Mailman 3 March 2010 - SciPy-User

problems with splrep,splev
by Christian K. 08 Jun '10

08 Jun '10

Hi, I am having problems with the spline functions from scipy.interpolate. Reading the doc strings I expected the following to work: from scipy.interpolate import splev,splrep import numpy as N x = N.linspace(-4,10,5) y = x**2 t,c,l = splrep(x,y,xb=0,xe=4) Traceback (most recent call last): File "C:\pythonxy\python\lib\site-packages\scipy\interpolate\fitpack.py", line 406, in splrep n,c,fp,ier = dfitpack.curfit(task, x, y, w, t, wrk, iwrk, xb, xe, k, s) error: (xb<=x[0]) failed for 1st keyword xb Second, I want to fit several curves using the same knots, thus I tried the following: y2 = x**2-3*x+4 t2,c2,l2 = splrep(x,y2,t=t) Traceback (most recent call last): File "C:\pythonxy\python\lib\site-packages\scipy\interpolate\fitpack.py", line 418, in splrep raise _iermess[ier][1],_iermess[ier][0] ValueError: Error on input data Are those bugs or am I misunderstanding something? Eventually I would like to morph one spline into another, therefor I want to stick with the knots. If anyone has experiences with splines and morphing I would be very interested to hear about it. Regards, Christian

4 4

Bug in ndimage.map_coordinates with mode='wrap' ?
by Kurt Smith 12 May '10

12 May '10

Hi, Testing the example code in ndimage.map_coordinate's docstring, I can't get things to work with mode='wrap'. What am I doing wrong? In [31]: a Out[31]: array([[ 0., 1., 2.], [ 3., 4., 5.], [ 6., 7., 8.], [ 9., 10., 11.]]) In [32]: ndimage.map_coordinates(a, [range(5), [0]*5], order=1, mode='wrap') # should be 0, 3, 6, 9, 0 -- right? Out[32]: array([ 0., 3., 6., 9., 3.]) In [33]: ndimage.map_coordinates(a, [[0]*4, range(4)], order=1, mode='wrap') # should be 0, 1, 2, 0 -- right? Out[33]: array([ 0., 1., 2., 1.]) Here's the output when extending the sampling range: In [36]: ndimage.map_coordinates(a, [range(10), [0]*10], order=1, mode='wrap') # should be 0, 3, 6, 9, 0, 3, 6, 9, ... Out[36]: array([ 0., 3., 6., 9., 3., 6., 0., 3., 6., 0.]) In [37]: ndimage.map_coordinates(a, [[0]*8, range(8)], order=1, mode='wrap') Out[37]: array([ 0., 1., 2., 1., 0., 1., 0., 1.]) If it's a bug, where can I file a report, and what can I do to help fix it? Looks like the wrapping code is in a compiled extension module -- I'll take a look. Kurt

1 2

How do I use vonmises.fit()?
by David Ho 14 Apr '10

14 Apr '10

Hi all! I want to fit some data with a Von Mises distribution, and get the mean and "kappa" parameters for that distribution. I'm a little confused about scipy.stats.distributions.vonmises.fit(). Just as a test, I tried this: >>> vm_rvs = scipy.stats.vonmises.rvs(1.2, 2.3, size=10000) # the first argument appears to be "kappa", and the second argument is the mean. >>> scipy.stats.distributions.vonmises.fit(vm_rvs) array([ 1.17643696e-01, 3.38956854e-03, 1.27331662e-27]) I got an array of 3 values, none of which seem to be equal to the mean or kappa of the distribution. I looked around in some of the docstrings, but I couldn't find any clear documentation of what these values are supposed to correspond to. I would've expected vonmises.fit() to return something like: array([ 1.2, 2.3]) What am I doing wrong? Thanks for your help, --David Ho PS: I also don't fully understand the "scale" and "loc" parameters for all of the continuous random variables. Does "loc" always correspond to the mean, and "scale" always correspond to the square root of the variance, or something else?

3 6

No Wilcoxon test in scipy ?
by Axel Kowald 01 Apr '10

01 Apr '10

Hello everybody, I'm using scipy 0.7.1 and try to find the Wilcoxon test, i.e. a non-parametric test for paired observations. Am I right, that there is no Wilcoxon test in SciPy ?! In the function summary of http://www.scipy.org/doc/api_docs/SciPy.stats.stats.html I could not find anything with Wilcoxon. I'm only a bit confused, because at the beginning of this website the Wilcoxon test is listet: INFERENTIAL STATS: ttest_1samp ttest_ind ttest_rel chisquare ks_2samp mannwhitneyu ranksums wilcoxon <=============== kruskal friedmanchisquare So, is the Wilcoxon test implemented in SciPy or not ? Many thanks, Axel

4 5

MATLAB accumarray equivalent ?
by Mohammad Abdollahi 01 Apr '10

01 Apr '10

Hi, Does anyone know what is the equivalent function of MATLAB's "accumarray()" in scipy/numpy, with the exact same functionality ? Thanks

7 8

implementing recursions
by nicky van foreest 01 Apr '10

01 Apr '10

Hi, Just out of curiosity, is there a smarter way to implement the following recursion? grid =delta = 0.001 end = 20 np.arange(0,end+delta,delta) n = len(grid) gamma = 3*grid for j in range(n-1, -1, -1): gamma[j] += dot(gamma[j+1:],G[1:n-j]) # G is a precomputed np.array I still dislike the remaining for loop, but I do not see an easy way out. Of course, it is possible to write the above as Ax = b, and use np.solve, but then I have to build the matrix A, which takes more time as much more memory is claimed. (I also realize that I can use Z-transforms, but this appears somewhat of a hassle, to say the least.) Thanks Nicky

2 2

preconditioner for GMRES
by Darcoux Christine 31 Mar '10

31 Mar '10

I am developping a Jacobian-Free Newton-Krylov code to solve the Navier Stokes equations. In this code, the product of the Jacobian matrix with a given vector is represented by the matvec method of a "LinearOperator"object. The scipy.sparse.linalg.isolve.gmres method can takes an optional preconditioner as parameter which is either an object "LinearOperator" or a matrix. However, the scipy documentation does not explain how one can form this preconditioner in a (ideally) matrix-free manner given only the matvec method of the jacobian. Any suggestions on this would be very appreciated. Thanks, Christine

3 4

Kurtosis/Skewness
by Dan bole 31 Mar '10

31 Mar '10

Hi all, I am trying to create a series of random variables selected from a distribution. I would like this distribution to start as a normal distribution, but then be altered based on assumptions of skewness and kurtosis (so I am not calculating skewness/kurtosis from a dataset, but instead creating the probability density function from assumptions of skewness/kurtosis). I can create a normal distribution and then pull random variables from this, and was wondering if it is possible to create a distribution based on assumptions of skewness and kurtosis? Many thanks,Dan _________________________________________________________________ Hotmail: Trusted email with Microsoft’s powerful SPAM protection. http://clk.atdmt.com/GBL/go/210850552/direct/01/

4 5

How to estimate error in polynomial coefficients
by Bruce Peterson 30 Mar '10

30 Mar '10

Another way to estimate the variability of model parameters is to use resampling statistics such as the Jackknife or Bootstrap approach. Basically a Jackknife estimate of the parameter error is obtained by removing one point from your data set,fitting your model to the remaining data, record the estimated parameters, replacing the data point, remove another and refit the model and recording the new estimates--repeat until all data points are removed and replaced. The set of estimated parameter values can be to create the cumulative frequency distribution and thus the variability of the parameter estimates. see http://en.wikipedia.org/wiki/Resampling_(statistics) Bruce Peterson, Ph.D. Terastat, Inc Information Access Systems Voice (425) 466 7344

1 0

installation, Espoo-Helsinki
by gintare statkute 30 Mar '10

30 Mar '10

Hello, I tried to install scientific-python one - two months before, but got in problems with ATLAS installation. I tried several times, thus now LAPACK maybe also has to be reinstalled. I really would like to try some simulations with scientific python, maybe somebody could help me in installation. regards, gintare

2 1