<div dir="ltr"><div>On behalf of the SciPy development team I'm pleased to announce the availability of <span><span><span><span><span class="">SciPy</span></span></span></span></span>
<span><span class="">0</span></span>.13.<span><span class="">0</span></span>.
This release contains some interesting new features (see highlights below) and half a year's worth of maintenance work. 65 people
contributed to this release.<br><br>Some of the highlights are:<br><br> - support for fancy indexing and boolean comparisons with sparse matrices<br> - interpolative decompositions and matrix functions in the linalg module<br>
- two new trust-region solvers for unconstrained minimization<br><br>This
release requires Python 2.6, 2.7 or 3.1-3.3 and NumPy 1.5.1 or greater.
Support for Python 2.4 and 2.5 has been dropped as of this release.<br>
<br>
Sources and binaries can be found at <a href="http://sourceforge.net/projects/scipy/files/scipy/0.13.0/">http://sourceforge.net/projects/scipy/files/scipy/0.13.0/</a>, release notes are copied below. <br></div><div><br>Enjoy,<br>
</div>Ralf<br><br><br><br><br>==========================<br>SciPy 0.13.0 Release Notes<br>==========================<br><br>.. contents::<br><br>SciPy 0.13.0 is the culmination of 7 months of hard work. It contains<br>many new features, numerous bug-fixes, improved test coverage and<br>
better documentation. There have been a number of deprecations and<br>API changes in this release, which are documented below. All users<br>are encouraged to upgrade to this release, as there are a large number<br>of bug-fixes and optimizations. Moreover, our development attention<br>
will now shift to bug-fix releases on the 0.13.x branch, and on adding<br>new features on the master branch.<br><br>This release requires Python 2.6, 2.7 or 3.1-3.3 and NumPy 1.5.1 or greater.<br>Highlights of this release are:<br>
<br> - support for fancy indexing and boolean comparisons with sparse matrices<br> - interpolative decompositions and matrix functions in the linalg module<br> - two new trust-region solvers for unconstrained minimization<br>
<br><br>New features<br>============<br><br>``scipy.integrate`` improvements<br>--------------------------------<br><br>N-dimensional numerical integration<br>^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^<br><br>A new function `scipy.integrate.nquad`, which provides N-dimensional<br>
integration functionality with a more flexible interface than ``dblquad`` and<br>``tplquad``, has been added.<br><br>``dopri*`` improvements<br>^^^^^^^^^^^^^^^^^^^^^^^<br><br>The intermediate results from the ``dopri`` family of ODE solvers can now be<br>
accessed by a *solout* callback function.<br><br><br>``scipy.linalg`` improvements<br>-----------------------------<br><br>Interpolative decompositions<br>^^^^^^^^^^^^^^^^^^^^^^^^^^^^<br><br>Scipy now includes a new module `scipy.linalg.interpolative`<br>
containing routines for computing interpolative matrix decompositions<br>(ID). This feature is based on the ID software package by<br>P.G. Martinsson, V. Rokhlin, Y. Shkolnisky, and M. Tygert, previously<br>adapted for Python in the PymatrixId package by K.L. Ho.<br>
<br>Polar decomposition<br>^^^^^^^^^^^^^^^^^^^<br><br>A new function `scipy.linalg.polar`, to compute the polar decomposition <br>of a matrix, was added.<br><br>BLAS level 3 functions<br>^^^^^^^^^^^^^^^^^^^^^^<br><br>The BLAS functions ``symm``, ``syrk``, ``syr2k``, ``hemm``, ``herk`` and<br>
``her2k`` are now wrapped in `scipy.linalg`.<br><br>Matrix functions<br>^^^^^^^^^^^^^^^^<br><br>Several matrix function algorithms have been implemented or updated following<br>detailed descriptions in recent papers of Nick Higham and his co-authors.<br>
These include the matrix square root (``sqrtm``), the matrix logarithm<br>(``logm``), the matrix exponential (``expm``) and its Frechet derivative<br>(``expm_frechet``), and fractional matrix powers (``fractional_matrix_power``).<br>
<br><br>``scipy.optimize`` improvements<br>-------------------------------<br><br>Trust-region unconstrained minimization algorithms<br>^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^<br><br>The ``minimize`` function gained two trust-region solvers for unconstrained<br>
minimization: ``dogleg`` and ``trust-ncg``.<br><br><br>``scipy.sparse`` improvements<br>-----------------------------<br><br>Boolean comparisons and sparse matrices<br>^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^<br><br>All sparse matrix types now support boolean data, and boolean operations. Two<br>
sparse matrices `A` and `B` can be compared in all the expected ways `A < B`,<br>`A >= B`, `A != B`, producing similar results as dense Numpy arrays.<br>Comparisons with dense matrices and scalars are also supported. <br>
<br>CSR and CSC fancy indexing<br>^^^^^^^^^^^^^^^^^^^^^^^^^^<br><br>Compressed sparse row and column sparse matrix types now support fancy indexing<br>with boolean matrices, slices, and lists. So where A is a (CSC or CSR) sparse<br>
matrix, you can do things like::<br><br> >>> A[A > 0.5] = 1 # since Boolean sparse matrices work<br> >>> A[:2, :3] = 2<br> >>> A[[1,2], 2] = 3<br><br><br>``scipy.sparse.linalg`` improvements<br>
------------------------------------<br><br>The new function ``onenormest`` provides a lower bound of the 1-norm of a<br>linear operator and has been implemented according to Higham and Tisseur<br>(2000). This function is not only useful for sparse matrices, but can also be<br>
used to estimate the norm of products or powers of dense matrices without<br>explictly building the intermediate matrix.<br><br>The multiplicative action of the matrix exponential of a linear operator<br>(``expm_multiply``) has been implemented following the description in Al-Mohy<br>
and Higham (2011).<br><br>Abstract linear operators (`scipy.sparse.linalg.LinearOperator`) can now be<br>multiplied, added to each other, and exponentiated, producing new linear<br>operators. This enables easier construction of composite linear operations.<br>
<br><br>``scipy.spatial`` improvements<br>------------------------------<br><br>The vertices of a `ConvexHull` can now be accessed via the `vertices` attribute,<br>which gives proper orientation in 2-D.<br><br><br>``scipy.signal`` improvements<br>
-----------------------------<br><br>The cosine window function `scipy.signal.cosine` was added.<br><br><br>``scipy.special`` improvements<br>------------------------------<br><br>New functions `scipy.special.xlogy` and `scipy.special.xlog1py` were added.<br>
These functions can simplify and speed up code that has to calculate <br>``x * log(y)`` and give 0 when ``x == 0``.<br><br><br>``<a href="http://scipy.io">scipy.io</a>`` improvements<br>-------------------------<br><br>Unformatted Fortran file reader<br>
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^<br><br>The new class `scipy.io.FortranFile` facilitates reading unformatted<br>sequential files written by Fortran code.<br><br>``scipy.io.wavfile`` enhancements<br>^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^<br>
<br>`scipy.io.wavfile.write` now accepts a file buffer. Previously it only<br>accepted a filename.<br><br>`scipy.io.wavfile.read` and `scipy.io.wavfile.write` can now handle floating<br>point WAV files.<br><br><br>``scipy.interpolate`` improvements<br>
----------------------------------<br><br>B-spline derivatives and antiderivatives<br>^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^<br><br>`scipy.interpolate.splder` and `scipy.interpolate.splantider` functions <br>for computing B-splines that represent derivatives and antiderivatives<br>
of B-splines were added. These functions are also available in the <br>class-based FITPACK interface as ``UnivariateSpline.derivative`` and<br>``UnivariateSpline.antiderivative``.<br><br><br>``scipy.stats`` improvements<br>
----------------------------<br><br>Distributions now allow using keyword parameters in addition to<br>positional parameters in all methods.<br><br>The function `scipy.stats.power_divergence` has been added for the<br>Cressie-Read power divergence statistic and goodness of fit test.<br>
Included in this family of statistics is the "G-test"<br>(<a href="http://en.wikipedia.org/wiki/G-test">http://en.wikipedia.org/wiki/G-test</a>).<br><br>`scipy.stats.mood` now accepts multidimensional input.<br>
<br>An option was added to `scipy.stats.wilcoxon` for continuity correction.<br><br>`scipy.stats.chisquare` now has an `axis` argument.<br><br>`scipy.stats.mstats.chisquare` now has `axis` and `ddof` arguments.<br><br><br>
Deprecated features<br>===================<br><br>``expm2`` and ``expm3``<br>-----------------------<br><br>The matrix exponential functions `scipy.linalg.expm2` and `scipy.linalg.expm3`<br>are deprecated. All users should use the numerically more robust<br>
`scipy.linalg.expm` function instead.<br><br>``scipy.stats`` functions<br>-------------------------<br><br>`scipy.stats.oneway` is deprecated; `scipy.stats.f_oneway` should be used<br>instead.<br><br>`scipy.stats.glm` is deprecated. `scipy.stats.ttest_ind` is an equivalent<br>
function; more full-featured general (and generalized) linear model<br>implementations can be found in statsmodels.<br><br>`scipy.stats.cmedian` is deprecated; ``numpy.median`` should be used instead.<br><br><br>Backwards incompatible changes<br>
==============================<br><br>LIL matrix assignment<br>---------------------<br>Assigning values to LIL matrices with two index arrays now works similarly as<br>assigning into ndarrays::<br><br> >>> x = lil_matrix((3, 3))<br>
>>> x[[0,1,2],[0,1,2]]=[0,1,2]<br> >>> x.todense()<br> matrix([[ 0., 0., 0.],<br> [ 0., 1., 0.],<br> [ 0., 0., 2.]])<br><br>rather than giving the result::<br><br> >>> x.todense()<br>
matrix([[ 0., 1., 2.],<br> [ 0., 1., 2.],<br> [ 0., 1., 2.]])<br><br>Users relying on the previous behavior will need to revisit their code.<br>The previous behavior is obtained by ``x[numpy.ix_([0,1,2],[0,1,2])] = ...`.<br>
<br><br>Deprecated ``radon`` function removed<br>-------------------------------------<br><br>The ``misc.radon`` function, which was deprecated in scipy 0.11.0, has been<br>removed. Users can find a more full-featured ``radon`` function in<br>
scikit-image.<br><br><br>Removed deprecated keywords ``xa`` and ``xb`` from ``stats.distributions``<br>--------------------------------------------------------------------------<br><br>The keywords ``xa`` and ``xb``, which were deprecated since 0.11.0, have<br>
been removed from the distributions in ``scipy.stats``.<br><br>Changes to MATLAB file readers / writers<br>----------------------------------------<br><br>The major change is that 1D arrays in numpy now become row vectors (shape 1, N)<br>
when saved to a MATLAB 5 format file. Previously 1D arrays saved as column<br>vectors (N, 1). This is to harmonize the behavior of writing MATLAB 4 and 5<br>formats, and adapt to the defaults of numpy and MATLAB - for example<br>
``np.atleast_2d`` returns 1D arrays as row vectors.<br><br>Trying to save arrays of greater than 2 dimensions in MATLAB 4 format now raises<br>an error instead of silently reshaping the array as 2D.<br><br>``scipy.io.loadmat('afile')`` used to look for `afile` on the Python system path<br>
(``sys.path``); now ``loadmat`` only looks in the current directory for a<br>relative path filename.<br><br><br>Other changes<br>=============<br><br>Security fix: ``scipy.weave`` previously used temporary directories in an<br>
insecure manner under certain circumstances.<br><br>Cython is now required to build *unreleased* versions of scipy.<br>The C files generated from Cython sources are not included in the git repo<br>anymore. They are however still shipped in source releases.<br>
<br>The code base received a fairly large PEP8 cleanup. A ``tox pep8`` <br>command has been added; new code should pass this test command.<br><br>Scipy cannot be compiled with gfortran 4.1 anymore (at least on RH5), likely<br>
due to that compiler version not supporting entry constructs well.<br><br><br>Authors<br>=======<br><br>This release contains work by the following people (contributed at least<br>one patch to this release, names in alphabetical order):<br>
<br>* Jorge Cañardo Alastuey +<br>* Tom Aldcroft +<br>* Max Bolingbroke +<br>* Joseph Jon Booker +<br>* François Boulogne<br>* Matthew Brett<br>* Christian Brodbeck +<br>* Per Brodtkorb +<br>* Christian Brueffer +<br>* Lars Buitinck<br>
* Evgeni Burovski +<br>* Tim Cera<br>* Lawrence Chan +<br>* David Cournapeau<br>* Dražen Lučanin +<br>* Alexander J. Dunlap +<br>* endolith<br>* André Gaul +<br>* Christoph Gohlke<br>* Ralf Gommers<br>* Alex Griffing +<br>
* Blake Griffith +<br>* Charles Harris<br>* Bob Helmbold +<br>* Andreas Hilboll<br>* Kat Huang +<br>* Oleksandr (Sasha) Huziy +<br>* Gert-Ludwig Ingold +<br>* Thouis (Ray) Jones<br>* Juan Luis Cano Rodríguez +<br>* Robert Kern<br>
* Andreas Kloeckner +<br>* Sytse Knypstra +<br>* Gustav Larsson +<br>* Denis Laxalde<br>* Christopher Lee<br>* Tim Leslie<br>* Wendy Liu +<br>* Clemens Novak +<br>* Takuya Oshima +<br>* Josef Perktold<br>* Illia Polosukhin +<br>
* Przemek Porebski +<br>* Steve Richardson +<br>* Branden Rolston +<br>* Skipper Seabold<br>* Fazlul Shahriar<br>* Leo Singer +<br>* Rohit Sivaprasad +<br>* Daniel B. Smith +<br>* Julian Taylor<br>* Louis Thibault +<br>* Tomas Tomecek +<br>
* John Travers<br>* Richard Tsai +<br>* Jacob Vanderplas<br>* Patrick Varilly<br>* Pauli Virtanen<br>* Stefan van der Walt<br>* Warren Weckesser<br>* Pedro Werneck +<br>* Nils Werner +<br>* Michael Wimmer +<br>* Nathan Woods +<br>
* Tony S. Yu +<br><br>A total of 65 people contributed to this release.<br>People with a "+" by their names contributed a patch for the first time.<br><br></div>