Mailman 3 November 2009 - SciPy-User

Re: [SciPy-User] loading mat file in scipy
by Alireza Alemi-Neissi Nov. 5, 2009

Nov. 5, 2009

Robin wrote: > On Tue, Oct 20, 2009 at 6:24 PM, Matthew Brett <matthew.brett(a)gmail.com> wrote: >> If you are using 0.7.1, then maybe you can let me have a look at your >> file somehow - maybe by sharing a dropbox folder? > Here is the link of the data: http://rapidshare.com/files/296444142/nestedStructedSyptom.mat.html > Are you able to look at the file I provided: > http://www.robince.net/robince/structs_cells.mat > > At 900K I think it is a more reasonable size for testing, and in terms > of s/MB performs a lot slower than the OPs original example. I'm sure > the root cause for any slowness would be the same - it also has > structs of cells with structs nested many times. > I am sure the cause of slow loading of your file and that of mine are the same. It took ~56sec on my computer to load your data into the python. Cheers, Alireza ---------------------------------------------------------------- SISSA Webmail https://webmail.sissa.it/ Powered by SquirrelMail http://www.squirrelmail.org/

2 1

Reshaping Question
by Skipper Seabold Nov. 5, 2009

Nov. 5, 2009

My brain is failing me. Is there a clean way to reshape an array like the following? import numpy as np c = np.arange(16).reshape(4, 2, 2) In [209]: c Out[209]: array([[[ 0, 1], [ 2, 3]], [[ 4, 5], [ 6, 7]], [[ 8, 9], [10, 11]], [[12, 13], [14, 15]]]) So that c == d where d = np.array(([0, 1, 4, 5], [2,3,6,7], [8,9,12,13], [10, 11, 14, 15])) In [211]: d Out[211]: array([[ 0, 1, 4, 5], [ 2, 3, 6, 7], [ 8, 9, 12, 13], [10, 11, 14, 15]]) Cheers, Skipper

3 6

fmin_slsqp- Bounds are not obeyed
by jagan prabhu Nov. 4, 2009

Nov. 4, 2009

Dear users, Problem is " Bounded with inequality constrains", in slssqp often Bounds are not obeyed, its deviates the bounds. So if deviates i made it to come back with in the region(bounds). But i face a problem in execution. i get error like, File "/usr/lib/python2.5/site-packages/scipy/optimize/slsqp.py", line 277, in fmin_slsqp c_ieq = array([ ieqcons[i](x) for i in range(len(ieqcons)) ]) File "/usr/lib/python2.5/site-packages/scipy/optimize/optimize.py", line 97, in function_wrapper return function(x, *args) TypeError: <lambda>() takes exactly 1 argument (2 given) program will look like, #import os #import scipy.optimize from scipy import * import numpy from scipy import optimize from numpy import asarray from math import * def cst(aParams,bounds): aParams = numpy.asarray(aParams) for par in range(len(aParams)): if ((bounds[par][0]<= aParams[par]<= bounds[par][1])): pass else: if (aParams[par]< bounds[par][0]): aParams[par] = bounds[par][0] if (aParams[par]> bounds[par][1]): aParams[par] = bounds[par][1] x = aParams[0] y = aParams[1] z = aParams[2] # objective function eqn = -cos(x)*cos(y)*cos(z)*log(-((x-pi)**2-(y-pi)**2-(z-pi)**2)) return eqn #Initial guess Init = numpy.array([5.0,15.0,17.0]) # parameters x,y,z bounds = [(2.0, 20000.0),(4.0, 50000.0),(5.0, 60000.0)] # inequality constraints x must be least,y larger than x smaller than z,and z the largest of all con1 = lambda x:numpy.asarray(x[1]-x[0], x[2]-x[1]) opt = fmin_slsqp(cst,Init,ieqcons= [con1] , bounds=bounds, fprime = None, args=(bounds,), full_output=True, iter=20000, iprint=2, acc=0.001) print '****************************************' print opt[0] print opt[1] print opt[2] print opt[4] Problems are, 1, bounds i could not able to pass to the function as args( ). 2, Whether implementation of the ineq. constraints are correct? any better way? 3, How to avoid bounds deviation? Please help me. Regards, Prabhu Now, send attachments up to 25MB with Yahoo! India Mail. Learn how. http://in.overview.mail.yahoo.com/photos

3 3

Wrapping C/C++ Code
by Lorenzo Isella Nov. 4, 2009

Nov. 4, 2009

Dear All, I hope this is not too off-topic. If you were asked to wrap C/C++ codes into a Python application (potentially relying on NumPy/SciPy) which route would you follow? Bear in mind that the initial C/C++ code is a standalone program which was not written having Python in mind at all. Many thanks Lorenzo

5 4

GPG key issues when trying to install on RHEL
by Girish Venkatasubramanian Nov. 3, 2009

Nov. 3, 2009

Hello, I followed your instructions and copied the appropriate .repo file to /etc/yum.repos.d/ on my RHEL5 x86_64 machine. But when I try to install the python-numpy and python-scipy using yum, I get the following error Warning: rpmts_HdrFromFdno: Header V3 DSA signature: NOKEY, key ID eda8433f GPG key retrieval failed: [ Errno 4] IOError: <urlopen error (110, 'Connection timed out')> I am trying to install via a proxy - and I have configured the yum.conf and set proxy_http and proxy_ftp directives. Any help is appreciated. Thanks

2 2

Re: [SciPy-User] reverse Cuthill-McKee
by Dominique Orban Nov. 3, 2009

Nov. 3, 2009

> ---------- Forwarded message ---------- > From: Robert Cimrman <cimrman3(a)ntc.zcu.cz> > To: SciPy Users List <scipy-user(a)scipy.org> > Date: Mon, 02 Nov 2009 17:46:57 +0100 > Subject: [SciPy-User] reverse Cuthill-McKee > Hi, > > I need an implementation of the (symmetric) reverse Cuthill-McKee matrix reordering algorithm. Is anyone aware of an implementation callable from Python? A scipy CSR/CSC matrix-based one would be the best, of course. > > thanks, > r. Hi Robert, I just pushed my repo to GitHub so you can try it out. I'm using the implementation from the Harwell Subroutine Library which you'll need to grab from their website. Specify your source dir in site.cfg and you should be good to go. git clone git://github.com/dpo/pyorder.git Cheers, Dominique

2 1

linear algebra: quadratic forms without linalg.inv
by josef.pktd＠gmail.com Nov. 2, 2009

Nov. 2, 2009

This is just an exercise. In econometrics (including statsmodels) we have a lot of quadratic forms that are usually calculate with a matrix inverse. I finally spend some time to figure out how to do this with cholesky or LU composition which should be numerically more stable or accurate (and faster). "Don't let that INV go past your eyes" (matlab file exchange) Josef """ Use cholesky or LU decomposition to calculate quadratic forms different ways to calculate matrix product B.T * inv(B) * B Note: calling convention in sparse not consistent, sparse requires loop over right hand side Author: josef-pktd """ import numpy as np from scipy import linalg B = np.ones((3,2)).T B = np.arange(6).reshape((3,2)).T print 'using inv' Ainv = linalg.inv(A) print np.dot(Ainv, B[:,0]) print np.dot(Ainv, B) print reduce(np.dot, [B.T, Ainv, B]) print 'using cholesky' F = linalg.cho_factor(A) print linalg.cho_solve(F, B[:,0]) print linalg.cho_solve(F, B) print np.dot(B.T, linalg.cho_solve(F, B)) print 'using lu' F = linalg.lu_factor(A) print linalg.lu_solve(F, B[:,0]) print linalg.lu_solve(F, B) print np.dot(B.T, linalg.lu_solve(F, B)) from scipy import sparse Asp = sparse.csr_matrix(A) print 'using sparse symmetric lu' F = sparse.linalg.splu(A) print F.solve(B[:,0]) #print F.solve(B) # wrong results but no exception AiB = np.column_stack([F.solve(Bcol) for Bcol in B.T]) print AiB print np.dot(B.T, AiB) #not: #Bsp = sparse.csr_matrix(B) #print B.T * F.solve(Bsp) # argument to solve must be dense array print 'using sparse lu' F = sparse.linalg.factorized(A) print F(B[:,0]) #print F(B) # wrong results but no exception AiB = np.column_stack([F(Bcol) for Bcol in B.T]) print np.dot(B.T, AiB)

6 17

reverse Cuthill-McKee
by Robert Cimrman Nov. 2, 2009

Nov. 2, 2009

Hi, I need an implementation of the (symmetric) reverse Cuthill-McKee matrix reordering algorithm. Is anyone aware of an implementation callable from Python? A scipy CSR/CSC matrix-based one would be the best, of course. thanks, r.

1 0

least-square filter design
by Neal Becker Nov. 2, 2009

Nov. 2, 2009

Anyone have code for least square (minimum mean square error) FIR filter design?

2 4

scipy.linalg.det TypeError
by Norm Wood Nov. 1, 2009

Nov. 1, 2009

Hi, I'm trying to use the scipy code for kernel density estimation and am running into a problem with the calculation of a determinant. In scipy/linalg/basic.py, the determinant code has a couple of lines: fdet, = get_flinalg_funcs(('det',),(a1,)) a_det,info = fdet(a1,overwrite_a=overwrite_a) These are at lines 484 and 485 in the scipy 0.7.1 basic.py source file. Execution of the second line produces a TypeError: TypeError: 'NoneType' object is not callable I expect this means that the first line failed to "get" the linalg function called 'det'. I'd appreciate any suggestions about why this is happening. The odd thing is that the numpy routine for calculating a determinant works fine. Here's simple code that shows the problem: #!/usr/bin/env /usr/bin/python import numpy import scipy.linalg import numpy.linalg A = numpy.matrix([[1.1, 1.9],[1.9,3.5]]) #This works fine y = numpy.linalg.det(A) #This throws a TypeError y = scipy.linalg.det(A) I'm using LAPACK 3.1.1, Atlas 3.8.0, numpy 1.2.1, and I've tried both scipy 0.7.0 and 0.7.1, all built with gcc 3.3.6 & g77. Platform is Linux (Slackware 10.2), kernel 2.4.31. Python is version 2.4.1, built with gcc 3.3.5 Here's the Atlas build info: ATLAS version 3.8.0 built by norm on Thu Dec 6 17:18:59 MST 2007: UNAME : Linux wombat 2.4.31-1 #1 SMP Wed Dec 7 16:52:41 MST 2005 i686 unknown unknown GNU/Linux INSTFLG : -1 0 -a 1 ARCHDEFS : -DATL_OS_Linux -DATL_ARCH_P4E -DATL_CPUMHZ=3014 -DATL_SSE3 -DATL_SSE2 -DATL_SSE1 -DATL_GAS_x8632 F2CDEFS : -DAdd__ -DF77_INTEGER=int -DStringSunStyle CACHEEDGE: 524288 F77 : g77, version GNU Fortran (GCC) 3.3.6 F77FLAGS : -O -m32 SMC : gcc, version gcc (GCC) 3.3.6 SMCFLAGS : -fomit-frame-pointer -mfpmath=387 -O2 -falign-loops=4 -m32 SKC : gcc, version gcc (GCC) 3.3.6 SKCFLAGS : -fomit-frame-pointer -mfpmath=387 -O2 -falign-loops=4 -m32 The output of numpy/distutils/system_info.py is pretty long. If there's particular information from that which would be useful, I'll be glad to send it, but it looks to me as if numpy and scipy are using the same atlas libraries. Here's the atlas_info from system_info.py atlas_info: libraries f77blas,cblas,atlas not found in /usr/local/lib libraries lapack_atlas not found in /usr/local/lib libraries lapack_atlas not found in /home/norm/lusr/lib __main__.atlas_info FOUND: libraries = ['lapack', 'f77blas', 'cblas', 'atlas'] library_dirs = ['/home/norm/lusr/lib'] language = f77 include_dirs = ['/home/norm/lusr/include'] And here's the atlas_blas_info from the scipy build: atlas_blas_info: libraries f77blas,cblas,atlas not found in /usr/local/lib FOUND: libraries = ['f77blas', 'cblas', 'atlas'] library_dirs = ['/home/norm/lusr/lib'] language = c include_dirs = ['/home/norm/lusr/include'] Thanks for any suggestions. Norm

3 3