Mailman 3 October 2007 - NumPy-Discussion

Numpy problem with SVD
by axel.breuer＠bnpparibas.com 18 Oct '07

18 Oct '07

Hi, I tried the function numpy.linalg.svd(X) on an array X with 100 rows and 50 columns. I got an Error message: ** On entry to DGESDD parameter number 12 has an illegal value Then I got kicked out of python. I tried numarray.mlab.svd(X) and it worked fine. Is there a workaround to perform svd with numpy on reasonably sized arrays ? Thanks in advance for your support Axel Breuer This message and any attachments (the "message") is intended solely for the addressees and is confidential. If you receive this message in error, please delete it and immediately notify the sender. Any use not in accord with its purpose, any dissemination or disclosure, either whole or partial, is prohibited except formal approval. The internet can not guarantee the integrity of this message. BNP PARIBAS (and its subsidiaries) shall (will) not therefore be liable for the message if modified. --------------------------------------------- Ce message et toutes les pieces jointes (ci-apres le "message") sont etablis a l'intention exclusive de ses destinataires et sont confidentiels. Si vous recevez ce message par erreur, merci de le detruire et d'en avertir immediatement l'expediteur. Toute utilisation de ce message non conforme a sa destination, toute diffusion ou toute publication, totale ou partielle, est interdite, sauf autorisation expresse. L'internet ne permettant pas d'assurer l'integrite de ce message, BNP PARIBAS (et ses filiales) decline(nt) toute responsabilite au titre de ce message, dans l'hypothese ou il aurait ete modifie.

2 1

design patterns for computing
by Renato Serodio 17 Oct '07

17 Oct '07

Hello there, some helpful pointers provided in the list have put me on track to ask further questions. Now that I'm acquainted with simulating interfaces and duck-typing (which is still a bit fuzzy to me) I think the information I've been looking for all along are the so-called 'design patterns'. Looking at the Optimizer and Stats examples, I could see how new objects were created that built upon a basis class. Stats even had a class to contain calculation results, which brought me even closer to the point. But now I'm stuck again.. I imagined three interfaces to be designed, through which the main program controls the number-crunching without constraining its implementation. These would be: - getInputs: responsible for passing/preparing data and parameters for the analysis/computation - Analysis: which implements all the crunching from FFTs to esoteric stuff like genetic algorithms (and which should be considered opaque from the main program POV) - AnalysisResults: to somehow get the results back into the main program. An additional class would cater for plotting properties, which are usually associated with a particular analysis. Finally, anticipating that I may want to cascade analyses, the inputs and results must be made uniform - must be of the same form, likely numpy arrays. Well, with this pattern, to implement an analysis, I'd have to define derivated getInputs and AnalysisResults classes, and code Analysis to accept and return arrays. A flag to the analysis object could be used to generate plot information or not. (Hey, writing this email is already helping, I'll put the gist in my documentation!) So, after this brainstorming I'll do some more on the framework. I'd be most grateful for comments on the design pattern - is it sensible, could it be better? Are there any good literature sources on patterns for this type of work? Thanks, Renato

2 1

appending extra items to arrays
by Adam Mercer 17 Oct '07

17 Oct '07

Hi In some code I have, I need to append some extra data to a given array. At the moment I construct the data in a list, append the extra information I need and then convert the final list to an array. Is there a way that I can append extra information to an existing array thereby negating the need for the list, as the array object doesn't seem to have an append() method? Cheers Adam

10 12

A basic question on the dot function
by Julien Hillairet 17 Oct '07

17 Oct '07

Hello, First of all, I'm sorry if this question had already been asked. I've searched on the gmane archive and elsewhere on internet, but I didn't found the answer to my question. As expected, the dot product of 2 'classical' vectors works fine : In [50]: a0 = numpy.array([1,2,3]) In [51]: numpy.dot(a0,a0) Out[51]: 14 What I don't understand, is why the dot product of a (3,N) vector gives an error : In [52]: a1 = numpy.array([[1,2,3],[1,2,3],[1,2,3],[1,2,3]]) In [54]: numpy.dot(a1,a1) <type 'exceptions.ValueError'> Traceback (most recent call last) <type 'exceptions.ValueError'>: objects are not aligned instead of what I've expected ; an array([14 14 14 14]). Of course, I've found an alternative in doing : In [61]: numpy.sum(a1*a1,axis=1) Out[61]: array([14, 14, 14, 14]) But I like to understand my mistakes or misunderstands :) So, could someone explained me what I've missed ? Thanks in advance. Best regards, JH

6 7

abstraction and interfaces
by Renato Serodio 17 Oct '07

17 Oct '07

Hello there, I use NumPy/SciPy to process telemetry, producing some metrics that I plot with matplotlib. Well, I want to package my scripts into a sort of all-wrapping application, i.e, the user just calls the app, selects which metrics to produce and waits to obtain the plots, without being exposed to anything more dangerous than a zoom button.. The scripts that produce these metrics use Scipy/Numpy functions that operate on data conveniently converted to numpy arrays. They're quite specific, and I tend to produce/tweak a lot of them. So, to fit in this application someone suggested I programmed 'interfaces' (in java jargon) to them - that way I could develop the whole wrapper application without giving much thought to the actual number-crunching bits. Problem is, I learnt numerics using Fortran, and though I understand the idea of objects, I have quite some difficulty imagining the 'patterns' that allow OO to be so flexible. So I was wondering if anyone experienced with OO Numerical computing could provide some pointers. For context, here are my thoughts: 1. the core of the application is responsible for interaction with the outside world. Crunching blocks are provided data and return matplotlib objects that go into GTK. To write procedures that abide to these parameters is my first (fortranic) impulse. But then I need to change the core everytime I add a procedure, and I'm limited to numpy-IN/matplolib-OUT. 2. if I could force every analysis procedure to be a class that implements its own data retrieval and outputs matplolib objects, my core no longer controls the whole communication with the outside world. And i still have to edit the core everytime a new analysis is required. A very interesting thing is that with these classes I could create new objects with different attributes, rather than pass a host of parameters to a procedure.. So, I'm a bit at odds with this brave new world I discovered. I'd be quite grateful for pointers on literature or examples that more or less lead to a solution. So far my searches have justified the need for objects, but not helped address this modularity requirement... Cheers, Renato

3 3

setting the attributes of an array of object
by jelle 16 Oct '07

16 Oct '07

Hi, I'm wondering whether i can re-write the following idiom with numpy arrays: for i in some_list: i.some_attr = some_value it would be wonderful if one was able to write this idiom as arr[just_these].some_attr = some_value or setattr(arr[just_these], 'some_attr', some_value) since often expensive loops through lists of object could be avoided. any thoughts on this are much appreciated, thanks, -jelle

4 3

Partially merging scons into trunk ?
by David Cournapeau 16 Oct '07

16 Oct '07

Hi, Following the previous thread on using scons within distutils in numpy, I would like to know if it is ok to merge some of the code from numpy.scons branch into the trunk ? I don't want to merge everything, just the part which has little chance to break anything (basically up to rev 4178 in numpy.scons branch; the code has been sucessfully tested on most major platforms). The only downside I can see is that it adds ~ 1 Mb of code into numpy because I included a local copy on scons. cheers, David

2 3

Avoiding distributing config.h in numpy.core ?
by David Cournapeau 16 Oct '07

16 Oct '07

Hi, I spent the last WE trying to understand the whole numpy build process (all this related to my project of "sconsification" of numpy). One thing which I came across, and is a bit problematic is the fact that numpy/core build process relies on a config.h which is meant to be distributed (by distributed, I mean that the config,h is an installed file). Would a patch to remove config.h from the generated files be considered useful ? (this change would of course be 100 % backward compatible: it would be totally transparent to users of numpy headers). To be more precise: this config.h is obviously inspired by autoheader facilities, but the config.h generated by the autotools should not be distributed: http://sourceware.org/autobook/autobook/autobook_96.html#SEC96 Incidently, the problems mentioned in the above link happen in numpy, where we conditionally define some symbols (various SIZEOF_*), because python itself does not follow the rule. A config.h should really only contains #define/#undef lines, nothing else; it would also avoid polluting the namespace (the config.h symbols should not be visible from users of numpy headers as they are now). Finally, this would also make the sconsification process easier, cheers, David

1 0

Re: [Numpy-discussion] .bytes
by Nadav Horesh 15 Oct '07

15 Oct '07

array(1, dtype=float32).itemsize Nadav -----Original Message----- From: numpy-discussion-bounces(a)scipy.org on behalf of Yves Revaz Sent: Sun 14-Oct-07 16:44 To: numpy-discussion(a)scipy.org Subject: [Numpy-discussion] .bytes Dear list, I'm translating codes from numarray to numpy. Unfortunately, I'm unable to find the equivalent of the command that give the number of bytes for a given type : using numarray I used : >>> Float32.bytes 4 I'm sure there is a solution in numpy, but I'm unable to find it. Thanks, Yves -- (o o) --------------------------------------------oOO--(_)--OOo------- Yves Revaz Lerma Batiment A Tel : ++ 33 (0) 1 40 51 20 79 Observatoire de Paris Fax : ++ 33 (0) 1 40 51 20 02 77 av Denfert-Rochereau e-mail : yves.revaz(a)obspm.fr F-75014 Paris Web : http://www.lunix.ch/revaz/ FRANCE ---------------------------------------------------------------- _______________________________________________ Numpy-discussion mailing list Numpy-discussion(a)scipy.org http://projects.scipy.org/mailman/listinfo/numpy-discussion

3 2

.bytes
by Yves Revaz 14 Oct '07

14 Oct '07

Dear list, I'm translating codes from numarray to numpy. Unfortunately, I'm unable to find the equivalent of the command that give the number of bytes for a given type : using numarray I used : >>> Float32.bytes 4 I'm sure there is a solution in numpy, but I'm unable to find it. Thanks, Yves -- (o o) --------------------------------------------oOO--(_)--OOo------- Yves Revaz Lerma Batiment A Tel : ++ 33 (0) 1 40 51 20 79 Observatoire de Paris Fax : ++ 33 (0) 1 40 51 20 02 77 av Denfert-Rochereau e-mail : yves.revaz(a)obspm.fr F-75014 Paris Web : http://www.lunix.ch/revaz/ FRANCE ----------------------------------------------------------------

2 1