Mailman 3 February 2007 - SciPy-User

Re: [SciPy-user] Any Books on SciPy?
by Andres Gonzalez-Mancera 28 Feb '07

28 Feb '07

I agree that there is a huge void in proper Scipy documentation which is essential for broader adoption. I know you can always go to the source files but this is not something everybody will do. Although the Documentation has been growing over the past year we're still missing some central 'official' documentation or at least tutorial. While I'm on the subject, I bought Travis' book early last year. I understood that we should have received updates as they became available. I never heard or received anything. Have there been any updates in the past year or was I left out of the loop? The reason I ask this is because I understand there were substantial changes to numpy in the road to 1.0. Are these changes documented anywhere other than the developers mailing lists? Thank you very much, Andres On 2/28/07, scipy-user-request(a)scipy.org <scipy-user-request(a)scipy.org> wrote: > Message: 8 > Date: Tue, 27 Feb 2007 23:19:51 -0700 > From: "Fernando Perez" <fperez.net(a)gmail.com> > Subject: Re: [SciPy-user] Any Books on SciPy? > To: "SciPy Users List" <scipy-user(a)scipy.org> > Message-ID: > <db6b5ecc0702272219k3a5d67dft7d2f1980b86950d1(a)mail.gmail.com> > Content-Type: text/plain; charset=ISO-8859-1; format=flowed > > On 2/27/07, John Hunter <jdh2358(a)gmail.com> wrote: > > On 2/27/07, Robert Love <rblove(a)airmail.net> wrote: > > > Are there any good, up to date books that people recommend for > > > numerical work with Python? > > > > > > I see the book > > > > > > Python Scripting for Computational Science > > > Hans Petter Langtangen > > > > > > Does anyone have opinions on this? Is it current? Are there better > > > books? > > > > I specifically do not recommend this book -- I own it but in my > > opinion it is outdated and is more a collection of the author's > > personal idioms than the current common practice in the scientific > > python community. For numerical work in python most people use > > I happen to share John's opinion, and I also have a copy of this book. > While it's technically correct, well written and fairly comprehensive > (probably /too/ much, since it's a bit all over the map), I strongly > dislike his approach. Much of the book uses his custom, home-made > collection of scripts and tools, which you can only download if you go > to a site and type a word from a certain page in the book (a simple > 'protection' system). > > Now you have an unmaintained, unreleased (publicly), set of tools to > learn from that don't have any licensing explicitly specified. > > Oh, and a good chunk of the tools in his distribution (since I have > the book, I have the code) use Perl. Go figure (there's also a tcl > directory thrown in for good measure). > > One of Python's main strengths for scientific work is precisely the > openness and interoperability of the various tools, and we all do our > part to help that be the case. The fact that this book follows an > approach more or less orthogonal to those ideas makes me very much > uninterested in using it. > > > There is a lot more, particularly for domain specific stiff, but these > > links are good starting points. Unfortunately, there is no > > one-stop-shop for a guide to scientific computing in python - Travis' > > documentation is the closest thing we have but it pretty much just > > covers numpy which is *the* core package. Fernando Perez and I have a > > very brief and limited started guide covering multiple packages > > (ipython, numpy, matplotlib, scipy, VTK) but I don't have the PDF > > handy (Fernando, do you have the roadshow doc handy?). > > Well, you asked for it :) > > http://amath.colorado.edu/faculty/fperez/tmp/py4science.pdf > > It's worth stressing, in the strongest possible terms, that this > should NOT, in any way, shape or form, be considered anything beyond a > pre-pre-alpha, pre-draft of a project for a possible book :) Besides, > it's already outdated in several important places (numpy, mayavi, no > TVTK,...). > > After all I said about the Langtangen book, at least it's a real one. > Our pdf draft is most certainly not. So if you need a book, with all > of its limitations, Langtangen's is currently the only game in town > that covers the whole spectrum of python for scientific computing. If > John and I ever end up stranded on a desert island for 3 months with > great internet access and poor diving gear, we might actually finish > ours, but don't hold your breath. > > Honestly, I think that today your best bet is: > > 1. Buy Travis' book. It's fantastic, has everything you need to know > about numpy, and you'll be supporting numpy itself. > > 2. Print Perry Greenfield's tutorial > (http://new.scipy.org/wikis/topical_software/Tutorial). I think he's > updating it for numpy now. > > 3. Have a look at some of the other info in > http://new.scipy.org/Documentation, in particular D. Kuhlman's course > is very nice. > > > Regards, > > f > -- Andres Gonzalez-Mancera Biofluid Mechanics Lab Department of Mechanical Engineering University of Maryland, Baltimore County andres.gonzalez(a)umbc.edu 410-455-3347

3 2

Re: [SciPy-user] Numpy eBook
by Paul Ray 28 Feb '07

28 Feb '07

On Feb 28, 2007, at 11:50 AM, scipy-user-request(a)scipy.org wrote: > While I'm on the subject, I bought Travis' book early last year. I > understood that we should have received updates as they became > available. I never heard or received anything. Have there been any > updates in the past year or was I left out of the loop? The reason I > ask this is because I understand there were substantial changes to > numpy in the road to 1.0. Are these changes documented anywhere other > than the developers mailing lists? I got one update in late December 2005, and none since then. I have certainly been hoping for a post-1.0 revision as well. Cheers, -- Paul

2 1

Optimizers and the number of iterations
by Nils Wagner 28 Feb '07

28 Feb '07

Hi all, AFAIK, only the number of function and gradient evaluations are directly available via full_output=1. Is there a better way to get the number of iterations ? What I have used so far, is x_opt, allvec=optimize.fmin_cg(func,x_0,retall=1) print 'The number of iterations is',shape(allvec)[0]-1 Nils

1 0

Transforming 1-d array to 2-d array
by Rich Shepard 28 Feb '07

28 Feb '07

I have this array (called barEco): [[ 2.29075869 2.12453058 3.06339111 2.88526612 3.32199956 2.96319486 3.12649018 2.81580625 3.23207315 2.493608 2.49459335 2.86643834 2.77111816 2.26500627 2.4902972 2.81149761 2.46213192 2.80432329 2.86150888 3.1135404 1.96135592 3.34577184 3.27458386 2.90845738 2.88745987 2.51245188 2.91666234 2.97438117]] and I want to convert it to: [[ -- 2.29075869 2.12453058 3.06339111 2.88526612 3.32199956 2.96319486 3.12649018] [ -- -- 2.81580625 3.23207315 2.493608 2.49459335 2.86643834 2.77111816] [ -- -- -- 2.26500627 2.4902972 2.81149761 2.46213192 2.80432329] [ -- -- -- -- 2.86150888 3.1135404 1.96135592 3.34577184] [ -- -- -- -- -- 3.27458386 2.90845738 2.88745987] [ -- -- -- -- -- -- 2.51245188 2.91666234] [ -- -- -- -- -- -- -- 2.97438117] [ -- -- -- -- -- -- -- -- ]] with 1.00 as the diagonal. It appears that the eye() function is the tool, but when I try foo = eye(barEco,8,8,1) print foo python responds Traceback (most recent call last): File "/data1/eikos/scopingPage.py", line 184, in OnCalcWeights inpWts = functions.weightcalc() File "/data1/eikos/functions.py", line 164, in weightcalc foo = eye(barEco,8,8,1) File "/usr/lib/python2.4/site-packages/numpy/lib/twodim_base.py", line 48, in eye m = equal(subtract.outer(arange(N), arange(M)),-k) TypeError: only length-1 arrays can be converted to Python scalars So, either I used eye() incorrectly, or that's not how to make the conversion. What should I be doing? Rich -- Richard B. Shepard, Ph.D. | The Environmental Permitting Applied Ecosystem Services, Inc. | Accelerator(TM) <http://www.appl-ecosys.com> Voice: 503-667-4517 Fax: 503-667-8863

2 10

Do I have LAPACK or not?
by Gary Pajer 28 Feb '07

28 Feb '07

When I compile I get the output below. Has LAPACK been found or not? I'm guessing yes, but I have some doubt ... thanks, gary ----------------------------------------------------- atlas_info: libraries f77blas,cblas,atlas not found in /usr/lib/atlas libraries lapack_atlas not found in /usr/lib/atlas libraries lapack not found in /usr/lib/sse2 libraries f77blas,cblas,atlas not found in /usr/lib libraries lapack_atlas not found in /usr/lib numpy.distutils.system_info.atlas_info /usr/lib/python2.4/site-packages/numpy/distutils/system_info.py:903: UserWarning: ********************************************************************* Could not find lapack library within the ATLAS installation. ********************************************************************* warnings.warn(message) FOUND: libraries = ['f77blas', 'cblas', 'atlas'] library_dirs = ['/usr/lib/sse2'] language = c define_macros = [('ATLAS_WITHOUT_LAPACK', None)] lapack_info: FOUND: libraries = ['lapack'] library_dirs = ['/usr/lib'] language = f77 FOUND: libraries = ['f77blas', 'cblas', 'atlas', 'lapack'] library_dirs = ['/usr/lib/sse2', '/usr/lib'] language = f77 define_macros = [('ATLAS_WITHOUT_LAPACK', None), ('ATLAS_INFO', '"\\"3.6.0\\""')] ATLAS version 3.6.0

2 2

Proper Use of NumPy's eig()
by Rich Shepard 28 Feb '07

28 Feb '07

I start the module with from numpy import linalg and I need to find the principal Eigenvector of the matrix E. But, when I try eigE = eig(E) python responds "NameError: global name 'eig' is not defined." Do I need another module from NumPy, should I explicitly cast E to a matrix, or am I calling the function incorrectly? Rich -- Richard B. Shepard, Ph.D. | The Environmental Permitting Applied Ecosystem Services, Inc. | Accelerator(TM) <http://www.appl-ecosys.com> Voice: 503-667-4517 Fax: 503-667-8863

6 14

no linalg module and failed check_integer test
by Jeremy Conlin 27 Feb '07

27 Feb '07

I just installed scipy from svn and I am having some problems. The most vexing is that I can no longer load the linalg package, at least I can't do anything with it. For example: Python 2.5 (r25:51918, Sep 19 2006, 08:49:13) [GCC 4.0.1 (Apple Computer, Inc. build 5341)] on darwin Type "help", "copyright", "credits" or "license" for more information. >>> import scipy >>> scipy.pkgload('linalg') >>> help(scipy.linalg) Traceback (most recent call last): File "<stdin>", line 1, in <module> AttributeError: 'module' object has no attribute 'linalg' >>> A = scipy.ones((10,10)) >>> scipy.linalg.eig(A) Traceback (most recent call last): File "<stdin>", line 1, in <module> AttributeError: 'module' object has no attribute 'linalg' >>> I didn't notice any errors during the installation. I know this isn't a lot of information to go on, but does anyone know what the problem is? Secondly, and this might be related, when I run scipy.test(1,10) I get the error shown below. Could this be the cause of my linalg problem? Thanks in advance, Jeremy test_init (scipy.io.tests.test_npfile.test_npfile) ... ok test_parse_endian (scipy.io.tests.test_npfile.test_npfile) ... ok test_read_write_array (scipy.io.tests.test_npfile.test_npfile) ... ok test_read_write_raw (scipy.io.tests.test_npfile.test_npfile) ... ok test_remaining_bytes (scipy.io.tests.test_npfile.test_npfile) ... ok ====================================================================== ERROR: check_integer (scipy.io.tests.test_array_import.test_read_array) ---------------------------------------------------------------------- Traceback (most recent call last): File "/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/scipy/io/tests/test_array_import.py", line 55, in check_integer from scipy import stats File "/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/scipy/stats/__init__.py", line 7, in <module> from stats import * File "/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/scipy/stats/stats.py", line 190, in <module> import scipy.special as special File "/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/scipy/special/__init__.py", line 8, in <module> from basic import * File "/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/scipy/special/basic.py", line 8, in <module> from _cephes import * ImportError: dlopen(/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/scipy/special/_cephes.so, 2): Symbol not found: ___dso_handle Referenced from: /Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/scipy/special/_cephes.so Expected in: dynamic lookup ---------------------------------------------------------------------- Ran 554 tests in 2.424s FAILED (errors=1) <unittest.TextTestRunner object at 0x2a49fb0>

2 4

Using mpfit
by Jared Peacock 27 Feb '07

27 Feb '07

I'm trying to use mpfit but I can't seem to get it to work. I'm testing the code using this program, which is similar to the example from the mpfit text: import mpfit2 import Numeric x=Numeric.arange(1.,stop=100.) p0=[5.7, 2.2, 500., 1.5, 2000.] y=6.+3.5*x+200.*x**2+2.7*Numeric.sqrt(x)+2700.*Numeric.log(x) err=.1 def myfunct(p,fjac=None,x=None,y=None,err=None): f=p[0]+p[1]*x+p[2]*x**2+p[3]*Numeric.sqrt(x)+p[4]*Numeric.log(x) status=0 return [status,(y-f)/err] fa = {'x':x, 'y':y, 'err':err} m = mpfit2.mpfit(myfunct, xall=p0, functkw=fa) print 'status = ', m.status #if (m.status <= 0): print 'error message = ', m.errmsg print 'parameters = ', m.params But when I run it I get this error: m = mpfit2.mpfit(myfunct, xall=p0, functkw=fa) File "mpfit2.py", line 1007, in __init__ [self.status, fvec] = self.call(fcn, self.params, functkw) TypeError: unpack non-sequence I'm not sure what this means or how to fix it. Does anybody have and incite to this problem and using mpfit? J. Peacock

1 0

Combining Arrays
by Rich Shepard 27 Feb '07

27 Feb '07

I have two arrays: one is the upper triangular half with 1s in the diagonal, the other has the values for the lower triangular half. I also have an array that specifies how the values in the lower half array are to be assigned to rows and columns of the 2D array. However, as this is my first use of NumPy, I am stuck at how to put these together to form what I need. Here's the upper array (call it topArr): [[ 1. 2.29075869 2.12453058 3.06339111 2.88526612 3.32199956 2.96319486 3.12649018] [ 0. 1. 2.81580625 3.23207315 2.493608 2.49459335 2.86643834 2.77111816] [ 0. 0. 1. 2.26500627 2.4902972 2.81149761 2.46213192 2.80432329] [ 0. 0. 0. 1. 2.86150888 3.1135404 1.96135592 3.34577184] [ 0. 0. 0. 0. 1. 3.27458386 2.90845738 2.88745987] [ 0. 0. 0. 0. 0. 1. 2.51245188 2.91666234] [ 0. 0. 0. 0. 0. 0. 1. 2.97438117] [ 0. 0. 0. 0. 0. 0. 0. 1. ]] Here's the lower array (call it botArr): [ 0.4365366 0.47069221 0.32643563 0.34658848 0.30102352 0.33747359 0.31984748 0.35513807 0.30939894 0.40102534 0.40086694 0.348865 0.36086516 0.44149988 0.4015585 0.35568232 0.40615208 0.35659227 0.34946598 0.32117778 0.50985137 0.2988847 0.30538231 0.34382488 0.34632516 0.39801757 0.34285765 0.33620439] And, using 'udx = tril(ones([8,8])-eye(8)).nonzero()' this is udx: (array([1, 2, 2, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6, 7, 7, 7, 7, 7, 7, 7]), array([0, 0, 1, 0, 1, 2, 0, 1, 2, 3, 0, 1, 2, 3, 4, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 6])) What I need is the array: [[ 1.00 2.29 2.12 3.06 ... ] [ 0.44 1.00 2.82 3.23 ... ] [ 0.47 0.22 1.00 2.26 ... ] ... ]] How do I do this, please? Rich -- Richard B. Shepard, Ph.D. | The Environmental Permitting Applied Ecosystem Services, Inc. | Accelerator(TM) <http://www.appl-ecosys.com> Voice: 503-667-4517 Fax: 503-667-8863

2 2

A NifTI-1 I/O library for Numpy wrapped through ctypes
by Krish Subramaniam 27 Feb '07

27 Feb '07

Hello Folks I am new here. Whipped up a quick tutorial "A NifTI-1 I/O library for Numpy wrapped through ctypes" for beginners since I am not an advanced user myself. I have to thank the Numpy developers for their wonderful job. This is a write-up on how to get a library wrapped in ctypes and "translate" the data to a Numpy N-D object. First ever draft. Would it find a way into your wiki or something? so that people who are in the same background as mine ( scientific, wrap an existing C module, seriously contemplate using Scipy instead of Matlab) would benefit. You can find it here : http://krish.caltech.edu/niftiio.pdf ( Should I have mailed the devel list instead ? ) Regards Krish Subramaniam P.S ( I think my previous mail didn't get through. If this is the 2nd mail you are receiving, please pardon )

2 2