Mailman 3 November 2007 - SciPy-User

Any image deblurring and wavelets packages?
by Ying Wai (Daniel) Fan 21 Nov '07

21 Nov '07

Hi All, I am a graduate student working on image deblurring. Is there any Python package doing that? I am also interested in wavelets. The PyWavelets package hasn't been updated for a year. Is it still being developed? Are there other Python wavelets packages? I want to persuade my adviser to switch from Matlab to Scipy. Without an image deblurring package/toolbox on Scipy, it is hard for him to switch. I am planning to rewrite some of his old codes in Scipy over the break and release them to the public. I just want to know if there is already some image deblurring Python codes out there? Regards, Daniel

2 1

Sparse matrix advice
by Anand Patil 21 Nov '07

21 Nov '07

Hi all, I'm looking for a 'row-ish' sparse matrix format that I can operate on (including making new nonzero elements) quickly in C++ or Pyrex. Could anyone recommend one? The LIL format looks nice because it can be expanded without any copying, but I'm worried that having to call PyList_GetItem (or whatever it is), inc/decreffing and converting from Python to C integers will take up a lot of time. PyTrilinos' Epetra looks pretty cool but I'm not sure how much distributed memory stuff I'll be doing and it would take me a long time to understand. I'm thinking that a good solution might be Pyrex wrappers for matrices represented as C++ vectors of rows, which are themselves vectors; would this actually be a good solution, and is anything like this out there already? I'm a total sparse linear algebra neophyte, so if I'm missing some boat entirely please help me get a clue. Many thanks in advance, Anand Patil

2 1

Questions on leastsq
by Stuart Wilkins 20 Nov '07

20 Nov '07

Hi, I am having problems with the least squares fitting. This may or may not be a bug. I am trying to get an idea of the errors from fitting data. I am usualy fitting to Gaussian Lorentzian, Lor**2 etc. When I try to compute the errors from sigma = sqrt(diag(cov_x)) where cov_x is the covariance matrix I get results which are wildly different than anything before. Including some code I had a long time ago in Matlab. Does anyone have any idea if the form of cov_x is correct or any success in using this? Thanks, Stuart

1 0

reading data with gaps and converting
by Tim Michelsen 20 Nov '07

20 Nov '07

Hello, I encountered another difficulty converting my files from one format to the other. Below you may find a simplified example. When I try to load for instance row 1 on the input file the pylab.load or scipy.io.read_array complains that the index is out of range or that a sequence cannot be converted into a array. I assume that the problem are the cells without any data. The file I use was exported from Excel. There may be some of you who have had to deal with similar format conversion issues. Therefore I would be very happy someone could indicate me a method how to read such structured data efficiently into a array. Thanks in advance. Kind regards, Tim reduced example data: input file structure parameter1 parameter1 parameter2 parameter2 year month day hour1 hour2 hour1 hour2 2007 11 20 8 2 2007 11 21 9 3 4 4 2007 11 22 2 8 4 8 2007 11 23 3 3 6 51 2007 11 24 1 72 34 2007 11 25 7 9 34 24 2007 11 26 6 65 12 1 2007 11 27 7 0 19 2 2007 11 28 44 21 2 output file structure year month day hour parameter1 parameter2 2007 11 20 1 nodata 2 2007 11 20 2 8 nodata 2007 11 21 1 9 4 2007 11 21 2 3 4 2007 11 22 1 2 4 2007 11 22 2 8 8

1 0

Problem with NLP in OpenOpt
by Dark Wind 19 Nov '07

19 Nov '07

Hi, I am using NLP of OpenOpt to solve a NonLinear programming problem. I am using scipy_cobyla solver. The initial point that I have provided is feasible, but I am getting a message: ------------------------------------------message--------------------------------------------------------- starting solver scipy_cobyla (license: BSD ) with problem unnamed solver scipy_cobyla has finished solving the problem unnamed istop: -100 Solver: Time Elapsed = 0.75 CPU Time Elapsed = 0.750110139697 NO FEASIBLE SOLUTION is obtained (max residual = 1.29567664921e-005) ----------------------------------------------------------------------------------------------------------------- All other solvers cannot take h constraint and some others require extra additions. Following is my code: --------------------------------------------------code------------------------------------------------- from scikits.openopt import NLP f = lambda x: 1*x[0] + 7*x[5] + 3*x[10] + 8*x[15] x0 = [1,0,0,0,0,1,0,0,0,0,1,0,0,0,0,1] p=NLP(f,x0,maxIter=1e7,maxFunEvals=1e7) h1 = lambda x: x[0]**2 + x[1]**2 + x[2]**2 + x[3]**2 -1 h2 = lambda x: x[0]*x[4]+x[1]*x[5]+x[2]*x[6]+x[3]*x[7] h3 = lambda x: x[0]*x[8]+x[1]*x[9]+x[2]*x[10]+x[3]*x[11] h4 = lambda x: x[0]*x[12]+x[1]*x[13]+x[2]*x[14]+x[3]*x[15] h5 = lambda x: x[4]**2+x[5]**2+x[6]**2+x[7]**2-1 h6 = lambda x: x[4]*x[8]+x[5]*x[9]+x[6]*x[10]+x[7]*x[11] h7 = lambda x: x[4]*x[12]+x[5]*x[13]+x[6]*x[14]+x[7]*x[15] h8 = lambda x: x[8]**2+x[9]**2+x[10]**2+x[11]**2-1 h9 = lambda x: x[8]*x[12]+x[9]*x[13]+x[10]*x[14]+x[11]*x[15] h10 = lambda x: x[12]**2+x[13]**2+x[14]**2+x[15]**2-1 p.h=[h1,h2,h3,h4,h5,h6,h7,h8,h9,h10] r=p.solve('scipy_cobyla') ----------------------------------------------------------------------------------------------------------- The initial value x0 satisfies the constraints hi's but i am still getting that there is no feasible solution. Can anyone help me with this? Thank you

3 3

Pylab plotting: math range errors unrecoverable
by Adam Ginsburg 19 Nov '07

19 Nov '07

I've been encountering these errors quite often: <type 'exceptions.OverflowError'>: math range error > /usr/lib/python2.5/site-packages/matplotlib/ticker.py(791)scale_range() 790 else: --> 791 ex = divmod(math.log10(-meanv), 1)[0] 792 offset = -10**ex They happen when I try to plot an array containing "inf" values - no surprise there. Two questions, though: 1. is there any way to make pylab ignore inf values, i.e. plot zeroes or quit cleanly instead of giving a messy error? I've looked at masked arrays, but they don't seem to do it, and I'd like something that does not require modifying my data 2. when I get the above error, then try to plot something WITHOUT infinities, it fails with the exact same error, and when I ran the debugger it claimed that meanv was still inf. Is there any way to fix that short of quitting and reentering the python command line? Thanks, Adam

3 2

ode on complex systems
by lorenzo bolla 19 Nov '07

19 Nov '07

Hi all, can I use any of scipy.integrate.ode/odeint etc. for complex systems? How? Here is what I'm doing, but I only get r.y real, instead of complex (and the result is different from Matlab). ---------------------------------------- import numpy import scipy.integrate.ode def f(t, y, betaA, betaB, CAB): M = numpy.array([[1j * betaA, CAB], [-CAB, 1j * betaB]]) return numpy.dot(M, y) betaA = 1. betaB = 1. CAB = 1. t0 = 0. t1 = 1. dt = .01 y0 = numpy.array([1, 0]).astype(complex) r = scipy.integrate.ode(f).set_integrator('vode').set_initial_value(y0, t0).set_f_params(betaA, betaB, CAB) while r.successful() and r.t < t1: r.integrate(r.t + dt) print r.t, r.y ---------------------------------------- Thank you in advance, Lorenzo

5 7

Re: [SciPy-user] Pb with scipy.stats
by Youngsu Park 18 Nov '07

18 Nov '07

Hi, In the scipy/stats directory, there is distribution.py file. Several distribution is defined in it. The betaprime is a instance of betaprime_gen that inherit rv_continuous. The ppf method is not defined In betaprime_gen class. ## Beta Prime class betaprime_gen(rv_continuous): def _rvs(self, a, b): u1 = gamma.rvs(a,size=self._size) u2 = gamma.rvs(b,size=self._size) return (u1 / u2) def _pdf(self, x, a, b): return 1.0/special.beta(a,b)*x**(a-1.0)/(1+x)**(a+b) def _cdf(self, x, a, b): x = where(x==1.0, 1.0-1e-6,x) return pow(x,a)*special.hyp2f1(a+b,a,1+a,-x)/a/special.beta(a,b) def _munp(self, n, a, b): if (n == 1.0): return where(b > 1, a/(b-1.0), inf) elif (n == 2.0): return where(b > 2, a*(a+1.0)/((b-2.0)*(b-1.0)), inf) elif (n == 3.0): return where(b > 3, a*(a+1.0)*(a+2.0)/((b-3.0)*(b-2.0)*(b-1.0)), inf) elif (n == 4.0): return where(b > 4, a*(a+1.0)*(a+2.0)*(a+3.0)/((b-4.0)*(b-3.0) \ *(b-2.0)*(b-1.0)), inf) else: raise NotImplementedError betaprime = betaprime_gen(a=0.0, b=500.0, name='betaprime', shapes='a,b', extradoc="""~~~""") so it uses the rv_continuous class's method. def ppf(self,q,*args,**kwds): """Percent point function (inverse of cdf) at q of the given RV. *args ===== The shape parameter(s) for the distribution (see docstring of the instance object for more information) **kwds ====== loc - location parameter (default=0) scale - scale parameter (default=1) """ loc,scale=map(kwds.get,['loc','scale']) args, loc, scale = self.__fix_loc_scale(args, loc, scale) q,loc,scale = map(arr,(q,loc,scale)) args = tuple(map(arr,args)) cond0 = self._argcheck(*args) & (scale > 0) & (loc==loc) cond1 = (q > 0) & (q < 1) cond2 = (q==1) & cond0 cond = cond0 & cond1 output = valarray(shape(cond),value=self.a*scale + loc) place(output,(1-cond0)+(1-cond1)*(q!=0.0), self.badvalue) place(output,cond2,self.b*scale + loc) goodargs = argsreduce(cond, *((q,)+args+(scale,loc))) scale, loc, goodargs = goodargs[-2], goodargs[-1], goodargs[:-2] place(output,cond,self._ppf(*goodargs)*scale + loc) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ return output .... sgf = vectorize self.vecfunc = sgf(self._ppf_single_call,otypes='d') ... def _ppf(self, q, *args): return self.vecfunc(q,*args) ~~~~~~~~~~~~~~~~~~~~ ... def _ppf_single_call(self, q, *args): return scipy.optimize.brentq(self._ppf_to_solve, self.xa, self.xb, args=(q,)+args, xtol=self.xtol) I'm not sure but I think that the problem comes from the vectorization of defalut _ppf_single_call funstion. But I don'k know how to fix it. Is there someone to fix it?

1 0

Installing numpy
by David Arnold 18 Nov '07

18 Nov '07

All, Mac OS X 10.4.11 PPC. Installed python from this: http://www.python.org/ftp/python/2.5.1/python-2.5.1-macosx.dmg Installation finishes with "There were errors installing the software. Please try installing again." Installing again makes no difference, but Python appears to be running: darnold $ python Python 2.5.1 (r251:54869, Apr 18 2007, 22:08:04) [GCC 4.0.1 (Apple Computer, Inc. build 5367)] on darwin Type "help", "copyright", "credits" or "license" for more information. >>> Downloaded numpy from: http://downloads.sourceforge.net/numpy/ numpy-1.0.4.tar.gz?use_mirror=internap After unpacking and changing directories, simply did this: numpy-1.0.4 $ python setup.py install Seem to finish OK. But testing according to README.txt resulted in: numpy-1.0.4 $ python -c 'import numpy; numpy.test()' Running from numpy source directory. Traceback (most recent call last): File "<string>", line 1, in <module> AttributeError: 'module' object has no attribute 'test' Any ideas or advice? Thanks David.

3 3

Pb with scipy.stats
by LB 18 Nov '07

18 Nov '07

Hi, I've got some problem with some of the continuous distributions defined in scipy.stats : >>> from scipy import stats, __version__ >>> __version__ '0.7.0.dev3440' >>> beta = stats.betaprime(5., 5.) >>> beta.pdf( 0.5 ) array(0.68282274043590918) >>> beta.cdf( 0.5>>> from scipy import stats, __version__ >>> __version__ '0.7.0.dev3440' >>> beta = stats.betaprime(5., 5.) >>> beta.pdf( 0.5 ) array(0.68282274043590918) >>> beta.cdf( 0.5 ) array(0.14484580602550423) So, PDF and CDF methods work, but not PPF : >>> beta.ppf( 0.5 ) Traceback (most recent call last): File "<stdin>", line 1, in <module> File "/home/loic/tmp/pycvs/lib/python2.6/site-packages/scipy/stats/ distributions.py", line 112, in ppf return self.dist.ppf(q,*self.args,**self.kwds) File "/home/loic/tmp/pycvs/lib/python2.6/site-packages/scipy/stats/ distributions.py", line 563, in ppf place(output,cond,self._ppf(*goodargs)*scale + loc) File "/home/loic/tmp/pycvs/lib/python2.6/site-packages/scipy/stats/ distributions.py", line 382, in _ppf return self.vecfunc(q,*args) File "/home/loic/tmp/pycvs/lib/python2.6/site-packages/numpy/lib/ function_base.py", line 949, in __call__ raise ValueError, "mismatch between python function inputs"\ ValueError: mismatch between python function inputs and received arguments It seems the same problem appears with at least foldcauchy, foldnorm and genexpon Is there any way to avoid this ? -- LB

1 0