Mailman 3 March 2001 - NumPy-Discussion

possible bug: pickling MaskedArrays with the latest cvs version of MA
by Tavis Rudd 24 Mar '01

24 Mar '01

Hi, I'm having difficulty pickling with the new implementation of activeattributes in the cvs version of MA. With the standard pickle module I get the this traceback: >>> import MA >>> a = MA.arange(3400) >>> fp = open('a.pickle','w') >>> from pickle import dump, load >>> dump(a,fp) Traceback (most recent call last): File "<stdin>", line 1, in ? File "/usr/local/lib/python2.1/pickle.py", line 932, in dump Pickler(file, bin).dump(object) File "/usr/local/lib/python2.1/pickle.py", line 109, in dump self.save(object) File "/usr/local/lib/python2.1/pickle.py", line 211, in save f(self, object) File "/usr/local/lib/python2.1/pickle.py", line 471, in save_inst getstate = object.__getstate__ File "/usr/local/lib/python2.1/site-packages/MA/activeattr.py", line 39, in __getattr__ return d['_get'](self, name) File "/usr/local/lib/python2.1/site-packages/MA/activeattr.py", line 80, in basic_get return self.__dict__[name] KeyError: __getstate__ With the cPickle module I'm getting: >>> from cPickle import dump, load >>> dump(a,file) Traceback (most recent call last): File "<stdin>", line 1, in ? cPickle.UnpickleableError: Cannot pickle <type 'instance method'> objects >>> This is with python 2.1b1 and the latest cvs version of Numeric. Has anyone else had this problem? Tavis

2 1

Handling underflow
by victor＠idaccr.org 22 Mar '01

22 Mar '01

Is there some way of having calculations which cause underflow automatically set their result to 0.0? For example when I take exp(a), where a is a floating point array. -- Victor S. Miller | " ... Meanwhile, those of us who can compute can hardly victor(a)idaccr.org | be expected to keep writing papers saying 'I can do the CCR, Princeton, NJ | following useless calculation in 2 seconds', and indeed 08540 USA | what editor would publish them?" -- Oliver Atkin

8 10

How operating arrays in a determined axis?
by Aureli Soria Frisch 19 Mar '01

19 Mar '01

Hello again, I have forgotten in my previous mail that I want the procedure to run independent of the value of N... So the value of N is not known during the implementation. Sorry about the duplication... >Hello! > >I have a N-dimensional array A and I want to operate in one of the axis >(k) with a 1 dimensional array (for instance, subtracting an array B of >length k). I have looked for some solutions in the manual and did not >found any. > >So after testing a lot I found following solution: > >A=Numeric.array(a) > >A.shape=(a1,a2,...,ak,...,aN) #where N is any int value > >B=Numeric.array(b) >B.shape=(ak,) > > >A_modified=Numeric.reshape(A, (a1*a2*...*aN,ak)) #ak is not included in >the product > >result=[] >for i in A_modified: > result.append(i - B) > > >and finally reshaping the result appropriately. But it does not seem >neither elegant nor simple. > >Is there any more elegant solution? The key point is: how to operate a N-D >array with a 1D in one determined axis? > >Thanks in advance. >Regards, >Aureli > ################################# Aureli Soria Frisch Fraunhofer IPK Dept. Pattern Recognition post: Pascalstr. 8-9, 10587 Berlin, Germany e-mail:aureli@ipk.fhg.de fon: +49 30 39 00 61 50 fax: +49 30 39 17 517 #################################

1 0

Release 18.1
by Paul F. Dubois 16 Mar '01

16 Mar '01

Release 18.1 is available as a Windows zip and source tar.gz I have pasted in the change notes for both Numpy itself and MA. RELEASE 18 REQUIRES PYTHON 2.1 Documents have been updated. Homepage (only) resited to a place I can work on more easily.

2 1

Numpy and linking....
by Karl Bellve 15 Mar '01

15 Mar '01

One of the things I am confused about is how to link to numpy on a windows NT. In the docs, it basically just says to include objectarray.h. But, I assume I must also include a static lib or a DLL. I also assume that the pyd files are just dlls under windows. Since they don't have the dll extension, I must manually load them. Is this correct? I just plan to export a 2D array via PyArray_FromDimsAndData. The let a user run a python script and wait for it to finish. -- Cheers, Karl Bellve ICQ # 13956200 Biomedical Imaging Group TLCA# 7938 University of Massachusetts Email: Karl.Bellve(a)umassmed.edu Phone: (508) 856-6514 Fax: (508) 856-1840 PGP Public key: finger kdb(a)molmed.umassmed.edu

1 0

FW: numpy array matching
by Paul F. Dubois 14 Mar '01

14 Mar '01

The obvious approach using equal doesn't seem to work. I'll forward your request. -- Paul -----Original Message----- From: Sam Tannous [mailto:stannous@cisco.com] Sent: Wednesday, March 14, 2001 3:31 AM To: Paul F. Dubois Subject: numpy array matching Paul, I can't seem to get my question posted to the discussion list. Could you please post it for me? (Or even answer it for me?) Thanks, Sam ------------------------------------------- If I have a character array t: >>> t = array(['abcd','aacd','ddfa','qqcd']) >>> print t [[a b c d] [a a c d] [d d f a] [q q c d]] I'm trying to find the row numbers that have a 'c' in the third column (and I'm trying to avoid a 'for' loop). What's the most efficient way to create an array that contains the indexes of the rows that match a particular pattern? (using masks?) The answer should be an array ([0,1,3]) Thanks for any tips, Sam

2 2

Bug? Product of complex arrays
by Aureli Soria Frisch 14 Mar '01

14 Mar '01

Hi, It seems there is a problem when multiplying complex arrays, but maybe that is already solved for the new NumPy 17.3, is it? The product of complex arrays does not seem to be right. Code lines: >>> a=a_conj[:] >>> a array([[ 1.-0.j, 1.-1.j, 1.-2.j], [ 1.-3.j, 1.-4.j, 1.-5.j], [ 1.-6.j, 1.-7.j, 1.-8.j]]) >>> a.imag=a.imag*-1 >>> a array([[ 1.+0.j, 1.+1.j, 1.+2.j], [ 1.+3.j, 1.+4.j, 1.+5.j], [ 1.+6.j, 1.+7.j, 1.+8.j]]) >>> Numeric.multiply(a,a_conj) array([[ 1. +0.j, 0. +2.j, -3. +4.j], [ -8. +6.j, -15. +8.j, -24.+10.j], [-35.+12.j, -48.+14.j, -63.+16.j]]) where the answer should be [[1+0j,2+0j,5+0j][10+0j,... ################################# Aureli Soria Frisch Fraunhofer IPK Dept. Pattern Recognition post: Pascalstr. 8-9, 10587 Berlin, Germany e-mail:aureli@ipk.fhg.de fon: +49 30 39 00 61 50 fax: +49 30 39 17 517 #################################

2 1

conversion to complex
by Eric Hagemann 14 Mar '01

14 Mar '01

I have an array of floats that is an interleaved stream of complex (real/imag) values that I want to promote to complex. My current method uses the following a=array([1,2,3,4]) b = a[0:4:2] + cmath.sqrt(-1)*a[1:4:2] Is there a more elegant way ? I am hoping to avoid explicit for loops etc

3 2

Numpy for Windows
by Karl Bellve 13 Mar '01

13 Mar '01

I am using Numpy and Python 2.1b1 as an internal scripting language for an NT application. I have the just installed the Numpy 18 Windows binaries but I believe I need the source, including arrayobject.c and arrayobject.h. I need them to link them to my application, so I can pass Numpy Arrays to python. I am not having any luck using anonymous CVS to download the current source. I wonder if the person who just released the windows binary version could also release the header files and libraries that I need to link to? -- Cheers, Karl Bellve

1 0

How to uninstall Numeric-17.3.0.tar.gz
by R. Lahaye 13 Mar '01

13 Mar '01

Hi, [ I'm not a member of this list, please forward reply also to my personal email address ] First of all, I'm not at all familiar with python or numeric python. I need Numeric for testing a software package on my Linux PC. I only needed to try the sofware package, so I wanted to temporarily install Numeric, How do I do that? I mean, there seems not to be a script for uninstall!?! The scripts in 17.3.0 only allow me to install. Why is the uninstall missing? Or am I overlooking something? Thanks, Rob. PS: don't tell me to go in such and such directory and delete such and such files. That would exactly be what scripts are supposed to do.

2 1