy/arrayobject.h, but that is not what happens. Here is the result of a "find": > mcxr2 : 17>find /sw -name arrayobject.h -print > /sw/include/python2.4/numarray/arrayobject.h > /sw/include/python2.4/Numeric/arrayobject.h > /sw/lib/python2.4/site-packages/scipy/base/include/scipy/arrayobject.h Is there a problem with the scipy_core installer? Shouldn't it create /sw/include/python2.4/scipy and populate it? Or, do I need to add /sw/lib/python2.4/site-packages/scipy/base/include/scipy to my include search path when building C extensions against the new scipy_core? Also, if anyone has successfully built ppgplot against the new scipy_core (it normally builds against Numeric or numarray), I'd love to know the code changes required. Cheers, -- Paul -- Dr. Paul S. Ray E-mail: Paul.Ray(a)nrl.navy.mil Naval Research Laboratory WWW : http://xweb.nrl.navy.mil/ personnel/paulr/ Code 7655 Phone : (202) 404-1619 Washington, DC 20375 AIM : NRLPSR

2 1

Does the new (Travis') scipy_core need Numeric?
by rcsqtc＠iiqab.csic.es Dec. 29, 2005

Dec. 29, 2005

Hi all, I may be asking a very simple question, but I could not find a clear explanation anyware on the scipy web or the new scipy_core web. I've installed Travis scipy_core 0.8.4, which is supposed to substitute the Numeric package (or that's what I understood). But scipy_core does not provide scipy.linalg, scipy.stats, etc which are still in the scipy distribution (scipy-0.3.2 in my case). Now, reading the documentation (and trying as well) it seems that I need Numeric to install scipy-0.3.2. So do I need to install Numeric AND scipy_core 0.8.4? Is this because the transition is not over yet? Maybe I misunderstood something. It's a pity that the documentation is split into two different pages: http://numeric.scipy.org/ and http://www.scipy.org/.Thanks for your consideration, Ramon

3 2

Manipulating Bit-Masked Data
by Rich Shepard Dec. 29, 2005

Dec. 29, 2005

I need to read bit-masked bytes from an Optical Mark Recognition card reader. I'm quite new to Python and haven't found any documentation on how to manipulate bits. NumPy has masked arrays, but that's a different situation. My rummaging around the SciPy wiki and a search of the mail list archives turned up nothing. I'd appreciate a pointer to documentation that will show me how to manipulate bit-masks in Python. Thanks, Rich -- Richard B. Shepard, Ph.D. | Author of "Quantifying Environmental Applied Ecosystem Services, Inc. (TM) | Impact Assessments Using Fuzzy Logic" <http://www.appl-ecosys.com> Voice: 503-667-4517 Fax: 503-667-8863

4 4

Example of power of new data-type descriptors.
by Travis Oliphant Dec. 27, 2005

Dec. 27, 2005

I'd like more people to know about the new power that is in scipy core due to the general data-type descriptors that can now be used to define numeric arrays. Towards that effort here is a simple example (be sure to use latest SVN -- there were a coupld of minor changes that improve usability made recently). Notice this example does not use a special "record" array subclass. This is just a regular array. I'm kind of intrigued (though not motivated to pursue) the possibility of accessing (or defining) databases directly into scipy_core arrays using the record functionality. # Define a new data-type descriptor >>> import scipy >>> dtype = scipy.dtypedescr({'names': ['name', 'age', 'weight'], 'formats': ['S30', 'i2', 'f4']}) >>> a = scipy.array([('Bill',31,260),('Fred',15,135)], dtype=dtype) # the argument to dtypedescr could have also been placed here as the argument to dtype >>> print a['name'] [Bill Fred] >>> print a['age'] [31 15] >>> print a['weight'] [ 260. 135.] >>> print a[0] ('Bill', 31, 260.0) >>> print a[1] ('Fred', 15, 135.0) It seems to me there are some very interesting possibilities with this new ability. The record array subclass adds an improved scalar type (the record) and attribute access to get at the fields: (e.g. a.name, a.age, and a.weight). But, if you don't need attribute access you can use regular arrays to do a lot of what you might need a record array to accomplish for you. I'd love to see what people come up with using this new facility. The new array PEP for Python basically proposes adding a very simple array object (just the basic PyArrayObject * of Numeric with a bare-bones type-object table) plus this new data-type descriptor object to Python and a very few builtin data-type descriptors (perhaps just object initially). This would basically add the array interface to Python directly and allow people to start using it generally. The PEP is slow going because it is not on my priority list right now because it is not essential to making scipy_core work well. But, I would love to have more people ruminating on the basic ideas which I think are crystallizing. Best wishes for a new year, -Travis Oliphant

2 4

output formatting of scipy arrays
by Andrew Jaffe Dec. 26, 2005

Dec. 26, 2005

Hi all- Sorry as usual if this is a faq, but is there any easy way to change the way arrays appear when printed (presumably, this means affecting the behavior of __str__?). Currently, I get things like [[ 1. 0.71375029 0.01215192 -0.90644004 0.02740346 0.66145675 0.85629017 0.96413003 0.65755463] [... when what I'd really prefer is something like [[ 1. 0.714 0.012 -0.906 0.027 0.661 0.856 0.964 0.658] [... Specifically, I'd like to get - less wide floating-point numbers - more entries per line Thanks, Andrew p.s. I wonder if the first part of the request is something that's possible to change more-or-less globally -- i.e. for all float -> string conversions -- in python?

2 2

scipy install on mac os x
by Evan Monroig Dec. 26, 2005

Dec. 26, 2005

Hi, summary: almost successful install on Tiger, only clapack/cblas/fftpack seem not to work properly I have already installed scipy on my linux box, and now I am trying also to use it on a Mac Os X Tiger. I followed the recent discussions on the mailing list, and tried to use them and here is my experience. 1) I used the python2.4.1 mac installer from here http://python.org/ftp/python/2.4.1/MacPython-OSX-2.4.1-1.dmg 2) I used the python2.4 fix for Tiger http://pythonmac.org/packages/TigerPython24Fix-r2.zip 3) I changed my path so that /usr/local/bin appears before /usr/bin (the objective being that launching "python" in the shell goes to python2.4 and not python2.3) 4) I used the 4 installers provided by on http://trichech.us/mac http://fisher.forestry.uga.edu/files/scipy_core-0.8.6.1693-py2.4-macosx10.4… http://fisher.forestry.uga.edu/files/scipy-0.4.3.1493-py2.4-macosx10.4.dmg http://fisher.forestry.uga.edu/files/matplotlib-0.85.1.cvs.dmg http://fisher.forestry.uga.edu/files/numarray-1.5.0.dmg One possible concern is that they were built for python2.4.2 and not python2.4.1 but I think this should not pose problems 5) I ran the tests in the command line: ---- EvanL:~ evanmonroig$ python Python 2.4.1 (#2, Mar 31 2005, 00:05:10) [GCC 3.3 20030304 (Apple Computer, Inc. build 1666)] on darwin Type "help", "copyright", "credits" or "license" for more information. >>> from scipy import * >>> test() [.....] Ran 1372 tests in 8.686s FAILED (failures=11) ---- I put all details in the attached files, but to sum up the failures are mostly due to: a) clapack module is empty b) cblas module is empty c) fftpack gives wrong results Any thoughts? Should I install the lapack and blas library by myself, or are they already on my system (it is likely so because I previously installed Octave)? Thanks, Evan

1 0

mixing scipy and Numeric
by Christian Kristukat Dec. 26, 2005

Dec. 26, 2005

Hi, again I've got problems with old code that relies on Numeric when imported in some other program that uses scipy_core. Numeric.log10 has problems with a type called <type 'float64_arrtype'>. A ValueError is raised when passing a variable of that type. The strange thing is that that variable has been created by pure Numeric code. Where does that type belong to? scipy or Numeric or plain python? What does the suffix _arrtype suggest since it seems to be an ordinary float? Regards, Christian

2 1

Packaging Changes to Core in SVN
by Travis Oliphant Dec. 26, 2005

Dec. 26, 2005

Hi folks and Merry Christmas, I hope this is the last of the major changes to the scipy core structure. After playing around with various alternative homes for the fft, linear algebra, and random number routines, I think (with Robert's suggestiong) we've found a permanent home for them under scipy.corefft scipy.corelinalg scipy.corerandom In these packages we will try to import the full scipy versions of any routines (fft and inv in particular) so that full scipy installs can have improved (but possibly difficult to install) versions of these basic routines. I am of a mind to get rid of lib out of scipy_core and move mtrand to scipy. Then move dotblas to base and be done with the lib directory in scipy_core. I'm beginning to see the wisdom in "flat is better than nested" If we can get the any wrinkles out of the new structure, I'd like to make another release after the new year so that the improved packaging can start getting used. Best, -Travis

2 3

Building scipy_core 0.6.1 and 0.8.4 from source fails on Windows XP Pro / cygwin
by George Christianson Dec. 25, 2005

Dec. 25, 2005

Good morning, I tried to build scipy_core versions 0.6.1 and 0.8.4 from SourceForge sources. My platform runs Windows XP version 2002, service pack 2, with a 4 processor 3GHZ pentium. I build under cygwin 1.5.18, gcc version 3.4.4, python 2.4.1. In each of the two versions, I unpacked the gzipped source files and ran 'python setup.py build. In eack case, the builds proceeded normally until this point: customize GnuFCompiler using build_ext building 'scipy.base.umath' extension compiling C sources gcc options: '-fno-strict-aliasing -DNDEBUG -g -O3 -Wall -Wstrict-prototypes' compile options: '-Ibuild/src/scipy/base/src -Iscipy/base/include -Ibuild/src/sc ipy/base -Iscipy/base/src -I/usr/include/python2.4 -c' gcc -shared -Wl,--enable-auto-image-base build/temp.cygwin-1.5.18-i686-2.4/build /src/scipy/base/src/umathmodule.o -L/usr/lib/python2.4/config -lpython2.4 -o bui ld/lib.cygwin-1.5.18-i686-2.4/scipy/base/umath.dll build/temp.cygwin-1.5.18-i686-2.4/build/src/scipy/base/src/umathmodule.o: In fun ction `UBYTE_multiply': ... (many undefined references ) /tmp/scipy_core-0.8.4/scipy/base/src/ufuncobject.c:528: undefined reference to ` _feclearexcept' build/temp.cygwin-1.5.18-i686-2.4/build/src/scipy/base/src/umathmodule.o: In fun ction `construct_reduce': /tmp/scipy_core-0.8.4/scipy/base/src/ufuncobject.c:528: undefined reference to ` _fetestexcept' /tmp/scipy_core-0.8.4/scipy/base/src/ufuncobject.c:528: undefined reference to ` _feclearexcept' collect2: ld returned 1 exit status Could somebody suggest where I should look to resolve these references? Thanks George Christianson

2 1