bug,
not worth looking into. linalg.svd is ok though, odd.
Summary: if you run linalg.lstsq on big arrays,
either check max |x|
or regularize, do lstsq( vstack( A, weight * eye(dim) ),
hstack( b, zeros(dim) ))
cheers
-- de
On 24/10/2012 19:59, josef.p...@gmail.com wrote:
> On Wed, Oct 24, 2012 at 1:33 PM, denis wrote:
>> Folks,
>> np.linalg.lstsq of a random-uniform A 50 x 32 with 3 columns all 0
>> returns x[:3] 0 as expected,
>> but 4 columns all 0 => huge x:
> lstsq
t 3)]
How is the code in the .dmg checking for python.org python ?
Thanks,
cheers
-- denis
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion
d matplotlib .dmg s too)
cheers
-- denis
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion
how to do that, but
"note that this is no longer an orthogonal diagonalization";
orthogonal can do it iff A and B commute.
cheers
-- denis
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion
ot;where is equivalent to ..." ?
Thanks,
cheers
-- denis
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion
Folks,
is there a way to speed up Array.take( floatindices.astype(int) ) ?
astype(int) makes a copy, floor() returns floats.
(Is there a wiki of NumPy one-liners / various tricks ?
would sure beat googling.)
Thanks,
cheers
-- denis
___
NumPy
ely cast to required type
cheers
-- denis
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion
aultdict
import numpy as np
__version__ = "2010 7jan denis"
def pairs( s, Len=2 ):
""" repeated non-overlapping pairs (substrings, subwords)
"abracadabra" -> ab ra [[0 7] [2 9]], not br
Len=3: triples, 4 ...
"""
to log a ticket -- experts please advise ?
cheers
-- denis
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion
cipy.sparse.linalg.dsolve',
DeprecationWarning)
scipy.stats error numpy.dtype does not appear to be the correct type object
cheers
-- denis
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion
import scipy.stats then works.
Similarly import scipy.cluster => trace
File "numpy.pxd", line 30, in scipy.spatial.ckdtree
(scipy/spatial/ckdtree.c:6087)
ValueError: numpy.dtype does not appear to be the correct type object
I like the naming convention xx_cython.so.
cheers
--
em in the process;
I realize that testing is hard work, no glory.)
cheers
-- denis
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion
existing GUIs and desiderata,
sort by sum(features) / time-to-write-a-manual (not time-to-hack).
Or,
o'er forms of doc let fools contest,
what's best written is the best.
cheers
-- denis
___
NumPy-Discussion mailing list
NumP
le / smooth: must exist, physicists ?
- adaptive binning, e.g. percentiles then uniform
- interpolate: fill holes, then *linear or spline
+= data is nice, but seems orthogonal to histogramming --
why not just subclass histogram ?
cheers
-- denis
__
On 04/05/2010 14:09, Neal Becker wrote:
> denis wrote:
>> Neal,
>> I like the idea of a faster np.histogram / histogramdd;
>> but it would have to be compatible with numpy and pylab
>> or at least a clear, documented subset (doc first).
>
> The poin
dots = np.vstack([ dot( x[j:j+4] .T, imatrix ) .T for j in ...])
cheers
-- denis
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion
(dot = inner ? math. ?))
but has a clean api.
@Anne, excellent description; can you find time to work up your notes
to a say 5-page intro ? Would sell like hotcakes
cheers
-- denis
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion
t64(1) + p
array([2, 2])
>>> np.int(1) + p
poly1d([1, 2])
>>> np.int64(1)*p
array([1, 1])
What's happening here? I'd expect operations with polynomials would
return a polynomial.
--
Denis
___
NumPy-Dis
Folks,
here's a simple adaptive interpolator;
drop me a line to chat about it
adalin2( func, near, nx=300, ny=150, xstep=32, ystep=16,
xrange=(0,1), yrange=(0,1), dtype=np.float, norm=abs )
Purpose:
interpolate a function on a regular 2d grid:
take func() where it changes ra
in(max(j, 0), m-1), min(max(k, 0), n-1)]
This works but is both ugly and 5* slower than plain a[j][k].
Is there a better way ?
(Sorry if this is a duplicate, must come up often.)
cheers
-- denis
___
NumPy-Discussion mailing list
NumPy-Discussion@s
Robert Kern gmail.com> writes:
> Looks good! Where can we get the code? Can this be specialized for 1D
functions?
Re code: sure, I'll be happy to post it if anyone points me to a real test
case or two, to help me understand the envelope -- 100^2 -> 500^2 grid ?
(Splines on regular grids are fa
Robert Kern gmail.com> writes:
> http://svn.scipy.org/svn/scikits/trunk/delaunay/scikits/delaunay/testfuncs.py
Thank you Robert, that looks nice.
I've put 1d adalin1.py in http://drop.io/denis_adalin ;
have 2d, but can someone please comment on {content, style, direction}
of this simple 1d (150 l
Does anyone have a program to generate a file with one line per Numpy function
/ class / method, for local grepping ?
It might be useful for any package with thousands of functions too.
(Grepping a Pypi summary to see "what the heck is ..." takes < 1 second.)
Sorry if this is a duplicate, must exi
denis bzowy t-online.de> writes:
>
> Does anyone have a program to generate a file with one line per Numpy function
> / class / method, for local grepping ?
Sorry I wasn't clear: I want just all defs, one per long line, like this:
...
PyQt4.QtCore.QObject.findChildren(type typ
Folks,
this simple timeit -> a largish speed ratio that surprised me --
but perhaps I've done something stupid ?
""" timeit one big vs many little a[ogrid] = f(ogrid)
Consider evaluating a function on an NxN grid, in 2 ways:
a) in one shot:
y,x = ogrid[0:N, 0:N]
a[y,x] = f(x,y)
b) piece
Added: an inline grid
y,x = np.ogrid[ j:j+n, k:k+n ]
a[ j:j+n, k:k+n ] = f(x,y)
is 3* faster than a[y,x] = f(x,y) for 256x256,
about the same for little 8x8 squares (on mac ppc.)
So ogrids are not "objects" -- you can't
g = xxgrid[j:j+n, k:k+n]
...
use g, pass it just like i
> Russell E. Owen wrote:
> > All the official numpy 1.3.0 Mac binaries are labelled "macosx10.5".
> > Does anyone know if these are backwards compatible with MacOS X 10.4
numpy-1.3.0-py2.5-macosx10.5.dmg works fine on macosx 10.4.11 ppc
(wit
jah gmail.com> writes:
>
> Hi,Suppose I have a set of x,y,c data ... matplotlib.pyplot.contour() ).
>
JAH, is griddata() working and fast enough for you ?
How many points are you contouring ?
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy
jah gmail.com> writes:
> Thanks all. Robert, griddata is exactly what I was looking for. David, I
think that should work too. And Denis, griddata is sufficiently fast that I am
not complaining---contouring about 1e6 or 1e7 points typically.
>
Fyinfo, take a look at http://yt.enzo
er do experts use ?
cheers
-- denis
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion
Le samedi 12 décembre 2009 à 01:55 -0800, T J a écrit :
> Is there a better way?
You may have a look at http://scipy.org/Cookbook/MultiDot
Several alternatives are proposed.
Cheers,
--
Denis
___
NumPy-Discussion mailing list
NumPy-Discuss
Folks,
numpy/scipy builds on my mac 10.4.9 exec g++ not gcc
but g++ assumes c++ even for .c files => lts of errors.
I changed the link /usr/bin/g++ --> gcc, must be a better way ?
Seems to me that g++ on numpy C is a bug:
numpy builds on macs should run either
gcc
or g++ -x c
Same for MNE-Python:
https://github.com/mne-tools/mne-python/blob/master/appveyor.yml
Denis
2015-03-05 20:42 GMT+01:00 Stefan van der Walt :
> Hi Chuck
>
> On 2015-03-05 10:09:08, Charles R Harris
> wrote:
> > Anyone familiar with appveyor <http://www.appveyor.com/>?
34 matches
Mail list logo