Mailman 3 March 2009 - NumPy-Discussion

loadtxt slow
by Gideon Simpson March 2, 2009

March 2, 2009

So I have some data sets of about 160000 floating point numbers stored in text files. I find that loadtxt is rather slow. Is this to be expected? Would it be faster if it were loading binary data? -gideon

6 6

populating an array
by Brennan Williams March 2, 2009

March 2, 2009

Ok... I'm using Traits and numpy. I have a 3D grid with directions I,J and K. I have NI,NJ,NK cells in the I,J,K directions so I have NI*NJ*NK cells overall. I have data arrays with a value for each cell in the grid. I'm going to store this as a 1D array, i.e. 1....ncells where ncells=NI*NJ*NK rather than as a 3D array Apart from lots of other data arrays that will be read in from external files, I want to create I, J and K data arrays where the 'I' array contains the I index for the cell, the 'J' array the J index etc. I haven't used numpy extensively and I'm coming from a Fortran/C background so I'm hideously inefficient. At the moment I have something like.... self.iarray=zeros(self.ncells,dtype=int) for icell in arange(1,self.ncells+2): # icell=i+(j-1)*ni+(k-1)*ni*nj i,j,k=rn2ijk(icell,self.ni,self.nj) self.yarray[icell]=i rn2ijk is defined as... def rn2ijk(n,ni,nj): nij=ni*nj k=(n+nij-1)/nij ij=n-(k-1)*nij j=(ij+ni-1)/ni i=ij-(j-1)*ni return i,j,k Obviously I can improve my use of rn2ijk as for a start I'm recalculating nij for every cell (and I can have anything from a few thousand to a few million cells). But I'm sure I could do this more efficiently by probably starting off with a 3d array, looping over i,j,k and then reshaping it into a 1d array. Ideas? Brennan

2 1

saving an array of strings
by Timmie March 2, 2009

March 2, 2009

Hello, can numpy.savetxt save an array of strings? I got the following arror when saving an array containing strings formatted from datetime objects: File "C:\Programme\pythonxy\python\lib\site-packages\numpy\lib\io.py", line 542, in savetxt fh.write(format % tuple(row) + '\n') TypeError: float argument required Thanks for any help & regards, Timmie

2 1

Call for testing: full blas/lapack builds of numpy on windows 64 bits
by David Cournapeau March 2, 2009

March 2, 2009

Hi, That's a call for testing for 64 bits windows users out there: please try the following binary with the test suite: http://www.ar.media.kyoto-u.ac.jp/members/david/archives/numpy/numpy-1.3.0.… python -c "import numpy; numpy.test()" Report any crash. I am particularly interested in vista and higher versions without visual studio installed - it looks like those cause a lot of trouble wrt the C runtime. *THIS IS ONLY FOR TESTING* - it is still far from being stable. Details: This one is built with free compilers (gcc + gfortran 4.4), and BLAS/LAPACK built with gfortran. This should enable openmp usage too (4.4 supports openmp3.0). It also means supporting scipy is not too far away - there are still some issues with C++, but hopefully, those are compiler bugs, and all numpy issues will have been sorted out. cheers, David

9 15

Bilateral filter: bug corrected
by Nadav Horesh March 2, 2009

March 2, 2009

Attached here a code for bilateral filter: 1. The code is composed of a cython back end (bilateral_base.pyx) and a python front end (bilateral.py) 2. I do not have the tools to make windows binaries (I run it on gentoo linux) . 3. It is not hard to strip the cython code to get a pure (and slow) python implementation. 4. If someone finds the licensing inadequate, I have no problem to re-license. Any comments are (more then) welcome Nadav

3 20