scipy / scipy.org

Scipy.org web site source code

scipy.org

The SciPy website is built on the scientific-python-hugo-theme and served using Hugo.

Build

After installing Hugo and Dart Sass, build locally:

git submodule update --init
make html

The pages are in public/.

To run the development server, which hosts the site and recompiles automatically on edits:

make serve

Team lists

To update the teams, the team_query.py provided by the theme is used. It needs a GitHub token with read:org permissions. The token has to be exported as GH_TOKEN.

GH_TOKEN=xxxxxxxxxx make teams

Analytics

The service Plausible.io is used to gather simple and privacy-friendly analytics for the site. The dashboard can be accessed here. The Scientific-Python community is managing the account.

scipy.org's Issues

Trouble using bioread to convert files from .acq to .mat

Hello,

I am a new Python user (2.7.10) attempting to use bioread (https://pypi.python.org/pypi/bioread/0.9.5) to convert files from aqknowledge to matlab. I am using a 64-bit PC, and I have downloaded Matlab, Python, numpy, scipy and bioread. Can someone walk me through the installation process for this package? I can't seem to get it to work.

Thank you so much for your help!!

IPython FITS file plotting problem

Dear all,

I am facing a problem while running a script ( please find the attached files in the zip folder - I am attaching both the ipython notebook version and also the python script ).

I am trying to plot an array of values, write it into a FITS file format, read it back again and plot it ---> I don't get the same plots!

If you could please help me with this it would be great.

The following are the versions of my packages and compiler:

matplotlib : '2.0.0b1'

numpy : '1.11.0'

astropy : u'1.1.2'

python : 2.7

Sincerely,
Anik Halder
python_fits_file_problem.zip

Error in numpy.random.gumbel doc examples

In the second example code at

http://docs.scipy.org/doc/numpy/reference/generated/numpy.random.gumbel.html

The following line

beta = np.std(maxima)*np.pi/np.sqrt(6)

should be

beta = np.std(maxima) * np.sqrt(6) / np.pi

Rgds

SciPy website FAQ "Where to get help" empty

Hi,

The FAQ entry on the scipy website has a "Where to get help" entry, but it's empty. I expected to see some contact details, a mailing list or something like that.

MathJax fails to load, due to mixed content rules

In Firefox 23 & above (and possibly other browsers), pages served over HTTPS will refuse to load content served over HTTP. Since MathJax is loaded over HTTP in the documentation, formulas fail to render.

For example, see this page: https://docs.scipy.org/doc/scipy/reference/tutorial/linalg.html#decompositions

dev docs not updated since ~4 months

I was looking for 'differential evolution' docs on
http://docs.scipy.org/doc/scipy-dev/reference/optimize.html
but these dev docs appear to not have been updated since March 16. If someone has time, could the website possibly be updated to reflect, for example, the changes to https://github.com/scipy/scipy/blob/master/scipy/optimize/__init__.py over the last few months?

Link Rot

Algorete Loopy link on install.rst is no longer working -- suggest removing it from the list

cKDTree

Hi all,
I have two array (A,B) containing: ID, x, y, z of the same number of points but slightly differents.
I would like to have an array in which each row has the ID x y z of the two nearest points of the two array.
At the moment I have this:

import numpy as np
from scipy.spatial import cKDTree
A = np.loadtxt('A.txt')
B = np.loadtxt('B.txt')
tree = cKDTree( B[:,[1,2,3]] )
d, inds = tree.query( A[:,[1,2,3]], k=1, p=np.inf, eps=0.0)
A_new = A[inds]
xyz_near = np.hstack(( B[:,0:4], A_new[:,0:4] ))

But the array "xyz_near" does not contain the right couple:

As you can see the first two row are right but not the next three.

If I do the same thing in matlab with dsearchn:

that is right.
I have tried to change p to 1, 2 and np.inf but same result.

Thanks

improve rendering of LaTeX math

tag: wishlist

Currently, TeX math looks resterized and clunky as compared to eg IPython notebooks or MathOverflow.

Updating the FAQ regarding FITS

FAQ, Section "I want to load data from a text file. How do I make this code more efficient?", regarding the FITS format: the pyfits package is no longer independently developed and has been subsumed into the astropy project.

Solving the Discrete Lyapunov Equation does not work with matrix input

Function scipy.linalg.solve_discrete_lyapunov does not work when the second argument (Right-hand side square matrix) is of type numpy.matrixlib.defmatrix.matrix. Example:

>>> from numpy import array, matrix
>>> from scipy.linalg import solve_discrete_lyapunov
>>> a = matrix([[0, 1],[-1/2, -1]])
>>> b = matrix([0, 3]).T
>>> solve_discrete_lyapunov(a, b*b.T)
Traceback (most recent call last):
  File "<ipython-input-244-c3af118b7628>", line 5, in <module>
    x = solve_discrete_lyapunov(a, b*b.T)
  File "C:\Miniconda3\lib\site-packages\scipy\linalg\_solvers.py", line 145, in solve_discrete_lyapunov
    x = solve(lhs, q.flatten())
  File "C:\Miniconda3\lib\site-packages\scipy\linalg\basic.py", line 80, in solve
    raise ValueError('incompatible dimensions')
ValueError: incompatible dimensions

Casting the second argument to an array (numpy.ndarray) solves the problem, but this is somewhat unexpected behaviour.

>>> solve_discrete_lyapunov(a, array(b*b.T))
array([[ 21.6, -14.4],
       [-14.4,  21.6]])

wiki.scipy.org is 403 Forbidden

please fix

Generated Documentation Images 404

On this page, the links on the (Source code, png, pdf) all result in 404 errors.

I haven't checked any other pages.

error in the doc of scipy.signal.convolve

Hi,

in the doc http://docs.scipy.org/doc/scipy/reference/tutorial/signal.html#filtering
for the function scipy.signal.convolve it says:

"If the flag has a value of ‘same’ then only the middle K values are returned starting at y[...] so that the output has the same length as the largest input"

But when I run it it seems that the output has the same length as the first input.

x = np.array([1.0, 2.0, 3.0])
h = np.array([0.0, 1.0, 0.0, 0.0, 0.0])

len(signal.convolve(h, x, 'same'))
5
len(signal.convolve(x, h, 'same'))
3

pip upgrade of scipy requires BLAS and LAPACK on Fedora/CentOS

https://www.scipy.org/install.html#fedora advises users to use yum to install scipy and then pip to upgrade it. However, upgrading via pip fails unless the user has BLAS and LAPACK devel packages installed.

Per http://stackoverflow.com/a/25147728/500207, the solution is to first

$ sudo yum install blas-devel lapack-devel

and only then sudo pip install --upgrade scipy.

scipy.stats

hello, i use python 3.4. But when i use scipy.stats ,i am comfused to unable to load this ?

scipy.optimize.basinhopping: accept_test returning numpy.bool_ raises an error

Dear Scipy team

When using scipy.optimize.basinhopping, if the accept_test function returns an object of type numpy.bool_, an error occurs. Is this an intentional feature? If not, I think it can be removed allowing accept_test to return numpy.bool_ object.

Thanks
Ajay

List of Scikits 404

scikits.appspot.com returns 404, so the list of scikits (scikits.appspot.com/scikits) cannot be accessed.

mail.scipy.org is down?

scipy.org site down : Unable to connect

Not sure if this is the place to inform you, but seems that the scipy.org has been down for a couple hours.

error in numpy mask doc

In http://docs.scipy.org/doc/numpy/reference/maskedarray.generic.html#the-numpy-ma-module
Indexing and slicing, the example is:

x = ma.array([1, 2, 3, 4, 5], mask=[0, 1, 0, 0, 1])
mx = x[:3]
mx
masked_array(data = [1 -- 3],
mask = [False True False],
fill_value = 999999)
mx[1] = -1
mx
masked_array(data = [1 -1 3],
mask = [False True False],
fill_value = 999999)

However, for mx, given we assigned mx[1] as -1 ,so the output is supposed to be unmasked for mx[1]. I tested it with Python 2.7.11, here is the output:

import numpy.ma as ma
x = ma.array([1, 2, 3, 4, 5], mask=[0, 1, 0, 0, 1])
mx = x[:3]
mx
masked_array(data = [1 -- 3],
mask = [False True False],
fill_value = 999999)

mx[1] = -1
mx
masked_array(data = [1 -1 3],
mask = [False False False],
fill_value = 999999)

Bug reports regarding documentation and web site

Maybe www/bug-report.rst should explain where and how people are supposed to file bug reports about the web site and the scipy/numpy documentation.

mstats.mannwhitneyu and stats.mannwhitneyu return inconsistent pvalues

Assuming sample_population and base_population are masked arrays:

mstats.mannwhitneyu(sample_population, base_population)

and

stats.mannwhitneyu(sample_population.compressed(), base_population.compressed())

Return the same mann-whitney statistic but differing p-values. The pvalue from mstats is twice the pvalue from stats suggesting that the mstats implementation is returning a 2 sided pvalue while the stats implementation returns the 1-sided pvalue - but there's nothing in the scipy documentation to indicate that this is the case.

SSL certificate expired

The SSL certificate has expired today, now browsers display warnings when browsing the site

*.scipy.org does not support any secure versions of TLS

scipy.org and docs.scipy.org web servers only support SSL v3 and TLS v1.0, neither of which is secure.

https://www.ssllabs.com/ssltest/analyze.html?d=docs.scipy.org

The situation is so bad that my browser refuses to connect to the server — clicking the "numpy functions by category" link from numpy.org homepage gives this:

Typo on website

On http://docs.scipy.org/doc/scipy/reference/sparse.linalg.html

one finds:

"Return a fuction for solving a sparse..."

Best,
-Matyas

Forbidden You don't have permission to access /Cookbook on this server.

At http://wiki.scipy.org/Cookbook I see

Forbidden

You don't have permission to access /Cookbook on this server.

mismatch between doc an source code of optimize.leastsq

http://docs.scipy.org/doc/scipy-0.14.0/reference/generated/scipy.optimize.leastsq.html#scipy.optimize.leastsq

For the parameter "maxfev" it is written that is is set to 100*(N+1) if unspecified. In the source code it is either set to 200*(n + 1) in case of "Dfun is None" or to 100*(n + 1) in the case of given "Dfun"

List of scikits page is broken

When accessing http://scikits.appspot.com/scikits I get

Traceback (most recent call last):
  File "/base/data/home/runtimes/python/python_lib/versions/1/google/appengine/ext/webapp/_webapp25.py", line 715, in __call__
    handler.get(*groups)
  File "/base/data/home/apps/s~scikits-hrd/1.387270795685241486/scikits.py", line 207, in get
    packages = sorted(Package.packages().values())
  File "/base/data/home/apps/s~scikits-hrd/1.387270795685241486/scikits.py", line 326, in packages
    results = server.search(dict(name="scikit"))
  File "/base/data/home/runtimes/python/python_dist/lib/python2.5/xmlrpclib.py", line 1147, in __call__
    return self.__send(self.__name, args)
  File "/base/data/home/runtimes/python/python_dist/lib/python2.5/xmlrpclib.py", line 1437, in __request
    verbose=self.__verbose
  File "/base/data/home/apps/s~scikits-hrd/1.387270795685241486/tools.py", line 230, in request
    response.headers)
ProtocolError: <ProtocolError for pypi.python.org/pypi: 403 >

Unused variable declaration in generated f2pywrappers2

I am using f2py for a project where the coding standards forbid the declaration of unused variables. This is enforced by automated builds which consider warnings as errors. As such, I am prevented from successfully compiling the *-f2pywrappers2.f90 code due to the declaration of an unused variable.

The variable "j" is declared in the wrappers generated to provide the dimensions of arrays. The declaration comes in the form of "integer r,i,j" but "j" is never used. I have tracked this down to "numpy/f2py/f90mod_rules.py", specifically line 52 in numpy version 1.10.2, in the definition of "fgetdims1". The declaration seems to have existed since at least numpy version 1.4.1. Removal of the declaration of "j", while leaving "r" and "i", eliminates the unused variable warnings.

Improving topical software

The topical software page is a direct translation of a wiki page.

• It is very long, and not very easy to read.
• It is not regularly updated --- authors of some projects add their packages there, but this does not seem to occur very commonly.
• It is far from comprehensive, not listing e.g. stuff on pypi

There are probably better ways to do it, potentially including
(i) splitting the long list to one metadata file per entry,
(ii) replacing sections by hierarchical tags,
(iii) automatically importing data from pypi (whole Science/Research category), using the local metadata for tagging.

FAQ entry about matmul operator is out of date

This one: Why not just have a separate operator for matrix multiplication?

Python 3.5 will have the @ operator suggested in PEP 465, if I'm not mistaken.

error in site documentation for pmf() method for Poisson distribution

Probability mass function method (pmf) for Poisson distribution claims first argument to be x, whereas the actual argument is k, as evident in

>>> help(poisson.pmf)
Help on method pmf in module scipy.stats._distn_infrastructure:

pmf(self, k, *args, **kwds) method of scipy.stats._discrete_distns.poisson_gen instance
    Probability mass function at k of the given RV.

    Parameters
    ----------
    k : array_like
        quantiles

http://www.scipy.org/ is down inaccessbile

Bounds not respected in scipy.optimize.minimize?

The following code did not produce the results that I expected. I don't know whether it's a bug or not, but I suspect that it is. Problematically (for me at least), the behavior is very different when undefined is set to None, 0.0, 1.0, etc. I expected that the bounds parameter would ensure that f(x) was not invoked with an invalid value of x. SLSQP is not the only minimizer that exhibits this problem, and other on the Internet appear to have received similar results.

import scipy.optimize
undefined = None
def f(x):
    if x > 1.0:
        print("x > 1.0, x=%12.10f" % x)
        return undefined
    print("f(%f)=1.0" % x)
    return 1.0
# The bounds are not properly respected!
print scipy.optimize.minimize(f, 1.0, method='SLSQP', bounds=[(0.0, 1.0)])

I don't necessarily expect a fix, I just wanted to report the issue.

Contact [email protected] not responsible

The contact: [email protected] listed at https://planet.scipy.org/ is not responsible.
Please update. I received this email:

-------- Forwarded Message --------
Subject: Fwd: http://wiki.scipy.org/Cookbook Forbidden
Date: Tue, 6 Oct 2015 11:41:22 +0300
From: Lisandro Dalcin [email protected]
To: Frank Breitling [email protected]

Dear Frank, I got two emails in the forward below, they may help you
to fix the issues in the scipy.org site.

---------- Forwarded message ----------
From: Brian Granger [email protected]
Date: 2 October 2015 at 22:01
Subject: Re: http://wiki.scipy.org/Cookbook Forbidden
To: Lisandro Dalcin [email protected]

Hi!

I think it is these folks:

Allen Parker [email protected] with a Cc to Didrik Pinte
[email protected]

Let me know if that doesn't work.

Cheers,

Brian

On Wed, Sep 30, 2015 at 12:56 PM, Lisandro Dalcin [email protected] wrote:

Dear Brian, how are you? I got the complaint I'm forwarding.

Basically, if you go to http://scipy.org/ and then you click in the
top-right icon (Blog), then you should be redirected to
https://planet.scipy.org/ , but the mpi4py docs show up. I have no
idea who is the admin of scipy.org these days, any tip about how to
proceed to get this fixed?

---------- Forwarded message ----------
From: Frank Breitling [email protected]
Date: 30 September 2015 at 14:16
Subject: Re: http://wiki.scipy.org/Cookbook Forbidden
To: Lisandro Dalcin [email protected]

Your name is the first thing that appears at

https://planet.scipy.org/

You get there when you click on Blogs from (top right) on

http://scipy.org/

Then you should get this changed as soon as possible.

On 29.09.2015 13:15, Lisandro Dalcin wrote:

On 28 September 2015 at 17:02, Frank Breitling [email protected] wrote:

Alright, but then why is your name listed instead of the admin?
Maybe you should have somebody change this.

Where is my name listed? I never had admin access to that server, I
only had user-level access to the subdomain http://mpi4py.scipy.org/,
which I'm no longer using and goes down time to time. That subdomain
is there just to host documentation about one of my projects, but I
never handled the "wiki" subdomain.

Did you try to send an email to the scipy mailing list?
http://www.scipy.org/scipylib/mailing-lists.html

Lisandro Dalcin

Research Scientist
Computer, Electrical and Mathematical Sciences & Engineering (CEMSE)
Numerical Porous Media Center (NumPor)
King Abdullah University of Science and Technology (KAUST)
http://numpor.kaust.edu.sa/

4700 King Abdullah University of Science and Technology
al-Khawarizmi Bldg (Bldg 1), Office # 4332
Thuwal 23955-6900, Kingdom of Saudi Arabia
http://www.kaust.edu.sa

Office Phone: [phone]

Brian E. Granger
Associate Professor of Physics and Data Science
Cal Poly State University, San Luis Obispo
@ellisonbg on Twitter and GitHub
[email protected] and [email protected]

citing.rst in wrong place

The citing.rst talks about citing all of the core projects, but is located under /scipylib/ which only applies to Scipy library.

mail, wiki, and blogs are down

http://wiki.scipy.org/
http://planet.scipy.org/
http://mail.scipy.org/

Wiki and blogs are down

Currently, neither http://wiki.scipy.org/ nor http://planet.scipy.org/ can be accessed.

Wiki and blogs are down (again)

Currently, neither http://wiki.scipy.org/ nor http://planet.scipy.org/ can be accessed.

Same issue as this one: #85

Please update the Debian version in the install page

The "Ubuntu & Debian" section of http://www.scipy.org/install.html page refers to Debian 7.0 which is an old version. The current stable Debian version is 8.3 and was released on Jan 23rd, 2016 [1]. Could you please update the page accordingly?

[1] - https://www.debian.org/releases/stable/

matrix multiplication goes wrong

I tried to multiply two matrix, but the result is wrong.
Here is the instance I used:

public_base = np.matrix([[-1612927239, 1853012542, 1451467945], [-2137446623, 2455606985, 1923480029], [2762180674, -3173333120, -2485675809]])
original_signature = np.matrix([[87398273893], [763829184], [118237397273]])
point = public_base * original_signature

and the answer I got is
[[-4827367651133834946]
[ 5600303079090794586]
[ 426332741103555793]]

but since the correct answer is

1.0e+19 *

3.2066
4.2494
-5.4914

The icons at the top of the page are too close

On my version of Firefox, with slightly larger fonts, the text under the different icons is running together, and the icons themselves look squashed together with excess space on the margins.

scipy-svn mailing list?

The https://mail.scipy.org/mailman/listinfo/scipy-svn mailing list seems to be not accepting mails sent to it from Github. The similar list https://mail.scipy.org/mailman/listinfo/numpy-svn on the other hand is accepting mail.

UnivariateSpline: interpolator(a) raises error, but interpolator.integrate(a,b) does not

I would like to create an interpolator which also integrates. The UnivariateSpline class seems like a simple solution to this problem.

My problem is that I would like it to return nan values outside of the bbox range. I came up with a solution involving a try/except statement. This works for the original interpolator function, but not for its integral (!)

My hacky solution is to evaluate the interpolator to catch the error, but then evaluate the integral. This seems like a bug to me though, so I thought I would report it.

At the same time, I will ask if there is a better way to perform this operation.

x=np.r_[0:10:10j]
interpolator=UnivariateSpline(x, x, k=1, ext=2)
xNew=np.r_[5:15:10j]

integral=[]
for d in xNew:
    # fill extrapolated values with nan
    try:
        _=interpolator(d)
        integral.append(interpolator.integral(0,d))
    except ValueError:
        integral.append(np.nan)

Overloaded ‘isnan(double&)’ is ambiguous when compiling with gcc 4.8.1 -std=c++11

This may be a numpy issue rather than a scipy issue, nevertheless, when installing scipy with the command "sudo -H pip install" I get the following compilation error:

In file included from scipy/special/Faddeeva.cc:122:0:
scipy/special/Faddeeva.cc: In function ‘std::complex Faddeeva::erf(std::complex, double)’:
/usr/local/lib/python2.7/dist-packages/numpy/core/include/numpy/npy_math.h:171:41: error: call of overloaded ‘isnan(double&)’ is ambiguous

System details:
Ubuntu 12.04.2 LTS
gcc (Ubuntu 4.8.1-2ubuntu1~12.04) 4.8.1
numpy.version = '1.10.1'

It appears numpy is trying to use the math.h version of isnan. I suspect somewhere else in the build process cmath.h has also been included resulting in the ambiguity.

http://wiki.scipy.org/ down

The Wiki is down (since 2 weeks).
http://wiki.scipy.org/ says:

Forbidden

You don't have permission to access / on this server.

Homepage not updated with recent news

I notice that there are commits for adding new news items for releases of e.g. scipy 0.17,
but they are not visible on the homepage itself. The latest news item there is from January 7.

NumPy tutorial down

Error on: http://wiki.scipy.org/Tentative_NumPy_Tutorial linked from www.numpy.org

Forbidden

You don't have permission to access /Tentative_NumPy_Tutorial on this server.

Update Building From Source on Windows

Placing issue in scipy.org that can then be closed by my PR. Links to numpy issue #5479 and #7104.