Mailman 3 September 2008 - NumPy-Discussion

Vectorization of the product of several matrices ?
by oc-spam66 01 Oct '08

01 Oct '08

Hello, I have two lists of numpy matrices : LM = [M_i, i=1..N] and LN = [N_i, i =1..N] and I would like to compute the list of the products : LP = [M_i * N_i, i=1..N]. I can do : P=[] for i in range(N) : P.append(LM[i]*LN[i]) But this is not vectorized. Is there a faster solution ? Can I vectorize this operation ? How ? Will it be more efficient than the code above ? Thank you for your help, -- O.C. Créez votre adresse électronique prenom.nom(a)laposte.net 1 Go d'espace de stockage, anti-spam et anti-virus intégrés.

2 1

What is the sign of nan?
by Charles R Harris 01 Oct '08

01 Oct '08

Hi All, I've been cleaning up the ufunc loops and the sign function currently doesn't have a defined behavior for nans. This makes the results depend on the order/type of comparisons in the code, which looks fragile to me. So what should it return? I vote for nan but am open for suggestions. Chuck

5 35

Numpy Nu-bee: "forward fill" function
by Joshua Ford 30 Sep '08

30 Sep '08

Hello all! I'm a python and numpy newbie and I'm having difficulty doing something that seems to be relatively straightforward. So I figured I would ask the experts! I would like to conduct the equivalent of a "forward fill" function on an array, that repeats the last non-null value until the next non-null value is reached. I saw that the scipy TimeSeries has a "forward_fill" function: http://pytseries.sourceforge.net/lib/interpolation.html I would like to do exactly what the TimeSeries "forward_fill" function does - but I don't have a time series. I simply want to convert an array [1,2,3,4,5,6,-999,-999,-999,-999,7,8] to [1,2,3,4,5,6,6,6,6,6,7,8] Here's what I've tried: from numpy import * a = array([1,2,3,4,5,6,-999,-999,-999,-999,7,8]) adata = a for i in adata: if i != -999: j = i elif i == -999: j = adata[i-1] print j else: print 'end' Of course .... this doesn't work. Many thanks in advance for any help or suggestions! As most of you can probably tell from my ugly code, I'm not a programmer...

2 2

nonuniform scatter operations
by Geoffrey Irving 30 Sep '08

30 Sep '08

Hello, Is there an efficient way to implement a nonuniform gather operation in numpy? Specifically, I want to do something like n,m = 100,1000 X = random.uniform(size=n) K = random.randint(n, size=m) Y = random.uniform(size=m) for k,y in zip(K,Y): X[k] += y but I want it to be fast. The naive attempt "X[K] += Y" does not work, since the slice assumes the indices don't repeat. Thanks, Geoffrey

3 4

problem calling SD.SDim.setname in pyhdf
by Catherine Moroney 30 Sep '08

30 Sep '08

Hi, I recently upgraded pyhdf from 0.7-3 to 0.8-2, and have noticed some problems with the SD.SDim interface. Specifically, the short test code attached works fine with version 0.7-3 and fails with version 0.8-2 when it's trying to assign a name to the dimension. I get the following error (below). Does anybody else have similar problems while trying to attach a name to an SDim object in the most recent version of pyhdf? I'm running on a Linux box, python version 2.5. etting Dimension: Dimension0 Traceback (most recent call last): File "./Test_pyhdf.py", line 35, in <module> main() File "./Test_pyhdf.py", line 27, in main dim.setname(dname) File "/usr/lib64/python2.5/site-packages/pyhdf/SD.py", line 2856, in setname _checkErr('setname', status, 'cannot execute') File "/usr/lib64/python2.5/site-packages/pyhdf/error.py", line 24, in _checkErr raise HDF4Error, str pyhdf.error.HDF4Error: ('s', 'e', 't', 'n', 'a', 'm', 'e', ' ', ':', ' ', 'c', 'a', 'n', 'n', 'o', 't', ' ', 'e', 'x', 'e', 'c', 'u', 't', 'e') Thanks for any help, Catherine #! /usr/bin/env python from pyhdf import HDF from pyhdf import SD def main(): rank = (10,20) flhdf = SD.SD("Test_pyhdf.hdf",HDF.HC.CREATE|HDF.HC.WRITE| HDF.HC.TRUNC) sds = flhdf.create("Test SDS",HDF.HC.INT32,rank) for idim in range(0,2): dim = sds.dim(idim) dname = "Dimension" + str(idim) print "Setting Dimension: ",dname dtype = HDF.HC.FLOAT32 scale = [i for i in range(0,rank[idim])] dim.setname(dname) <----- CRASHES ON THIS LINE dim.units = "None" dim.setscale(dtype, scale) sds.endaccess() flhdf.end() if __name__ == "__main__": main()

2 1

numpy installation as non root using site.cfg
by Poul Georg Moses 30 Sep '08

30 Sep '08

Dear mailing list I am trying to install numpy-1.1.1 on a cluster where I do not have root permission. I have succesfully installed \ python2.5.2. I am using the cite.cfg file to include the blas and atlas libraries. I have done that by setting the\ defaults to: library_dirs = /usr/lib include_dirs = /usr/include where blas and atlas is located. Then I type: python setup.py install --home=$HOME and get the following output Running from numpy source directory. F2PY Version 2_5585 Traceback (most recent call last): File "setup.py", line 96, in <module> setup_package() File "setup.py", line 89, in setup_package configuration=configuration ) File "/data/pgmoses/SourceCode/numpy-1.1.1/numpy/distutils/core.py", line 150, in setup config = configuration() File "setup.py", line 56, in configuration config.add_subpackage('numpy') File "/data/pgmoses/SourceCode/numpy-1.1.1/numpy/distutils/misc_util.py", line 850, in add_subpackage caller_level = 2) File "/data/pgmoses/SourceCode/numpy-1.1.1/numpy/distutils/misc_util.py", line 833, in get_subpackage caller_level = caller_level + 1) File "/data/pgmoses/SourceCode/numpy-1.1.1/numpy/distutils/misc_util.py", line 780, in _get_configuration_from_s\ etup_py config = setup_module.configuration(*args) File "numpy/setup.py", line 9, in configuration config.add_subpackage('core') File "/data/pgmoses/SourceCode/numpy-1.1.1/numpy/distutils/misc_util.py", line 850, in add_subpackage caller_level = 2) File "/data/pgmoses/SourceCode/numpy-1.1.1/numpy/distutils/misc_util.py", line 833, in get_subpackage caller_level = caller_level + 1) File "/data/pgmoses/SourceCode/numpy-1.1.1/numpy/distutils/misc_util.py", line 780, in _get_configuration_from_s\ etup_py config = setup_module.configuration(*args) File "numpy/core/setup.py", line 299, in configuration blas_info = get_info('blas_opt',0) File "/data/pgmoses/SourceCode/numpy-1.1.1/numpy/distutils/system_info.py", line 267, in get_info return cl().get_info(notfound_action) File "/data/pgmoses/SourceCode/numpy-1.1.1/numpy/distutils/system_info.py", line 366, in __init__ self.parse_config_files() File "/data/pgmoses/SourceCode/numpy-1.1.1/numpy/distutils/system_info.py", line 372, in parse_config_files self.cp.read(self.files) File "/data/pgmoses/lib/python2.5/ConfigParser.py", line 267, in read self._read(fp, filename) File "/data/pgmoses/lib/python2.5/ConfigParser.py", line 462, in _read raise MissingSectionHeaderError(fpname, lineno, line) ConfigParser.MissingSectionHeaderError: File contains no section headers. file: /data/pgmoses/SourceCode/numpy-1.1.1/site.cfg, line: 60 'library_dirs = /usr/lib\n' Do you have any suggestion to what I might be doing wrong Best regards, Poul Georg Moses -- View this message in context: http://www.nabble.com/numpy-installation-as-non-root-using-site.cfg-tp19697… Sent from the Numpy-discussion mailing list archive at Nabble.com.

3 2

profiling line by line
by Robin 29 Sep '08

29 Sep '08

Hi, I am using the prun feature of Ipython which is very helpful. I was wondering though if theres anything for Python that would allow line-by-line profiling (ie a time for each line of code) like the MATLAB profiler? Cheers Robin

10 40

Help needed with the priority of 0d-arrays and np.scalars
by Pierre GM 29 Sep '08

29 Sep '08

All, In a recent post, I was pointing to a bug in numpy.ma. I think I can understand what happens, but I'd need a couple of clarifications. When multiplying a 0d ndarrray or one of its subclass by a numpy scalar, we end up with a numpy scalar cast to the highest type, right ? So, np.float64(1)*np.array(0.,dtype='g') is a np.float128 object, while np.float64(1)*np.array(0,dtype=np.float32) is a np.float64 object. What surprises me is that numpy scalars have a low priority (listed as -100 or -1000000 depending on the page of the numpy book), so if I have a 0d subclass of ndarray with a higher priority, I should end up with a 0d subclass of ndarray, but it doesn't seem to be the case. The problem comes and bites me with ma.masked, defined as a 0d MaskedArray with __array_priority__ of 15 and a np.float64 dtype. Doing ma.masked*np.float64(1.) yields ma.masked, as it should. Doing np.float64(1.)*ma.masked yields np.float64(0.) but shouldn't: I should access instead the __rmul__ method of ma.masked. Any comment/suggestion would be welcome.

1 0

Minimum distance between 2 paths in 3D
by Andrea Gavana 28 Sep '08

28 Sep '08

Hi All, I was wondering if someone had any suggestions/references/snippets of code on how to find the minimum distance between 2 paths in 3D. Basically, for every path, I have I series of points (x, y, z) and I would like to know if there is a better way, other than comparing point by point the 2 paths, to find the minimum distance between them. Any suggestion is welcome :-) Thank you. Andrea. "Imagination Is The Only Weapon In The War Against Reality." http://xoomer.alice.it/infinity77/

6 7

Cython docs including numpy info
by Fernando Perez 28 Sep '08

28 Sep '08

Hi all, I just found out something that might interest others on this list. Gabriel Gellner (forgive me if I'm not duly crediting others, I don't know the detailed history of this) has been leading the effort to produce self-contained Cython documentation, in sphix, that includes the new numpy support added recently to Cython: http://www.mudskipper.ca/cython-doc/ I'm looking at these docs right now, but I'm sure Gabriel and the rest of the Cython team would benefit from more feedback. Cheers, f

1 0