Mailman 3 January 2008 - NumPy-Discussion

load movie frames in python?
by Brian Blais 30 Jan '08

30 Jan '08

Hello, Is there a way to read frames of a movie in python? Ideally, something as simple as: for frame in movie('mymovie.mov'): pass where frame is either a 2-D list, or a numpy array? The movie format can be anything, because I can probably convert things, but most convenient would be avi, mov, and flv (for youtube videos). thanks, Brian Blais -- Brian Blais bblais(a)bryant.edu http://web.bryant.edu/~bblais

3 5

a view of a numpy object array
by YW 30 Jan '08

30 Jan '08

Hi, How would one deal with the problem of getting a view of a numpy object array? For example, In [49]: b Out[49]: TaskResultArray([[TaskResult[ID:0]:{'out': 22}, TaskResult[ID:1]: {'out': 22}, TaskResult[ID:2]:{'out': 16}, TaskResult[ID:3]:{'out': 16}, TaskResult[ID:4]:{'out': 16}], [TaskResult[ID:5]:{'out': 16}, None, None, None, None], [None, None, None, None, None], [None, None, None, None, None], [None, None, None, None, None]], dtype=object) In [50]: b.shape Out[50]: (5, 5) In [58]: b[0,0]['out'] Out[58]: 21 What would be the best way without looping and copying, to get a view of this array, say c, so that we have In [69]: c Out[69]: array([[ 21., 22., 16., 16., 16.], [ 16., NaN, NaN, NaN, NaN], [ NaN, NaN, NaN, NaN, NaN], [ NaN, NaN, NaN, NaN, NaN], [ NaN, NaN, NaN, NaN, NaN]]) it is desired that c is a view instead of a copy so no extra care of updating c is needed. Is the tools that does this available in current numpy already? thanks in advance

1 0

a view of a numpy object array
by YW 30 Jan '08

30 Jan '08

Hi, How would one deal with the problem of getting a view of a numpy object array? For example, In [49]: b Out[49]: TaskResultArray([[TaskResult[ID:0]:{'out': 22}, TaskResult[ID:1]: {'out': 22}, TaskResult[ID:2]:{'out': 16}, TaskResult[ID:3]:{'out': 16}, TaskResult[ID:4]:{'out': 16}], [TaskResult[ID:5]:{'out': 16}, None, None, None, None], [None, None, None, None, None], [None, None, None, None, None], [None, None, None, None, None]], dtype=object) In [50]: b.shape Out[50]: (5, 5) In [58]: b[0,0]['out'] Out[58]: 21 What would be the best way without looping and copying, to get a view of this array, say c, so that we have In [69]: c Out[69]: array([[ 21., 22., 16., 16., 16.], [ 16., NaN, NaN, NaN, NaN], [ NaN, NaN, NaN, NaN, NaN], [ NaN, NaN, NaN, NaN, NaN], [ NaN, NaN, NaN, NaN, NaN]]) it is desired that c is a view instead of a copy so no extra care of updating c is needed. Is the tools that does this available in current numpy already? thanks in advance

1 0

Read-only mercurial mirror of numpy trunk available
by David Cournapeau 30 Jan '08

30 Jan '08

Hi there, I got a mercurial mirror of numpy available. I put some basic info on the wiki http://scipy.org/scipy/numpy/wiki/HgMirror I don't know mercurial so much, so I only put some instructions to get started. It will force me to get more familiar with the mercurial features which are different from bzr, too (I will put more info then; in particular, merging works a bit differently). Some info: - the import was done with hgsvn. - The mirror contains only the trunk, and not from the start: it looks like hgsvn does not follow svn copy, and as such, it started importing from the scipy_core -> numpy rename. Still, there is the whole trunk history from 2 years. - The .hg directory only takes 17 Mb. With the working tree, it takes 27 Mb, that is only 3 Mb less than a recent svn checkout. - for some reasons, when you clone the hg repository, the tags with svn revisions are lost. If anyone knows how to avoid that (it is really useful to find a hg revision corresponding to a svn revision) cheers, David

3 6

argsort memory problem?
by Oriol Vendrell 30 Jan '08

30 Jan '08

Hi all, I've noticed something that looks like an odd behaviour in array.argsort(). # test1 --------------------- from numpy import array while True: a=array([8.0,7.0,6.0,5.0,4.0,2.0]) i=a.argsort() # --------------------------- # test2 --------------------- from numpy import array a=array([8.0,7.0,6.0,5.0,4.0,2.0]) while True: i=a.argsort() # --------------------------- test1 runs out of memory after a few minutes, it seems that in each cycle some memory is allocated and never returned back. test2 runs fine until killed. I'm unsure if I'm missing something or if this could be a bug. I'm using numpy 1.0.1 with python 2.4.4 in a debian stable system. Any suggestions? Thanks, -- Oriol Vendrell ovendrell(a)gmail.com

5 6

preparing for 1.0.5 release
by Jarrod Millman 29 Jan '08

29 Jan '08

Hello, I am pushing back the NumPy 1.0.5 release by at least 1 week (http://projects.scipy.org/scipy/numpy/milestone/1.0.5). After speaking with Travis O., we would like to see the following branches merged this week: http://projects.scipy.org/scipy/numpy/browser/branches/maskedarray http://projects.scipy.org/scipy/numpy/browser/branches/build_with_scons/ Once these merges occur we would like to wait at least 1 week, before branching for the 1.0.5 release. This will give us an opportunity to see much wider testing of this work before making the new release. Please let me know if there is any reason that we should delay merging these two branches. -- Jarrod Millman Computational Infrastructure for Research Labs 10 Giannini Hall, UC Berkeley phone: 510.643.4014 http://cirl.berkeley.edu/

4 4

difference numpy/matlab
by lorenzo bolla 29 Jan '08

29 Jan '08

I noticed that: min([1+1j,-1+3j]) gives 1+1j in matlab (where for complex, min(abs) is used) but gives -1+3j in numpy (where lexicographic order is used) shouldn't this be mentioned somewhere in "Numpy for Matlab users" webpage? -- Lorenzo Bolla lbolla(a)gmail.com http://lorenzobolla.emurse.com/

4 4

Numpy
by matthew yeomans 29 Jan '08

29 Jan '08

Hello Dear Members of Numpy. I experienced some problems using numpy when i tried to compile it with py2exe. How this is done? Thanks and Regrads Matthew -- Kollox Ghal Xejn

1 0

major bug in fromstring, ascii mode
by Eric Firing 29 Jan '08

29 Jan '08

In the course of trying to parse ascii times, I ran into a puzzling bug. Sometimes it works as expected: In [31]:npy.fromstring('23:19:01', dtype=int, sep=':') Out[31]:array([23, 19, 1]) But sometimes it doesn't: In [32]:npy.fromstring('23:09:01', dtype=int, sep=':') Out[32]:array([23, 0]) In [33]:npy.__version__ Out[33]:'1.0.5.dev4742' In [34]:npy.fromstring('23:09:01', dtype=int, sep=':', count=3) Out[34]:array([23, 0, 16]) In [35]:npy.fromstring('23 09 01', dtype=int, sep=' ', count=3) Out[35]:array([23, 0, 9]) In [36]:npy.fromstring('23 09 01', dtype=int, sep=' ') Out[36]:array([23, 0, 9, 1]) Maybe it is a problem specific to int conversion; examples that fail with int work with float: In [37]:npy.fromstring('23 09 01', dtype=float, sep=' ') Out[37]:array([ 23., 9., 1.]) In [38]:npy.fromstring('23:09:01', dtype=float, sep=':') Out[38]:array([ 23., 9., 1.]) Eric

4 10

/usr/bin/xlf: 1501-210 command option NO_ATLAS_INFO=1 contains an incorrect subargument
by Hans-Joachim Ehlers 28 Jan '08

28 Jan '08

Given: scipy-0.6.4 numpy 1.0.4 python 2.5.1 AIX 5.3 xlc v8 xlf v10 essel 4.2 lapack 3.1.1 with CCI Problem: During the scipy build - python setyp.py build - the following error shows up: .... building 'scipy.lib.blas.fblas' extension compiling C sources C compiler: cc_r -DNDEBUG -O2 -qmaxmem=-1 compile options: '-DNO_ATLAS_INFO=1 -Ibuild/src.aix-5.3-2.5 -I/usr/local/lib/python2.5/site-packages/numpy/core/include -I/usr/local/include/python2.5 -c' compiling Fortran sources Fortran f77 compiler: /usr/bin/xlf -qextname -O5 Fortran f90 compiler: /usr/bin/xlf90 -qextname -O5 Fortran fix compiler: /usr/bin/xlf90 -qfixed -qextname -O5 compile options: '-DNO_ATLAS_INFO=1 -Ibuild/src.aix-5.3-2.5 -I/usr/local/lib/python2.5/site-packages/numpy/core/include -I/usr/local/include/python2.5 -c' xlf:f77: build/src.aix-5.3-2.5/build/src.aix-5.3-2.5/scipy/lib/blas/fblas-f2pywrappers.f /usr/bin/xlf: 1501-210 command option NO_ATLAS_INFO=1 contains an incorrect subargument /usr/bin/xlf: 1501-210 command option NO_ATLAS_INFO=1 contains an incorrect subargument >>>>>>>>>>>>>>>>>> error: Command "/usr/bin/xlf -qextname -O5 -DNO_ATLAS_INFO=1 -Ibuild/src.aix-5.3-2.5 -I/usr/local/lib/python2.5/site-packages/numpy/core/include -I/usr/local/include/python2.5 -c -c build/src.aix-5.3-2.5/build/src.aix-5.3-2.5/scipy/lib/blas/fblas-f2pywrappers.f -o build/temp.aix-5.3-2.5/build/src.aix-5.3-2.5/build/src.aix-5.3-2.5/scipy/lib/blas/fblas-f2pywrappers.o" failed with exit status 40 <<<<<<<<<<<<<<<<<< I found a reference regarding the IBM XLF compiler running on mac http://projects.scipy.org/pipermail/scipy-user/2004-May/002884.html where Pearu says: ... it turns out that -D/-U options to xlf compiler are not for defining cpp macros... The following information from the xlf Compiler says regarding the -D option : ... -D Specifies whether the compiler compiles fixed source form lines with a D in column 1 or treats them as comments. Please note that in order to pass C-style -D macro definitions to the C preprocessor (e.g. compiling a file that ends with .F) use the -W option. For example: -WF,-DDEFINE_THIS -D is the short form of -qdlines ... -U Makes the compiler case-sensitive to identifier names. By default the compiler interprets all names as if they were lower-case. Solution: Changeing the Command line from : /usr/bin/xlf -qextname -O5 -DNO_ATLAS_INFO=1 ....../fblas-f2pywrappers.o to /usr/bin/xlf -qextname -O5 -WF,-DNO_ATLAS_INFO=1 .... /fblas-f2pywrappers.o allowed the command line to succed without an error. But i do not know if the -WF,DNO_ATLAS_INFO=1 will change the program logic nor where to change ( in case the -WF, is the solution ) this permanently. The only reference i found regarding -DNO_ATLAS_INFO=1 was within the file: ../site-packages/numpy/distutils/system_info.py tia Hajo

1 0