Mailman 3 January 2002 - NumPy-Discussion

Truncated Singular value decomposition
by Nils Wagner 28 Jan '02

28 Jan '02

Hi all, how can I obtain a truncated SVD (dyadic decomposition) of V,S,WT = singular_value_decomposition(Z) ? I am interested in a low rank approximation of Z. Thanks in advance. Nils

2 1

Matlab to Numpy
by Nils Wagner 28 Jan '02

28 Jan '02

Hi all, Who can send me a program written in Numpy (+Vpython for graphics), that reproduces the results of the following Matlab code x=linspace(0,1,25); t=linspace(0,2,50); [X,T] = meshgrid(x,t); z=exp(-abs((X-.5)*(T-1)))+sin(X.*T); subplot(3,2,1) surf(X,T,z) axis([0,1,0,2,0.4,2.1]) xlabel('x'),ylabel('t'),zlabel('z'),title('Actual surface') [u,s,v] = svd(z); for k =1:3 zz=u(:,1:k)*s(1:k,1:k)*v(:,1:k)'; subplot(3,2,k+1) surf(X,T,zz),axis([0,1,0,2,0.4,2.1]) xlabel('x'),ylabel('t'),zlabel('z') title(['Rank',num2str(k),' approximation']) Thanks in advance Nils

1 0

Character typecode
by Jochen Küpper 27 Jan '02

27 Jan '02

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 According to the manual there is a typecode 'Character'. But the following program gives an error: ,---- | import Numeric as num | data = num.array([1,2,3], num.Character) `---- ,---- | Traceback (most recent call last): | File "<stdin>", line 2, in ? | AttributeError: 'module' object has no attribute 'Character' `---- This is python-2.2, latest NumPy cvs. The following line fixes that. Any reason that isn't in there? Index: Lib/Precision.py =================================================================== RCS file: /cvsroot/numpy/Numerical/Lib/Precision.py,v retrieving revision 1.1.1.1 diff -u -r1.1.1.1 Precision.py - --- Lib/Precision.py 2000/01/13 21:23:06 1.1.1.1 +++ Lib/Precision.py 2002/01/27 15:30:50 @@ -33,6 +33,7 @@ return typecode raise PrecisionError, key+" of "+str(required_bits)+" bits not available on this system" +Character = 'c' UnsignedInt8 = 'b' try: Int0 = _lookup(_code_table, 'Integer', 0) Greetings, Jochen PS: The manual page http://pfdubois.com/numpy/html2/numpy-6.html is really messed up right at the spot where typecodes are explained. That is with Netscape-4.7 and Mozilla-0.9.7 on Linux and Win2000. PPS: What format is the manual master deocument in? Would it be feasible to generate info pages from that? - -- University of North Carolina phone: +1-919-962-4403 Department of Chemistry phone: +1-919-962-1579 Venable Hall CB#3290 (Kenan C148) fax: +1-919-843-6041 Chapel Hill, NC 27599, USA GnuPG key: 44BCCD8E -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.6-cygwin-fcn-1 (Cygwin) Comment: Processed by Mailcrypt and GnuPG <http://www.gnupg.org/> iD8DBQE8VB5XiJ/aUUS8zY4RAhIxAJ43vXeTvN0hlbVmCsSCGCGYrt2m/wCeOvH8 lodh8d0YmJy0eCqMyglNdj0= =z/Wo -----END PGP SIGNATURE-----

1 0

Column in IEEE trans Antennas and Propagation
by Rob 25 Jan '02

25 Jan '02

I've been asked to write a short column about Numpy and my web site in IEEE AP, for the Educational Corner. I still consider myself a Numpy neophyte, but perhaps thats the beauty of Python, being able to do neat things without being an expert. I am pondering what to write about on the Numpy side. I am soliciting any genious ideas for topics :) The first thing I thought of is that I just plain enjoy being able to tweek code without worrying about compilers. Second, the code ends up so much more readable than Fortran or C. Thirdly, vectorizing the ported code makes it look much more like the mathematical basis of the program. Lastly, I think its the perfect language for students learning algorithms. I'm open to any ideas. Thanks, Rob. -- The Numeric Python EM Project www.pythonemproject.com

1 0

memory-mapped Numeric arrays: arrayfrombuffer version 2
by Kragen Sitaker 25 Jan '02

25 Jan '02

I would be very happy if this got included in the Numpy distribution, so that people don't have to do an extra download to get this facility. I just submitted it as a patch on SourceForge, but my browser failed to upload the file. http://sourceforge.net/tracker/index.php?func=detail&aid=507847&group_id=13… There is one thing in arrayfrombuffer.c I'm not sure about, and I could use some help here. arrayobj is the return value from PyArray_FromDimsAndData: /* do I need to incref arrayobj?! fromstring does... */ return (PyObject*)arrayobj; } So, do I? Or not? It seems to work as it is. Following is the announcement I posted to comp.lang.python and (hopefully) comp.lang.python.announce. The 'arrayfrombuffer' package features support for Numerical Python arrays whose contents are stored in buffer objects, including memory-mapped files. This has the following advantages: - loading your array from a file is easy --- a module import and a single function call --- and doesn't use excessive amounts of memory. - loading your array is quick; it doesn't need to be copied from one part of memory to another in order to be loaded. - your array gets demand-loaded; parts you aren't using don't need to be in memory or in swap. - under memory-pressure conditions, your array doesn't use up swap, and parts of it you haven't modified can be evicted from RAM without the need for a disk write - your arrays can be bigger than your physical memory - when you modify your array, only the parts you modify get written back out to disk This is something that's been requested on the Numpy list a few times a year since 1999. arrayfrombuffer lives at http://pobox.com/~kragen/sw/arrayfrombuffer/ The current version is version 2; it is released under the X11 license (the BSD license without the advertising clause).

2 1

RE: memory-mapped Numeric arrays: arrayfrombuffer version 2
by Paul F. Dubois 25 Jan '02

25 Jan '02

I have verified that this package seems to work on Windows. I says seems only because I didn't try enough to uncover anything subtle. Unless or until we are convinced as a community that this is (a) the right way to do this and (b) that the package is portable, it would not be wise to put it in the main distribution. I would like to hear from the community about this so that I will know whether or not to add this package as a separate SourceForge 'package' within the Numerical Python area. Meantime I will add a link to the web page. -----Original Message----- From: python-announce-list-admin(a)python.org [mailto:python-announce-list-admin@python.org] On Behalf Of Kragen Sitaker Sent: Wednesday, January 23, 2002 9:40 PM To: python-announce-list(a)python.org Subject: memory-mapped Numeric arrays: arrayfrombuffer version 2 The 'arrayfrombuffer' package features support for Numerical Python arrays whose contents are stored in buffer objects, including memory-mapped files. This has the following advantages: - loading your array from a file is easy --- a module import and a single function call --- and doesn't use excessive amounts of memory. - loading your array is quick; it doesn't need to be copied from one part of memory to another in order to be loaded. - your array gets demand-loaded; parts you aren't using don't need to be in memory or in swap. - under memory-pressure conditions, your array doesn't use up swap, and parts of it you haven't modified can be evicted from RAM without the need for a disk write - your arrays can be bigger than your physical memory - when you modify your array, only the parts you modify get written back out to disk This is something that's been requested on the Numpy list a few times a year since 1999. arrayfrombuffer lives at http://pobox.com/~kragen/sw/arrayfrombuffer/ The current version is version 2; it is released under the X11 license (the BSD license without the advertising clause). <kragen(a)pobox.com> <P><A HREF="http://pobox.com/~kragen/sw/arrayfrombuffer/">arrayfrombuffer 2</A> - creates Numeric arrays from memory-mapped files. (23-Jan-02) -- http://mail.python.org/mailman/listinfo/python-announce-list

2 1

yacvsu
by Paul Dubois 23 Jan '02

23 Jan '02

yet-another-cvs-update Disregarding advice from wiser heads, I made setup.py generate rpm_install.sh so that it would execute setup.py with the "right" python. Works for me (TM) in the sense that the rpm builds. I'm afraid to test it because I don't want my system ruined. We need some words someplace about how one uses these rpms. I suppose the web site is the right place. Also new in this issue is an upgrade to MA.average to allow axial weights as Numeric now does. I replaced all memcpy's with memmove's due to a bug report. If someone thinks this is a hideous decision just let me know why.

1 0

cvs checkins for rpms
by Paul F. Dubois 21 Jan '02

21 Jan '02

Version 21.0a2 is now in cvs awaiting community testing. Especially to be tested are rpm creation, Windows installer creation. Includes new files setup.cfg and rpm_install.sh from Gerard Vermeulen.

2 2

Please help me test a new setup.py
by Paul F. Dubois 18 Jan '02

18 Jan '02

Working from something given to me by Gerard Vermeulen I have tried to improve Numeric's setup.py. This is now in CVS. I could use help testing it, particularly by the those whose brains are RPM-enabled. Those Numpy developers who have not done a damn thing lately are granted this chance for redemption. (:-> There is no more setup_all.py; rather, setup.py builds everything. It has, however, a structure that enables easy commenting-out of any optional packages. I have made and tested the Windows packages and installer; and made and tested Linux packages and have made but not tested the Linux RPM. (sh makedist.sh). If trying this on a Python will Numeric already installed, remove first from site-packages: Numeric.pth, Numeric, FFT, RNG, MA, kinds, PropertiedClasses Under the new scheme, Numeric.pth points to Numeric and the directories for the others are INSIDE Numeric. Everything should be backward compatible. I like this; it reduces the visible pollution in site-packages. The RPM produced does not contain the documentation (that runs into a distutils bug in older Pythons). It does not have a name beginning with python- because that screwed up something else. In short, it is a compromise. Once we're all convinced that what I have works then we can worry about the finer stuff. A nice side benefit is that there is now only one Windows .exe installer, and only one thing to uninstall in add/remove programs.

2 1

Re: viewing Gridded or Polar data?
by Rob Hooft 17 Jan '02

17 Jan '02

I wrote a program to display X-ray diffraction images. The display engine is available as "dis.py" from http://starship.python.net/crew/hooft/ Regards, Rob Hooft -- Rob W.W. Hooft || rob(a)hooft.net || http://www.hooft.net/people/rob/

1 0