Mailman 3 April 2004 - NumPy-Discussion

PyArray_ContiguousFromObject & PyDECREF
by Michiel Jan Laurens de Hoon 17 May '11

17 May '11

Dear pythoneers, In Chapter 12 ("Writing a C extension") in the Numerical Python manual, in the first example PyArray_ContiguousFromObject is used together with PyDECREF, while in the last example PyArray_ContiguousFromObject is used but the arrays are not PyDECREF'ed before the function returns. From Chapter 13 ("C API Reference") it seems that a PyDECREF is needed in both cases. Or am I missing something here? Is PyDECREF not needed for some reason in the last example? Thanks in advance, Michiel de Hoon U Tokyo First example: ============== static PyObject * trace(PyObject *self, PyObject *args) { PyObject *input; PyArrayObject *array; double sum; int i, n; if (!PyArg_ParseTuple(args, "O", &input)) return NULL; array = (PyArrayObject *) PyArray_ContiguousFromObject(input, PyArray_DOUBLE, 2, 2); if (array == NULL) return NULL; n = array->dimensions[0]; if (n > array->dimensions[1]) n = array->dimensions[1]; sum = 0.; for (i = 0; i < n; i++) sum += *(double *)(array->data + i*array->strides[0] + i*array->strides[1]); Py_DECREF(array); return PyFloat_FromDouble(sum); } Last example: ============= static PyObject * matrix_vector(PyObject *self, PyObject *args) { PyObject *input1, *input2; PyArrayObject *matrix, *vector, *result; int dimensions[1]; double factor[1]; double real_zero[1] = {0.}; long int_one[1] = {1}; long dim0[1], dim1[1]; extern dgemv_(char *trans, long *m, long *n, double *alpha, double *a, long *lda, double *x, long *incx, double *beta, double *Y, long *incy); if (!PyArg_ParseTuple(args, "dOO", factor, &input1, &input2)) return NULL; matrix = (PyArrayObject *) PyArray_ContiguousFromObject(input1, PyArray_DOUBLE, 2, 2); if (matrix == NULL) return NULL; vector = (PyArrayObject *) PyArray_ContiguousFromObject(input2, PyArray_DOUBLE, 1, 1); if (vector == NULL) return NULL; if (matrix->dimensions[1] != vector->dimensions[0]) { PyErr_SetString(PyExc_ValueError, "array dimensions are not compatible"); return NULL; } dimensions[0] = matrix->dimensions[0]; result = (PyArrayObject *)PyArray_FromDims(1, dimensions, PyArray_DOUBLE); if (result == NULL) return NULL; dim0[0] = (long)matrix->dimensions[0]; dim1[0] = (long)matrix->dimensions[1]; dgemv_("T", dim1, dim0, factor, (double *)matrix->data, dim1, (double *)vector->data, int_one, real_zero, (double *)result->data, int_one); return PyArray_Return(result); } -- Michiel de Hoon, Assistant Professor University of Tokyo, Institute of Medical Science Human Genome Center 4-6-1 Shirokane-dai, Minato-ku Tokyo 108-8639 Japan http://bonsai.ims.u-tokyo.ac.jp/~mdehoon

3 2

add.reduce for character arrays
by Faheem Mitha 10 May '04

10 May '04

Hi, The following does not work. I am trying to do something similar to add.reduce, except that I want strings concatenated across the appropriate axes. I was hoping that Python would "magically" recognise that addition in strings translated to concatenation. Incidentally, the error message is not terribly imformative. Is there any way to make this work, or any other method people can suggest, or do I need to write my own function? Thanks. Regards, Faheem. ********************************************************************** In [43]: s Out[43]: CharArray([['a', 'b', 'c'], ['d', 'e', 'f'], ['g', 'h', 'i']]) In [44]: numarray.add.reduce(s,1) --------------------------------------------------------------------------- error Traceback (most recent call last) /home/faheem/wc/corrmodel/boost/<console> error: getShape: sequence object nested more than MAXDIM

2 4

rationalizing functions that return number of elements
by Russell E Owen 29 Apr '04

29 Apr '04

I'm hoping we might rationalize the names of functions that return the number of elements in an array and fix the associated help strings. At present we have: numarray function size(array, axis=None) Seems to return # of elements in array, or in a particular axis, but its help says something else: "size returns the size in bytes of an array." numarray.ma has function size(a, axis=None) Seems the same as numarray.size but the help is wrong, and in a different way: "returns shape tuple for 'a'. MaskedArray version." nummarray.ma arrays have method "size(seolf, axis=None): The same as the others and correctly documented: "Number of elements in array, or in a particular axis." numarray arrays have method nelements() The same as "size" above, but has no axis argument: "nelements() -> total number of elements in array." I strongly encourage you to settle on one name (and one argument list, i.e. with "axis"). The current situation forces one to treat ma arrays differently than normal arrays in an area where there's no need for such a difference. It is also confusing. "size" is a perfectly good name and is already used in 3 out of 4 cases, so adopting it will break the least existing code and deprecate only one odd method. On the other hand, "nelements" has the advantage that it is less ambiguous (based on the incorrect help strings). So if you really think it's worth the pain you could switch to that. My personal leaning is towards "size", but it's not my decision. By the way, "itemsize" already exists as an array method, so "size" is NOT needed for that (and would be a wretched choice--ambigious and the potential to mysteriously break existing code). -- Russell P.S. I reported "nelements" as PR 934514 <http://sourceforge.net/tracker/index.php?func=detail&aid=934514&group_id=13…> and the incorrect help strings as PR 944678 <http://sourceforge.net/tracker/index.php?func=detail&aid=944678&group_id=13…> P.P.S. here's a log showing the tests that convinced me size works as I said, not as the help strings say: >>> a = num.array([[1.1, 2.2, 3.3], [4.4, 5.5, 6.6]]) >>> a array([[ 1.1, 2.2, 3.3], [ 4.4, 5.5, 6.6]]) >>> numarray.size(a) 6 >>> numarray.size(a, 1) 3 >>> aa = numarray.ma.array(a) >>> numarray.ma.size(aa) 6 >>> numarray.ma.size(aa, 1) 3 >>> aa.size() 6 >>> aa.size(1) 3 >>> a.nelements() 6

1 0

numarray on 64 bits
by Marc Poinot 29 Apr '04

29 Apr '04

Did anybody use numarray on a 64 bits platform ? (e.g. SGI/Irix) The package works on our 64 bits platforms... using 32 bits mode, but not using 64 mode. Big crash in _ufunc... Is there a flag to set somewhere, trick, hint, noway ? -MP- ----------------------------------------------------------------------- Marc POINOT Alias: marcvs Email: poinot(a)onera.fr ONERA -MFE/DSNA/ELSA Tel: 01.46.73.42.84 Info: elsa-info(a)onera.fr 29, Div. Leclerc Fax: 01.46.73.41.66 Site: 92322 Chatillon FRANCE Project: elsA Web: http://www.onera.fr

2 1

Your message to SciPy-user awaits moderator approval
by scipy-user-bounces＠scipy.net 28 Apr '04

28 Apr '04

Your mail to 'SciPy-user' with the subject RE: Incoming Msg Is being held until the list moderator can review it for approval. The reason it is being held: Post by non-member to a members-only list Either the message will get posted to the list, or you will receive notification of the moderator's decision. If you would like to cancel this posting, please visit the following URL: http://www.scipy.net/mailman/confirm/scipy-user/8c8e42fd74bfe658b790f95e8d6…

1 0

Einladung
by Das deutsche Panel 27 Apr '04

27 Apr '04

Können Sie uns Ihre Meinung mitteilen ? Das deutsche Panel ist eine Marktforschungswebsite, die mit dem Ziel errichtet worden ist, die Bevölkerung Deutschlands in Bezug auf verschiedene Bereiche, wie z.B. neue Fernsehwerbung zu befragen. Wenn Sie daran interessiert sind, Ihre Meinungen zu veröffentlichen, füllen Sie bitte unseren Fragebogen aus, bei : http://www.deutsches-panel.com/rec/index.php?id=numpy-discussion@lists.sour… Wenn Sie unser Panel besuchen und an unseren Studien teilnehmen, werden Sie automatisch bei einer Verlosung teilnehmen. Je nach Umfang der Studie, können Sie verschiedene Objekte im Wert zwischen 200 und 500 Euro gewinnen. So haben Sie die Möglichkeit verschiedene Objekte wie Playstations, Handies oder computerverbundene Geräte zu gewinnen. Falls Sie sich entscheiden den Wert des Objektes einer Hilfeorganisation zu spenden, werden wir Ihnen eine Liste mit unseren Partnerorganisationen zur Verfügung stellen. Dadurch ermöglichen wir Ihnen die Wahl einer Organisation, der Sie spenden wollen. Bestätigungen dieser Spenden werden auf Anfrage gewährt. Falls Sie zukünftig keine weiteren Einladungen von uns erhalten wollen, klicken Sie bitte auf folgende Link um Ihre Einschreibung abzubestellen : http://www.deutsches-panel.com/rec/unsubscribe.php?id=numpy-discussion@list… Die Ergebnisse der Verlosungen und Spenden werden Ihnen online auf dem Das deutsche Panel Site zur Verfügung stehen. Mit freundlichen Grüßen , Anna Klaus Das deutsche Panel

1 0

Re: [Numpy-discussion] Numarrays from C arrays without copying
by Todd Miller 22 Apr '04

22 Apr '04

On Thu, 2004-04-22 at 11:47, Paul F. Dubois wrote: > This is what PyArray_FromDimsAndData was for, and I believe it is fairly > heavily used. That's good to know. I was just discussing with Perry how to expand the C-API to support this stuff. Perhaps we don't have to. > Todd Miller wrote: > > On Thu, 2004-04-22 at 08:51, Ville Hakulinen wrote: > > > >>Dear all, > >> > >>I would like to turn a large C array allocated by a library into a > >>Numarray object without copying the contents. In other words I have > >>a given array of double precision floats and I want to get a Numarray > >>object, where the data pointer points to this array and no additional > >>memory is allocated. Naturally when the reference count of the Numarray > >>object gets back to zero, the object is freed, but the array itself > >>is left as a nuisance for the C code. > >> > >>Is there a standard way of doing this? > > > > > > No, not yet. You're not the first person to ask for this but I'd > > appreciate it if you'd explain why you need it. So far, not having > > that feature is a conscious omission to keep complexity down. > > > > Regards, > > Todd > > -- Todd Miller <jmiller(a)stsci.edu>

2 2

Numarrays from C arrays without copying
by Ville Hakulinen 22 Apr '04

22 Apr '04

Dear all, I would like to turn a large C array allocated by a library into a Numarray object without copying the contents. In other words I have a given array of double precision floats and I want to get a Numarray object, where the data pointer points to this array and no additional memory is allocated. Naturally when the reference count of the Numarray object gets back to zero, the object is freed, but the array itself is left as a nuisance for the C code. Is there a standard way of doing this? Thanks in advance, -- Ville

4 6

ANN: SciPy 0.3 Released
by Travis N. Vaught 21 Apr '04

21 Apr '04

Greetings, SciPy 0.3 has been released and binaries are available from the scipy.org site. http://www.scipy.org Changes since the 0.1 version (0.1 enjoyed a wide release, there was a version 0.2 that had limited exposure) include the following: - general improvements: Added utility functions for constructing arrays by concatenation, intended mainly for command-line use. Added bmat constructor for easy creation of block matrices. Added mat constructor for constructing matrices (where * is matrix multiplication). Added many PIL utility functions so that if the PIL is installed, image loading, saving, and other operations are available. Added scipy.info, which handles dynamic docstrings and class help better than python's help command. - documentation: much improved - sparse: superLU upgraded to 3.0 - optimize: Added non-linear conjugate gradient algorithm. Improvements to BFGS algorithm and Wolfe-condition line-search. Added two CONSTRAINED optimization techniques. Added simulated annealing and brute-force optimization strategies and Powell's method. Added many very good 1-D root-finding routines. - stats: Added many statistical distributions. Many continuous and discrete random variables are available with a simple mechanism for adding new ones. Added several statistical tests. - special: Added MANY new special functions. |general_function| renamed to |vectorize| and moved to scipy_base. Improved the way domain errors are handled (user specified display of problems). More tests on special functions added. - fftpack: replaced with fftpack2--can optionally be configured to support djbfft - io: Reading of MATLAB .mat files (version 4 and 5) supported. Writing version 4 .mat files supported. Added very flexible and general_purpose text file reader. Added shelving save operation to save variables into an importable module. Routines for reading and writing fortran-records. - linalg: Linear algebra is greatly improved over SciPy 0.1. ATLAS is now optional (but encouraged). Most lapack functions have simplified interfaces (all lapack and blas available). Matrix exponentials and other matrix functions added. - signal: Added support for filter design and LTI system analysis. - xplt: Updated xplt to use newer pygist so that Windows platform is supported. Addition of several plotting types. - integrate: added another ODE integrator. Added romberg integration and several other integration approaches. The complete release notes can be found here: http://www.scipy.org/download/scipy_release_notes_0.3.html You'll also notice that scipy.org is now a spanking new Plone portal (http://www.plone.org -- keep up the good work, plone folks). This is the first binary release in a long time and we hope to increase the frequency to every 6 months or so. If you'd like to follow or join the community, you can subscribe to the mailing lists here: http://www.scipy.org/mailinglists/ Best Regards, Travis BTW SciPy is: ------------- SciPy is an open source library of scientific tools for Python. SciPy supplements the popular Numeric module, gathering a variety of high level science and engineering modules together as a single package. SciPy includes modules for graphics and plotting, optimization, integration, special functions, signal and image processing, genetic algorithms, ODE solvers, and others.

1 0

Sale on OEM software XP, office, Adobe, all under 25-99$ per CD from Bowen's SoftShop
by Monroe Bean 19 Apr '04

19 Apr '04

1 0