DOC: update 0.14.0 release notes. #3286

Merged
merged 2 commits into from Feb 9, 2014

Projects

None yet

4 participants

@rgommers
Owner
rgommers commented Feb 5, 2014

@pv, @jnothman: could one of you suggest wording for the sparse matrix improvements?

@ev-br, @pv: same question for the BPoly/PPoly stuff

Anything else notable that I missed?

@rgommers rgommers added this to the 0.14.0 milestone Feb 5, 2014
@pv pv commented on the diff Feb 5, 2014
doc/release/0.14.0-notes.rst
New features
============
+``scipy.interpolate`` improvements
pv
pv Feb 5, 2014 Owner

Note: the new features list is probably easier to read if there is only one level of sections below it.
I.e., only ------- titles and not ^^^^^^. I.e., I'd do away with the "scipy.XXX improvements" titles.

rgommers
rgommers Feb 5, 2014 Owner

Agreed, I don't like the current formatting much. Will change it for the whole file.

@pv pv commented on an outdated diff Feb 5, 2014
doc/release/0.14.0-notes.rst
New features
============
+``scipy.interpolate`` improvements
+----------------------------------
+
+`scipy.interpolate.interp1d` now accepts non-monotonic inputs and sorts them.
+If performance is critical, sorting can be turned off by using the new
+``assume_sorted`` keyword.
+
+TODO: describe `scipy.interpolate.BPoly` and `scipy.interpolate.PPoly` work.
pv
pv Feb 5, 2014 Owner

"""Faster implementations of piecewise polynomials in power and Bernstein polynomial bases have been added as scipy.interpolate.PPoly and scipy.interpolate.BPoly. New users should use these in favor of scipy.interpolate.PiecewisePolynomial."""

Coverage Status

Changes Unknown when pulling c1f31ab on rgommers:014-relnotes into * on scipy:master*.

@pv pv commented on an outdated diff Feb 5, 2014
doc/release/0.14.0-notes.rst
@@ -57,6 +93,19 @@ Box-Cox transformation
The functions `scipy.special.boxcox` and `scipy.special.boxcox1p` compute
the Box-Cox transformation.
+
+``scipy.sparse`` improvements
+-----------------------------
+
+TODO: finish describing changes / new features in sparse matrix formats.
+
+- sparse matrices can now be indexed with 64-bit integers, and are therefore no
+ longer limited to ``2^31`` nonzero elements.
+- ``__numpy_ufunc__`` in combination with numpy 1.9
+- performance optimizations
pv
pv Feb 5, 2014 Owner

"""
Significant performance improvement in CSR, CSC, and DOK indexing speed.
"""
NB. the PR with the DOK changes is still waiting for review...

@pv pv commented on an outdated diff Feb 5, 2014
doc/release/0.14.0-notes.rst
@@ -57,6 +93,19 @@ Box-Cox transformation
The functions `scipy.special.boxcox` and `scipy.special.boxcox1p` compute
the Box-Cox transformation.
+
+``scipy.sparse`` improvements
+-----------------------------
+
+TODO: finish describing changes / new features in sparse matrix formats.
+
+- sparse matrices can now be indexed with 64-bit integers, and are therefore no
pv
pv Feb 5, 2014 Owner

"""
Sparse matrices are no longer limited to 2^31 nonzero elements. They automatically switch to using 64-bit index data type for matrices containing more elements. User code written assuming the sparse matrices use int32 as the index data type will continue to work, except for such large matrices. Code dealing with larger matrices needs to accept either int32 or int64 indices.
"""

@pv pv commented on an outdated diff Feb 5, 2014
doc/release/0.14.0-notes.rst
@@ -57,6 +93,19 @@ Box-Cox transformation
The functions `scipy.special.boxcox` and `scipy.special.boxcox1p` compute
the Box-Cox transformation.
+
+``scipy.sparse`` improvements
+-----------------------------
+
+TODO: finish describing changes / new features in sparse matrix formats.
+
+- sparse matrices can now be indexed with 64-bit integers, and are therefore no
+ longer limited to ``2^31`` nonzero elements.
+- ``__numpy_ufunc__`` in combination with numpy 1.9
pv
pv Feb 5, 2014 Owner

"""
When using Numpy >= 1.9 (to be released in MM 2014), sparse matrices function correctly when given to arguments of np.dot, and np.multiply, and other ufuncs. With earlier Numpy and Scipy versions, the results of such operations are undefined and usually unexpected.
"""

Owner
pv commented Feb 5, 2014

More stuff:

  • Condition number estimate for matrix exponential, scipy.linalg.expm_cond.
  • Evaluation of bivariate spline derivatives in scipy.interpolate
  • More controllable error estimation in scipy.optimize.curve_fit

Backward incompatible changes:

  • Eigenvectors in the case of generalized eigenvalue problem are normalized to unit vectors in 2-norm, rather than following the LAPACK normalization convention.
Owner
rgommers commented Feb 8, 2014

Thanks @pv, all comments addressed.

Coverage Status

Coverage remained the same when pulling 8a536bd on rgommers:014-relnotes into 63f951e on scipy:master.

@pv pv merged commit 47a89d5 into scipy:master Feb 9, 2014

1 check passed

default The Travis CI build passed
Details
Owner
pv commented Feb 9, 2014

Merging.

@jnothman jnothman commented on the diff Feb 10, 2014
doc/release/0.14.0-notes.rst
``scipy.special`` improvements
------------------------------
-Box-Cox transformation
-^^^^^^^^^^^^^^^^^^^^^^
+The functions `scipy.special.boxcox` and `scipy.special.boxcox1p`, which
+compute the Box-Cox transformation, have been added.
+
+
+``scipy.sparse`` improvements
+-----------------------------
+
+- Significant performance improvement in CSR, CSC, and DOK indexing speed.
+- When using Numpy >= 1.9 (to be released in MM 2014), sparse matrices function
+ correctly when given to arguments of ``np.dot``, ``np.multiply`` and other
jnothman
jnothman Feb 10, 2014 Contributor

We now know that ufuncs do not work for sparse matrices with duplicate indices at present... I'm not sure if this strong a statement is appropriate unless that is fixed before the release.

pv
pv Feb 10, 2014 Owner

On one hand, it's not a regression.

@BrianNewsom BrianNewsom added a commit to BrianNewsom/scipy that referenced this pull request Feb 11, 2014
@rgommers @BrianNewsom rgommers + BrianNewsom DOC: complete 0.14.0 release notes. Addresses review comments on gh-3286
.

Thanks to @pv.
de70f2f
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment