Mailman 3 November 2009 - SciPy-User

chebfun
by josef.pktd＠gmail.com Dec. 8, 2009

Dec. 8, 2009

I just came by chance across this (for matlab) http://www2.maths.ox.ac.uk/chebfun/index.html http://www2.maths.ox.ac.uk/chebfun/license.html The documentation looks helpful for someone like me who doesn't know enough about what Chebyshev polynomials are good for. Josef

3 3

ANN: graph-tool, a package for graph analysis and manipulation
by Tiago de Paula Peixoto Dec. 4, 2009

Dec. 4, 2009

Hi, This is a shameless plug for graph-tool, a python module for manipulation and statistical analysis of graphs. It is aimed at the so-called "network research", and contains many algorithms useful in this area, such as correlations, clustering, community detection, maximum flow, spectral properties, betweeness, pagerank, etc. It is written primarily in C++ with the Boost Graph Library, so it should scale quite well performance-wise when large graphs are used. The interface is of course in Python, and it integrates very well with scipy/numpy. More information, including the complete documentation, is available at: http://graph-tool.forked.de Cheers, Tiago

6 12

ANN: visvis
by Almar Klein Dec. 3, 2009

Dec. 3, 2009

Hi all, I am pleased to announce the first release of visvis, a Python visualization library for of 1D to 4D data. Website: http://code.google.com/p/visvis/ Discussion group: http://groups.google.com/group/visvis/ Since this is the first release, it hasn't been tested on a large scale yet. Therefore I'm specifically interested to know whether it works for everyone. === Description === Visvis is a pure Python visualization library that uses OpenGl to display 1D to 4D data; it can be used from simple plotting tasks to rendering 3D volumetric data that moves in time. Visvis can be used in Python scripts, interactive Python sessions (as with IPython or IEP) and can be embedded in applications. Visvis employs an object oriented structure; each object being visualized (e.g. a line or a texture) has various properties that can be modified to change its behaviour or appearance. A Matlab-like interface in the form of a set of functions allows easy creation of these objects (e.g. plot(), imshow(), volshow()). Regards, Almar

4 8

Dijkstra's algorithm on a lattice
by Zachary Pincus Dec. 3, 2009

Dec. 3, 2009

Hi all, A bit off-topic, but before I write some C or cython to do this, I thought I'd ask to see if anyone knows of existing code for the task of finding the shortest (weighted) path between two points on a lattice. Specifically, I have images with "start" and "end" pixels marked and I want to find the path through the image with the lowest integrated intensity. Trivial but tedious to implement, so if anyone has some good tips I'd be happy to know. (There's already a left-to-right- shortest-path-finder in the image scikit repository, but that's not quite what I need.) Thanks, Zach

5 14

scikits.timeseries question
by Christopher Barker Dec. 2, 2009

Dec. 2, 2009

HI all, Maybe I'm missing something, but I can't seem to get this to work as I'd like. I have a bunch of data that is indexed by "day since Jan 1, 2001". It seemed I should be able to do a DateArray like this: In [40]: import scikits.timeseries as ts In [41]: sd = ts.Date(freq='D', year=2001, month=1, day=1) In [42]: sd Out[42]: <D : 01-Jan-2001> In [43]: da = ts.date_array((1,2,3,4), start_date=sd) In [44]: da Out[44]: DateArray([1, 2, 3, 4], freq='U') but it looks like it didn't get the frequency ffomr teh start date, so I did: In [46]: da = ts.date_array((1,2,3,4), start_date=sd, freq='D') In [47]: da Out[47]: DateArray([01-Jan-0001, 02-Jan-0001, 03-Jan-0001, 04-Jan-0001], freq='D') Now it's got the frequency, but it's using year 0001, instead of 2001, which is the same as I get if I don't use a start_date at all. What am I missing? In [50]: ts.__version__ Out[50]: '0.91.3' -Chris -- Christopher Barker, Ph.D. Oceanographer Emergency Response Division NOAA/NOS/OR&R (206) 526-6959 voice 7600 Sand Point Way NE (206) 526-6329 fax Seattle, WA 98115 (206) 526-6317 main reception Chris.Barker(a)noaa.gov

4 14

isn't it a bug in scipy.sparse? + some questions
by Dmitrey Dec. 2, 2009

Dec. 2, 2009

Hi scipy.sparse developers and all other scipy users, I'm trying to take benefits for solving SLEs in FuncDesigner via involving scipy.sparse. Some examples are here http://openopt.org/FuncDesignerDoc#Solving_systems_of_linear_equations and example for sparse SLEs is here http://trac.openopt.org/openopt/browser/PythonPackages/FuncDesigner/FuncDes… It already works faster than using dense matrices, but I want to speedup it even more, so I have some questions and seems like bug report (scipy.__version__ 0.7.0): from scipy import sparse from numpy import * a=sparse.lil_matrix((3,1)) a[0:3,:] = ones(3) print a.todense() #prints [[ 1.] [ 0.] [ 0.]] while I expect all-ones Questions: 1) Seems like a[some_ind,:]=something works very, very slow for lil. I have implemented a workaround, but can I use a[some_ind,:] for another format than lil? (seems like all other ones doesn't support it). 2) What is current situation with matmat and matvec functions? They say "deprecated" but no alternative is mentioned. 3) What is current situation with scipy.sparse.linalg.spsolve? It says /usr/lib/python2.6/dist-packages/scipy/sparse/linalg/dsolve/linsolve.py:78: DeprecationWarning: scipy.sparse.linalg.dsolve.umfpack will be removed, install scikits.umfpack instead ' install scikits.umfpack instead', DeprecationWarning ) But I don't want my code to be dependent on a scikits module. Are there another default/autoselect solver for sparse SLEs? If no, which one would you recommend me to use as default for sparse SLEs - bicg, gmres, something else? Thank you in advance, D.

4 4

"profiling" a function
by Askey, Scott A Capt USAF AETC AFIT/ENY Nov. 30, 2009

Nov. 30, 2009

What are the tools available in Scipy for evaluating the (computational) cost of a function call? In particular I am solving nonlinear systems (with fsolve) and considering exact versus approximate Jacobians and trig functions versus their approximations. V/R Scott -----Original Message----- From: scipy-user-bounces(a)scipy.org on behalf of scipy-user-request(a)scipy.org Sent: Mon 11/30/2009 1:00 PM To: scipy-user(a)scipy.org Subject: SciPy-User Digest, Vol 75, Issue 70 Send SciPy-User mailing list submissions to scipy-user(a)scipy.org To subscribe or unsubscribe via the World Wide Web, visit http://mail.scipy.org/mailman/listinfo/scipy-user or, via email, send a message with subject or body 'help' to scipy-user-request(a)scipy.org You can reach the person managing the list at scipy-user-owner(a)scipy.org When replying, please edit your Subject line so it is more specific than "Re: Contents of SciPy-User digest..." Today's Topics: 1. chebfun (josef.pktd(a)gmail.com) 2. ANN: visvis (Almar Klein) 3. Parallel code (Jose Gomez-Dans) 4. Re: Parallel code (Robin) 5. Re: Parallel code (Sturla Molden) 6. Re: Parallel code (Sturla Molden) ---------------------------------------------------------------------- Message: 1 Date: Sun, 29 Nov 2009 20:42:08 -0500 From: josef.pktd(a)gmail.com Subject: [SciPy-User] chebfun To: SciPy Users List <scipy-user(a)scipy.org> Message-ID: <1cd32cbb0911291742m1d3f3ab8r886009c7fba3cab9(a)mail.gmail.com> Content-Type: text/plain; charset=ISO-8859-1 I just came by chance across this (for matlab) http://www2.maths.ox.ac.uk/chebfun/index.html http://www2.maths.ox.ac.uk/chebfun/license.html The documentation looks helpful for someone like me who doesn't know enough about what Chebyshev polynomials are good for. Josef ------------------------------ Message: 2 Date: Mon, 30 Nov 2009 10:05:23 +0100 From: Almar Klein <almar.klein(a)gmail.com> Subject: [SciPy-User] ANN: visvis To: SciPy Users List <scipy-user(a)scipy.org> Message-ID: <cc38d75f0911300105x5e7b8754gf0c6b21547de95d1(a)mail.gmail.com> Content-Type: text/plain; charset=UTF-8 Hi all, I am pleased to announce the first release of visvis, a Python visualization library for of 1D to 4D data. Website: http://code.google.com/p/visvis/ Discussion group: http://groups.google.com/group/visvis/ Since this is the first release, it hasn't been tested on a large scale yet. Therefore I'm specifically interested to know whether it works for everyone. === Description === Visvis is a pure Python visualization library that uses OpenGl to display 1D to 4D data; it can be used from simple plotting tasks to rendering 3D volumetric data that moves in time. Visvis can be used in Python scripts, interactive Python sessions (as with IPython or IEP) and can be embedded in applications. Visvis employs an object oriented structure; each object being visualized (e.g. a line or a texture) has various properties that can be modified to change its behaviour or appearance. A Matlab-like interface in the form of a set of functions allows easy creation of these objects (e.g. plot(), imshow(), volshow()). Regards, Almar ------------------------------ Message: 3 Date: Mon, 30 Nov 2009 12:05:57 +0000 From: Jose Gomez-Dans <jgomezdans(a)gmail.com> Subject: [SciPy-User] Parallel code To: SciPy Users List <scipy-user(a)scipy.org> Message-ID: <91d218430911300405w34759867x7145cab01938d6bd(a)mail.gmail.com> Content-Type: text/plain; charset="iso-8859-1" Hi! I want to run some code in parallel, and I have toyed with the idea of either using the multiprocessing module, or using ipython (which is quite easy to use). The main idea is to run a number of class methods in parallel (unsurprisingly!), fed with some arguments. However, these methods will need (read-)access to a rather large numpy array. Ideally (and since this is running on a SMP box), this could be a chunk of shared memory. I am aware of Sturla Molden's suggestion of using ctypes, but I guess that I was wondering whether some magic simple stuff is available off the shelf for this shared memory business? Thanks! J

2 1

stats, classes instead of functions for results MovStats
by josef.pktd＠gmail.com Nov. 30, 2009

Nov. 30, 2009

Following up on a question by Keith on the numpy list and his reminder that covariance can be calculated by the cross-product minus the product of the means, I redid and enhanced my moving stats functions. Suppose x and y are two time series, then the moving correlation requires the calculation of the mean, variance and covariance for each window. Currently in scipy stats intermediate results are usually thrown away on return (while rpy/R returns all intermediate results used for the calculation. Using a decorator/descriptor of Fernando written for nitime, I tried out to write the function as a class instead, so that any desired ( intermediate) calculations are only made on demand, but once they are calculated they are attached to the class as attributes or properties. This seems to be a useful "pattern". Are there any opinion for using the pattern in scipy.stats ? MovStats will currently go into statsmodels Below is the class (with cutting part of init), a full script is the attachment, including examples that test the class. about MovStats: y and x are tested for 2d, either (T,N) with axis=0 or (N,T) with axis=1, should (but may not yet) work for nd arrays along any axis (signal.correlate docstring) nans are handled by dropping the corresponding observations from the window, not adding any additional observations, not tested if a window is empty because it contains only nans, nor if variance is zero (kern is intended for weighted statistics in the window but not tested yet, I still need to decide on normalization requirements) requires scipy.signal, all calculations done with signal.correlate, no loops as often, functions are one-liners all results are returned for valid observations only, initial observations with incomplete window are cut bonus: slope of moving regression of y on x, since it was trivial to add still some cleaning and documentation to do usage: ms = MovStats(x, y, axis=1) ms.yvar ms.xmean ms.yxcorr ms.yxcov ... Josef class MovStats(object): def __init__(self, y, x=None, kern=5, axis=0): self.y = y self.x = x if np.isscalar(kern): ws = kern <... snip> @OneTimeProperty def ymean(self): ys = signal.correlate(self.y, self.kern, mode='same')[self.sslice] ym = ys/self.n return ym @OneTimeProperty def yvar(self): ys2 = signal.correlate(self.y*self.y, self.kern, mode='same')[self.sslice] yvar = ys2/self.n - self.ymean**2 return yvar @OneTimeProperty def xmean(self): if self.x is None: return None else: xs = signal.correlate(self.x, self.kern, mode='same')[self.sslice] xm = xs/self.n return xm @OneTimeProperty def xvar(self): if self.x is None: return None else: xs2 = signal.correlate(self.x*self.x, self.kern, mode='same')[self.sslice] xvar = xs2/self.n - self.xmean**2 return xvar @OneTimeProperty def yxcov(self): xys = signal.correlate(self.x*self.y, self.kern, mode='same')[self.sslice] return xys/self.n - self.ymean*self.xmean @OneTimeProperty def yxcorr(self): return self.yxcov/np.sqrt(self.yvar*self.xvar) @OneTimeProperty def yxslope(self): return self.yxcov/self.xvar

3 4

Parallel code
by Jose Gomez-Dans Nov. 30, 2009

Nov. 30, 2009

Hi! I want to run some code in parallel, and I have toyed with the idea of either using the multiprocessing module, or using ipython (which is quite easy to use). The main idea is to run a number of class methods in parallel (unsurprisingly!), fed with some arguments. However, these methods will need (read-)access to a rather large numpy array. Ideally (and since this is running on a SMP box), this could be a chunk of shared memory. I am aware of Sturla Molden's suggestion of using ctypes, but I guess that I was wondering whether some magic simple stuff is available off the shelf for this shared memory business? Thanks! J

3 3

BivariateSpline examples and my crashing python
by josef.pktd＠gmail.com Nov. 28, 2009

Nov. 28, 2009

I wanted to prepare some examples for the use of the Bivariate Spline classes, The second part of the attached script contains a translation of a scipy.tutorial example to using the 3 classes instead. However, this script keeps crashing on me. Initially I thought it is RectBivariateSpline, but now I think it might be matplotlib, TK backend and maybe my latest numpy build. I would like to know if the splines crash or if it is my current setup that causes the crash. It crashes in spyder, idle and when I close the windows when I run it on the commandline. The last might indicate that the problem is matplotlib related. Can someone run the script, preferably not in an interpreter where you want to keep your session alive? Josef

3 4