Mailman 3 June 2010 - SciPy-User

Using leastsq(), fmin(), anneal() to do a least squares fit
by Andreas 30 Jun '10

30 Jun '10

Hi there, I have an optimization problem in 12 variables. I first wrote a functino toBeMinimized(), which outputs these 12 variables as one array. Trying to solve this problem with leastsq(), I noticed that however i play around with the parameters, the function does not seem to find the global optimum. So I figured I'd try some other functions from scipy.optimize, starting with anneal(). I wrote a wrapper function around my original toBeMinimized(), doing nothing but call np.sum(toBeMinimized(params)**2). Now, however, the results I get from anneal vary widely, and don't seem to have anything in common with the results from leastsq(). Basically the same happens when I use fmin() instead of anneal(). I'm somewhat at a loss here. leastsq() seems to give the most consistent results, but still they vary too much to be too useful for me. Any ideas? Thanks for your insight, Andreas.

2 1

optimization w/ constraints
by Andreas 30 Jun '10

30 Jun '10

hi there, i need to minimize a function in 19 variables x_1, ... x_19, with constraints x_i * a = x_{i+8} i=4,...,11 so actually, it's a 12 variable problem, but some of the variables get multiplied with each other ... how would i do that using scipy? i noticed the scipy.optimize.slsqp, where i can specify the f_eqcons parameter. just wanted to make sure i'm actually on the right path ... thanks for you insight! cheers, andreas.

2 2

interpolation with inverse-distance weighting + KDTree
by denis 30 Jun '10

30 Jun '10

Folks, here's a tiny class Invdisttree for interpolation with inverse- distance weighting + KDTree. It's solid, pretty fast, local, works for scattered data in any number of dimensions, leverages the excellent KDTree module. Comments would be welcome; real test cases, 3+d, most welcome. (For interpolating 2d data to a fine uniform grid, matplotlib._delaunay.nn_interpolate_grid is ~ 10 times faster (on my old mac ppc One reason is that the dot( 1/dist, z[ix] ) takes over half the time; another may be that nn_grid caches the current triangle ? ) cheers -- denis import numpy as np from scipy.spatial import cKDTree as KDTree class Invdisttree: """ inverse-distance-weighted interpolation using KDTree: invdisttree = Invdisttree( X, z ) -- points, values interpol = invdisttree( q, k=6, eps=0 ) -- interpolate z from the 6 points nearest each q; q may be one point, or a batch of points """ def __init__( self, X, z, leafsize=10 ): self.tree = KDTree( X, leafsize=leafsize ) # build the tree self.z = z def __call__( self, q, k=6, eps=0 ): # k nearest neighbours of each query point -- self.distances, self.ix = self.tree.query( q, k=k, eps=eps ) interpol = [] # np.zeros( (len(self.distances),) + np.shape(z[0]) ) for dist, ix in zip( self.distances, self.ix ): if dist[0] > 1e-10: w = 1 / dist wz = np.dot( w, self.z[ix] ) / np.sum(w) # weight z s by 1/dist else: wz = self.z[ix[0]] interpol.append( wz ) return interpol

1 0

How can I build an rpm of lapack
by Nils Wagner 30 Jun '10

30 Jun '10

Hi all, SUSE (and Red Hat) regularly shipped versions of the BLAS library where some functions were missing. Hence I would like to build my own rpm's of lapack and blas. Where can I find some instructions to build rpms of lapack and blas ? Any pointer would be appreciated. Nils

2 4

ANN: Visvis version 1.3 (includes meshes and lighting)
by Almar Klein 29 Jun '10

29 Jun '10

Hi all, I am exited to announce version 1.3 of Visvis, the object oriented approach to visualization. Website: http://code.google.com/p/visvis/ Discussion group: http://groups.google.com/group/visvis/ Documentation: http://code.google.com/p/visvis/wiki/Visvis_basics The largest improvement is the Mesh class to represent triangular and quad meshes and surface data. The Axes class got a property to access 8 different light sources. These improvements enable numerous new possibilities to visualize data using Visvis. Further changes include the introduction of polar plotting and 3D bar charts. For a (more) complete list of changes see the release notes <http://code.google.com/p/visvis/wiki/releaseNotes>. === Description === Visvis is a pure Python library for visualization of 1D to 4D data in an object oriented way. Essentially, visvis is an object oriented layer of Python on top of OpenGl? <http://code.google.com/p/visvis/w/edit/OpenGl>, thereby combining the power of OpenGl?<http://code.google.com/p/visvis/w/edit/OpenGl>with the usability of Python. A Matlab-like interface in the form of a set of functions allows easy creation of objects (e.g. plot(), imshow(), volshow(), surf()). Regards, Almar

1 0

f2py: questions on array arguments
by Ramon Crehuet 29 Jun '10

29 Jun '10

Dear all, I have a couple of question on f2py. 1. The follwing fortran function: function f2(x,y) implicit none real,intent(in):: x,y real, dimension(3):: f2 f2(1)=x+y**2 f2(2)=sin(x*y) f2(3)=2*x-y end function f2 gives a segmentation fault when called from python if it is not in a fortran module. If it is contained in a fortran module, it works fine and returns an array. That makes sense because fortran modules automatically generate an interface. However, I don't see that reflected in the .pyf files generated by f2py. So, is there a way to "correct" the function outside the module to work with f2py? 2. I have read in several posts that automatic arrays do not work with f2py. So that something like: real function trace(m) real, dimension(:,:), intent(in) :: m Has to be converted into: real function trace2(m,n) integer, intent(in) :: n !f2py integer,intent(hide),depend(m) :: n=shape(m,0) real, dimension(n,n), intent(in) :: m which works fine with f2py but is not nice in fortran. I've tried to do something like: real function trace(m) !f2py integer,depend(m) :: n=shape(m,0) !f2py real, dimension(n,n), intent(in) :: m real, dimension(:,:), intent(in) :: m But it does not work. Is there a workaround to avoid passing the dimension of the matrix as a fortran argument? Thanks in advance! Ramon

1 0

Re: [SciPy-User] Matlab trademark - was: Re: SciPy-User Digest, Vol 82, Issue 49
by Joe Harrington 29 Jun '10

29 Jun '10

I am copying this to scipy-dev, where documentation policy discussions belong, as it proposes a policy decision on the use of trademarked words in the documentation. I have exchanged email with Natasha Hellerich in UCF's Office of General Counsel about how to refer to MATLAB® in our documentation. As noted previously, our use, without their specific permission, falls under "fair use". Fair use requires good faith, which in turn requires that we follow the trademark holder's (reasonable) requests for how to use the term and acknowledge that the trademark is theirs. It also requires that we restrict ourselves to descriptive use of the trademarked term. Calling a module "matlab", even as part of "scipy.io.matlab", might not be in line with the latter requirement, since the thing referred to is not a product of The MathWorks®, but rather our own module for reading the MATLAB file format. Possibilities that seem more nominative to me include "matlabcompat", "matlabfilecompat", or something else that indicates that the module is not MATLAB nor any part of it and was not written by/at The MathWorks. I have not consulted a lawyer about this specific issue but that might be advisable once the community has decided which naming alternatives would be OK. The advantage of doing this now would be to have some time with a deprecation warning on the old name. If we made the change in response to a cease-and-desist letter, we might not have that luxury. Picking a naming convention that works with other software's file formats that we could implement in the future (e.g., for IDL® save files; IDL is a registered trademark of ITT, Inc.) would be good. The two relevant messages from our lawyer (minus lengthy quoted discussion that appeared after her signature) are below, forwarded with her permission. The first includes instructions for the use of "MATLAB" and the text of the trademark statement, which differs from that proposed earlier on scipy-user. Note that the US URL for the page she references is: http://www.mathworks.com/company/pressroom/editorial_guidelines.html I tried hard to get her to agree that we did not need to use the R-in-a-circle symbol at all. However, her second email below notes a case in which someone was forced by a US court to include the R-in-a-circle symbol in a fair use of someone else's registered trademark: G.D. Searle & Co. v. Hudson Pharm Corp 715 F.2d 837, 839( 3rd Cir. 1983) However, use of the symbol is only required on the first use in a document, at least in this case. Regarding placement of the first use of the term and the trademark statement, NumPy is a single entity imported into Python in a single command: import numpy as np The docs are separately available as a single PDF and a book-format HTML document tree. Each entity is thus a single item requiring a single trademark acknowledgement statement and R-in-a-circle symbol. In the PDF and HTML, the right place for the trademark statement and the use with the ® symbol is in the front matter. Then, all the uses in the main text, including all docstrings, are simply "MATLAB", unadorned by the ® symbol. I would suggest putting the trademark statement on page 1 of the PDF version, below the release number and date, right before Chapter 1, but any location would be fine as long as it is before chapter 1 and before any other use of the term. I would also suggest adding a copyright statement and an appropriate Creative Commons or loosely similar license to the PDF and HTML. The help() function in the software itself is merely an index browser into the collection of docs, capable of jumping around in the docs at random but not of actualy reordering the docs. The notion of "first" seems best addressed by the help(np) page (np.__doc__), since the PDF/HTML front matter does not exist in the help() system and since that page offers somewhat of an index into the rest of the docs. I suggest ending np.__doc__ with: The NumPy documentation occasionally refers to MATLAB®, which is a registered trademark of The MathWorks, Inc. The UCF lawyer's recommendation seems well in line with the web sites cited earlier in this thread. If anyone has any reason to object, now is the time to do so. Otherwise, I propose that we make the lawyer's recommendation our policy. Note that this is all based on US trademark law. If the law is different in your country, please speak up now so we can see if there is a policy that satisfies all countries' laws. Thanks, --jh-- Prof. Joseph Harrington Planetary Sciences Group Department of Physics MAP 414 4000 Central Florida Blvd. University of Central Florida Orlando, FL 32816-2385 jh(a)physics.ucf.edu planets.ucf.edu Date: Mon, 21 Jun 2010 13:14:10 -0400 From: "Natasha Hellerich" <nhelleri(a)mail.ucf.edu> To: "gcounsel" <gcounsel(a)mail.ucf.edu>, "Tanya Perry" <tperry(a)mail.ucf.edu>,<jh(a)physics.ucf.edu> Subject: Re: Fwd: use of TM and R in computer documentation In-Reply-To: <wl839wgtki4.fsf(a)glup.physics.ucf.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Disposition: inline Just as is stated on their website http://www.mathworks.co.uk/company/pressroom/editorial_guidelines.html MATLAB® - MATLAB should always be written with all letters in uppercase. Use the ® symbol on the first reference. Also, the statement MATLAB® is a registered trademark of The MathWorks, Inc. should be included. Please do not include any additional statements that were included on your previous email, such as "We use this trademark without permission from The MathWorks, etc." I specifically recommend AGAINST this statement. With respect to the computer documentation the employee is writing, I don't really know what this looks like, so it is difficult for me to judge whether to consider that as a single entity for purposes of following The MathWorks, Inc. guidelines with respect to using the ® symbol on the first reference. You are in a better position to judge whether this documentation constitutes a single entity. If you consider it as such, you can then make that argument. Also, not knowing the particulars surrounding this, my advice is limited to the trademark/symbol usage issues raised. Natasha Date: Wed, 16 Jun 2010 13:24:36 -0400 From: "Natasha Hellerich" <nhelleri(a)mail.ucf.edu> To: <jh(a)physics.ucf.edu> Cc: "gcounsel" <gcounsel(a)mail.ucf.edu>, "Tanya Perry" <tperry(a)mail.ucf.edu> Subject: Re: Fwd: use of TM and R in computer documentation In-Reply-To: <wl8zkz5s8qr.fsf(a)glup.physics.ucf.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Disposition: inline Good afternoon, With respect to your follow-up question, I can offer the following information: As previously recommended, I took a look at the MathWorks website. They in fact provide the specific guidelines as to how they wish for others to refer to their trademarks, including when and how to include the circle R symbol. Please see the link below. http://www.mathworks.co.uk/company/pressroom/editorial_guidelines.html I recommend looking to other companies' web sites for guidance as well or contacting the entity at issue for the use of trademarks other than those owned by MathWorks. Generally, 17 United States Code (USC) Section 107 sets forth the concept of fair use. Fair use would be a defense to someone using another person's trademark. Fair use is the legal concept that would allow you to use another's trademark, within the parameters of fair use. So the fair use defense is one way to argue that you are not infringing a trademark, but it would require that the use is descriptive and in good faith and is used to describe the goods and services of another. If someone knows the other trademark is registered, then good faith use would show the use of the symbol. There is at least one court case G.D. Searle & Co. v. Hudson Pharm Corp 715 F.2d 837, 839( 3rd Cir. 1983) where a defendant in a trademark infringement case was ordered by the court to refer to his competitors' products using the registration symbol of R with a circle. I will ask Tanya Perry, our paralegal, to research additional case law regarding rules with respect to using another person's or entity's trademark in research articles or other commentary, as well as in other scenarios, i.e. whether there is additional case law out there that discusses the use of appropriate trademark symbols every time the trademark is mentioned vs. maybe just once at the beginning. Again, if a company already sets forth its own specific rules regarding use of trademark symbols when referring to their trademarks (such as MathWorks does via their web site), then those rules obviously should be followed. Tanya - please advise on what you can find regarding this issue. Thanks, Natasha

4 5

Fwd: [Scipy-tickets] [SciPy] #1212: Single precision FFT insufficiently accurate.
by Sebastian Haase 28 Jun '10

28 Jun '10

this workaround seems in-acceptable if single-precision was used (by the user of SciPy) because of memory constrains .... !!! Regards, -Sebastian Haase On Sun, Jun 27, 2010 at 2:53 PM, <scipy-tickets(a)scipy.org> wrote: > #1212: Single precision FFT insufficiently accurate. > ---------------------------+------------------------------------------------ > Reporter: rgommers | Owner: somebody > Type: defect | Status: new > Priority: normal | Milestone: 0.8.0 > Component: scipy.fftpack | Version: 0.7.0 > Keywords: | > ---------------------------+------------------------------------------------ > > Comment(by pv): > > Work-around for 0.8.x committed in r6570 > > -- > Ticket URL: <http://projects.scipy.org/scipy/ticket/1212#comment:1> > SciPy <http://www.scipy.org> > SciPy is open-source software for mathematics, science, and engineering. > _______________________________________________ > Scipy-tickets mailing list > Scipy-tickets(a)scipy.org > http://mail.scipy.org/mailman/listinfo/scipy-tickets >

6 16

Installing on osx 10.6, py 2.6
by Vincent Davis 28 Jun '10

28 Jun '10

Ok I am at it again. trying to build/install from source. It fails, full details available here http://pastebin.com/KT08eLiZ The last bit is below. I am trying to learn more about this but not there is a lot I don't know. Thanks Vincent ld: in /usr/lib/libSystem.B.dylib, missing required architecture ppc64 in file for architecture ppc64 collect2: ld returned 1 exit status lipo: can't open input file: /var/folders/2f/2fiXYQSSE+CgAzDQPp9+k++++TI/-Tmp-//ccs45Qej.out (No such file or directory) error: Command "/usr/local/bin/gfortran -Wall -arch ppc -arch i686 -arch x86_64 -arch ppc64 -Wall -undefined dynamic_lookup -bundle build/temp.macosx-10.4-x86_64-2.6/build/src.macosx-10.4-x86_64-2.6/scipy/fftpack/_fftpackmodule.o build/temp.macosx-10.4-x86_64-2.6/scipy/fftpack/src/zfft.o build/temp.macosx-10.4-x86_64-2.6/scipy/fftpack/src/drfft.o build/temp.macosx-10.4-x86_64-2.6/scipy/fftpack/src/zrfft.o build/temp.macosx-10.4-x86_64-2.6/scipy/fftpack/src/zfftnd.o build/temp.macosx-10.4-x86_64-2.6/build/src.macosx-10.4-x86_64-2.6/scipy/fftpack/src/dct.o build/temp.macosx-10.4-x86_64-2.6/build/src.macosx-10.4-x86_64-2.6/fortranobject.o -Lbuild/temp.macosx-10.4-x86_64-2.6 -ldfftpack -lfftpack -lgfortran -o build/lib.macosx-10.4-x86_64-2.6/scipy/fftpack/_fftpack.so" failed with exit status 1 MacBookPro-new-2:scipy-vmd-dev vmd$

1 1

scipy.sparse.csr_matrix.matmat deprecation question
by Andrew Schein 27 Jun '10

27 Jun '10

I would like to perform a matrix multiplication of the form A * B where A is dense and B is sparse CSR or COO. Does scipy.sparse have this capability and will it in the future? How fast is the scipy implementation in comparison to INTEL MKL? It appears that there is a .matmat function that has been deprecated. Does this reflect a retreat, or is the functionality found in some other place? Thanks, Andrew -- Andrew I. Schein www.andrewschein.com

2 1