Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP

Comparing changes

Choose two branches to see what's changed or to start a new pull request. If you need to, you can also compare across forks.

Open a pull request

Create a new pull request by comparing changes across two branches. If you need to, you can also compare across forks.
base fork: scipy/scipy
...
head fork: scipy/scipy
Checking mergeability… Don't worry, you can still create the pull request.
This comparison is big! We're only showing the most recent 250 commits
Commits on Apr 17, 2015
@insertinterestingnamehere insertinterestingnamehere MAINT: Require Cython 0.22 in the installation docs and the cythonize
script.
Also use Cython 0.22 in the benchmark suite.
9f4903e
@insertinterestingnamehere insertinterestingnamehere BUG: Fixed compilation failures with recent versions of Cython on win…
…dows.
1eb34a2
@argriffing argriffing Merge pull request #4699 from wuhaochen/master
DOC: cluster.hierarchy.linkage and sparse.linalg.spilu
6cdd087
@argriffing argriffing Merge pull request #4713 from ev-br/no_dtype
MAINT: interpolate: no need to pass dtype around if it can be inferred
989c75f
Commits on Apr 20, 2015
@alexbrc alexbrc DOC: optimize.leastsq default maxfev clarification 4702e2b
@alexbrc alexbrc MAINT: better generic initialization of location and scale parameters…
… for mle
4d60815
@argriffing argriffing Merge pull request #4718 from ev-br/fshapes
ENH: stats: can specify f-shapes to fix in fitting by name
4a41fff
@alexbrc alexbrc MAINT: count bfgs updates more carefully 4f4e14d
Commits on Apr 21, 2015
@larsmans larsmans Merge pull request #4751 from argriffing/bench-sparse-null-slice
BENCH: add sparse null slice benchmark
8c2ba28
@dlax dlax Merge pull request #4758 from argriffing/leastsq-docstring
DOC: optimize.leastsq default maxfev clarification
acd131d
Commits on Apr 22, 2015
@insertinterestingnamehere insertinterestingnamehere MAINT: Specify Cython version in Travis build. 2ccca3a
@rgommers rgommers Merge pull request #4753 from insertinterestingnamehere/compile_fix
BUG: Fixed compilation with recent Cython versions.
14a6a5b
@alexbrc alexbrc DOC: linprog copyediting 90986de
Commits on Apr 23, 2015
@alexbrc alexbrc MAINT: fix superficial linprog bugs 361ddfd
@dlax dlax Merge pull request #4763 from argriffing/linprog-cleanup
fix small linprog bugs
f9810b9
@Andreea-G Andreea-G BUG: Fixes #4594: scipy.optimize.linprog IndexError when a callback i…
…s provided

The bug appeared at the beginning of Phase 2, if Phase 1 successfully terminated with some of the artificial variables as pivots. These artificial variables were zero because the corresponding element in the last column of T was zero, which allowed all artificial variables to be eliminated and Phase 1 to terminate successfully. However, basis[:m] contained elements that were greater than the length of solution, which resulted in IndexError at solution[basis[:m]]. The remaining true (non-artificial) variables that are not included in basis[:m] are also zero. Therefore simply setting the length of solution to be large enough solves the problem.
9438fb7
@Andreea-G Andreea-G BUG: Fixes #4746: linprog returns solution violating equality constraint
This adds to pull request #4685. The problem happens if some artificial variable(s) remain basic at the end of Phase 1. PR #4685 solved the problem if the row in T associated with the artificial variable was zero (for all columns in T that correspond to a non-artificial variable). When there are non-zero elements in that row, then one of the non-zero elements is picked for pivot, and the artificial variable is replaced in the basis by the non-artificial variable. This is done until there are no more artificial variables in the basis that have non-zero rows in T.
260f6a8
Commits on Apr 24, 2015
@alexbrc alexbrc BENCH: add signal.lsim benchmark 7992499
@alexbrc alexbrc BENCH: add signal.lsim2 benchmark 8527360
@alexbrc alexbrc BENCH: reduce time for signal.lsim benchmarks 8b0af55
@argriffing argriffing Merge pull request #4762 from Andreea-G/linprog_artificial
BUGS: Fixes #4746 and #4594: linprog returns solution violating equality constraint and IndexError when a callback is provided
bca1d9c
@cwrowley cwrowley Deprecate nonuniform timesteps in lsim 822d7e6
Commits on Apr 25, 2015
@Andreea-G Andreea-G Fixes #4726: test_cobyla.test_vector_constraints
Set rhoend and tol as 1e-5 in test_cobyla for fmin_cobyla and minimize.
5b802a1
@dlax dlax Merge pull request #4769 from Andreea-G/cobyla_rhoend
Fixes #4726: test_cobyla.test_vector_constraints
67dce17
Commits on Apr 26, 2015
@astrofrog astrofrog BUG: io.idl: don’t crash if a pointer reference is not found in the h…
…eap.

Instead, set variable to None and emit a warning.  Closes gh-4613.
e83e981
@astrofrog astrofrog TST: added regression test for io.readsav issue with invalid pointers. d0645f7
@rgommers rgommers Merge pull request #4710 from astrofrog/fix-idl-pointer
scipy.io.idl: make reader more robust to missing variables in heap
100ce74
@rgommers rgommers Merge pull request #4721 from f0k/patch-1
DOC: document that imresize converts the input to a PIL image
e4b5787
@rgommers rgommers Merge pull request #4711 from ortylp/master
Fix crash for unknown chunks at the end of file
49f0cb3
@daggerdagr daggerdagr edited scipy/stats/stats.py to fix doctest for fisher_exact 5edc08c
@rgommers rgommers Merge pull request #4771 from daggerdagr/master
DOC: fix documentation error for fisher_exact
bcf5fcd
Commits on Apr 27, 2015
@Andreea-G Andreea-G BUG: Fixes #4733: optimize.brute finish option is not compatible with…
… optimize.minimize

Made optimize.brute compatible with optimize.minimize and other minimization functions that may now be passed as argument to `finish`. `finish` is now allowed to also return an `OptimizeResult` object.
d461e86
@alexbrc alexbrc TST: remove chi from list of failing fits 87d128c
@ev-br ev-br Merge pull request #4759 from argriffing/stats-fit-support
improved stats mle fit
fce7061
@sturlamolden sturlamolden ENH: rewrite/enhancement of scipy.spatial.cKDTree.
This PR does the following:
   1.  Store the entire kd-tree in a contiguous memory buffer (std::vector).
   2.  Remove remaining memory leaks in cKDTree.query by using a memory pool.
  3a.  Move all code and objects used by cKDTree.query to C++
  3b.  cKDTree.query now releases the GIL
  3c.  cKDtree.query is 50-100 % faster with one thread.
  4a.  Add support for pickling cKDTree on Python 3 (not working on Pyhton 2).
  4b.  Using cKDTree with multiprocessing is easier when pickle is supported
   5.  Multithreading cKDTree.query with n_jobs keyword (excellent scalability).
   6.  Increase default leafsize to 16.
   7.  Add option for constructing balanced kd-trees (used by default).
   8.  Add option for compacting nodes by recomputing boundaries (used by default).
   9.  The kd-tree is inspectable from Python through a new cKDTreeNode class.
  10.  Tests added for all new functionality.
The remaining query functions in cKDTree can be migrated to C++ for GIL release
in a similar manner. Subset queries proposed for scipy.spatial.KDTree can
e.g. be implemented in cKDTree using std::set.
752351c
@sturlamolden sturlamolden MAINT: address rgommers comments, C style updates ff493e7
@sturlamolden sturlamolden MAINT: remove NumPy deprecation define 0aa9886
@sturlamolden sturlamolden ENH: prefetch data during bruteforce search. Speedup ~20% 9916d51
@sturlamolden sturlamolden ENH: SIMD accelerate Euclidian distances, speeds up cKDTree b5b1033
@rgommers rgommers BLD: spatial.ckdtree: fix Bento build. Also update .gitignore eda16c3
@ev-br ev-br DOC: document how `svdvals` handles empty input 11b9ee2
@ev-br ev-br DOC: linalg: add funm_psd as a docstring example
For a hermitian matrix, using eigendecomosition is likely
faster and simpler than a full Schur-decomp based algorithm.

This turns the code by @argriffing from gh-3556 into an example.
e9bb211
@rgommers rgommers MAINT: ckdtree: fix some signatures, and PEP8 / doc formatting fixes. fc6cc8f
@rgommers rgommers MAINT: ckdtree: fix some build warnings 09f4785
@ev-br ev-br Merge pull request #4365 from argriffing/empty-svd
svdvals now returns a length-0 sequence of singular values given a size-0 2d array
20c9a1a
@dlax dlax Merge pull request #4756 from Andreea-G/brute
BUG: Fixes #4733: optimize.brute finish option is not compatible with optimize.minimize
1d61b06
@dlax dlax Merge pull request #4760 from argriffing/lbfgs-basinhopping-experiment
MAINT: count bfgs updates more carefully
e9eda32
@rgommers rgommers MAINT: basinhopping: fix error in tests and docstring. 1e7c7ce
Commits on Apr 28, 2015
@pbrod pbrod ENH: Refactored DCT and DST in order to accept int and complex types …
…like fft does.

Also refactored the code to allow padding/truncation:
If ``n < x.shape[axis]``, `x` is truncated.
If ``n > x.shape[axis]``, `x` is zero-padded.
This is similar to the fft function.
2af6bde
@ev-br ev-br Merge pull request #4381 from pbrod/ENH_make_DCT_accept_int_and_complex
ENH: Make DCT and DST accept int and complex types like fft
5fc2232
@dlax dlax Merge pull request #4717 from rgommers/basinhopping-issue
MAINT: basinhopping: fix error in tests, silence /0 warning, fix docstring
506d877
@rgommers rgommers DOC: update 0.16.0 release notes.
[ci skip]
3d6e084
@ev-br ev-br ENH: interpolate: add `axis` attribute to PPoly and BPoly 30448ac
@ev-br ev-br MAINT: interpolate: refactor PchipInterpolator to be a *Poly subclass c7ec61b
@ev-br ev-br MAINT: interpolate: slightly improve test_polyint
* slightly improve readability of a test
* clean up imports
611eb6b
@ev-br ev-br ENH: interpolate: add `axis` argument to Akima interpolator b0395b4
@ev-br ev-br BENCH: add a simple benchmark for PPoly 12944a1
@ev-br ev-br BUG: fix off-by-one error in axis handling in PPoly
require that 0 <= axis < c.ndim-1
c55032e
Commits on Apr 29, 2015
Anne Archibald Include apparently-forgotten functions in docs 61de9d0
@andreas-h andreas-h ENH: only rearrange output if necessary in _PPolyBase.__call__ faf68b9
@andreas-h andreas-h STY: PEP8 compliance for _monotone.py 8a6b6de
@ev-br ev-br Merge pull request #4780 from aarchiba/circdocfix
Include apparently-forgotten functions in docs
3af64e7
@ev-br ev-br DOC: reformat the Methods section of the KDE docstring
so that method docstrings render
ff07364
@alexbrc alexbrc MAINT: cdf and ppf for rice distribution 239a731
@alexbrc alexbrc TST: remove the rice distribution from the distmissing list 6abad09
@rgommers rgommers Merge pull request #4786 from ev-br/kde_docs
DOC: reformat the Methods section of the KDE docstring
a18db90
@jjgreen jjgreen Default epsilon fix for RBF interpolate
MAINT: rbf: add a regression test from original bug report, #4523

Make the test deterministic by using the random seed.
Also avoid runtime warnings by using xlogy in the thin-plate kernel.

TST: rbf: stop silencing RuntimeWarnings

removed (my) original stability test

ev-br's version is better and has the same coverage,
so make mine obsolete

removed some blank lines
ce66e32
Commits on Apr 30, 2015
@ev-br ev-br Merge pull request #4643 from jjgreen/RBF_default_epsilon_too_large_2020
RBF default epsilon too large 2020
b6a114c
@endolith endolith MAINT: Use a dictionary for function name synonyms
Gets rid of duplication of information and long elif chain, minimizing
future problems like
e-q@55c2ca8

(also some minor PEP8 fixes)
69fbb6f
@rgommers rgommers Merge pull request #4778 from endolith/window_synonyms
MAINT: signal.windows: use a dictionary for function name synonyms
89c7022
@ev-br ev-br CI: add a kludge for detecting test failures which try to disguise as…
… success

Save the output from runtests.py and look for the magic line nose prints
for a successful test run. See, e.g., #4736
for the motivation and idea.
1a1d9c7
Commits on May 01, 2015
@ev-br ev-br DOC: stats: fix py errors in docstring examples 02b45fb
@ev-br ev-br DOC: interpolate: fix py errors in docstring examples 7efaf7b
@ev-br ev-br DOC: linalg: fix py errors in docstring examples aaacbbb
@ev-br ev-br DOC: linalg: fix a typo in solve_circulant docstring a22af5b
@ev-br ev-br DOC: fix a typo in qr_insert docstring example f79b6b3
@ev-br ev-br DOC: special: fix some py errors in docstring examples 788d133
@ev-br ev-br DOC: fix some py errors in docstring examples
in spatial, integrate, signal
cffc9b4
@ev-br ev-br DOC: optimize: fix some py errors in docstring examples 9a135e9
@ev-br ev-br DOC: sparse: fix some py errors in docstring examples 0177788
@ev-br ev-br DOC: io.netcdf: fix a typo in a docstring example 36ff4fc
@ev-br ev-br DOC: fftpack: tweak syntax in docstring examples 0c43fa7
@ev-br ev-br MAINT: add a docstring-check option to refguide_check 9db1077
@ev-br ev-br CI: use python -u for unbuffered stdin/stderr on TravisCI 8c7a163
@ev-br ev-br CI: address review comments
use sys.exit in scripts
0f87b16
@ev-br ev-br CI: move validate_runtests_log.py to tools/ 6705975
@pv pv Merge pull request #4792 from ev-br/xerbla_try_harder
CI: add a kludge for detecting test failures which try to disguise as successes
ea41482
@perimosocordiae perimosocordiae TST: adding test case for #4791 406e88a
@perimosocordiae perimosocordiae BUG: numpoints wasn't incremented for new points
Fixes #4791
6115c09
@WarrenWeckesser WarrenWeckesser BUG: spatial: Handle 1-D data in pdist and cdist with the 'mahalanobi…
…s' metric.

There's no reason that pdist and cdist shouldn't handle observations that are
1-D when the Mahalanobis metric is used. Before this change, a call such as
    pdist([[1], [2], [3], [4]], metric='mahalanobis')
would crash with a LinAlgError exception, because np.linalg.inv(V) doesn't
handle a scalar argument.  The fix is to ensure that inv(V) is called with
a 2-D array, even when the data points are 1-D.
7369d65
@WarrenWeckesser WarrenWeckesser BUG: spatial: mahalanobis edge case in pdist and cdist
In pdist and cdist with the Mahalanobis metric, when the covariance matrix or
its inverse are not provided, it is computed from the given observations.
For the covariance matrix to be nonsingular, the number of observations
must be more than the dimension of the observations.  Before this change, this
conditions was not checked.  The function might have raised a LinAlgError if
the inv() function detected the noninvertibility.  If the noninvertibility was
not detected, the inverse computed by np.linalg.inv()--and therefore the values
returned by the function--would be garbage.

With this change, a ValueError is raised when there are too few observations
given to pdist or cdist.
84ef56d
Commits on May 02, 2015
@erosennin erosennin BUG: Fixes #4673: Mark FITPACK functions as thread safe. 402a1ec
@erosennin erosennin TST: Move knownfailure_overridable decorator to scipy._lib._testutils. e53bfac
@erosennin erosennin TST: Add a test for GIL release in scipy.interpolate.RectBivariateSpl…
…ine.
c59d272
@danielsjensen1 danielsjensen1 BUG: Fix TypeError in scipy.optimize._trust-region.py when disp=True.
See ticket #4798.
8e1b695
@ev-br ev-br Merge pull request #4770 from erosennin/fitpack-threadsafe
Mark FITPACK functions as thread safe.
0a4e34c
Commits on May 03, 2015
Anne Archibald DOC: add brief note about origin of Lena image
Also mention alternatives 'face' and 'ascent' and add those
to the function listing in the scipy.misc docstring.
ec85393
@rgommers rgommers Merge pull request #4785 from aarchiba/lena_note
DOC: add a brief note about origin of Lena image
d9b221f
@ev-br ev-br DOC: tweak the code sample in `KrylovJacobian` docstring
[ci skip]
6977e97
@rgommers rgommers Merge pull request #4768 from ev-br/docstr_ex
DOC: fix python syntax errors in docstring examples
1306448
Commits on May 04, 2015
@ewmoore ewmoore BUG: Relaxed stride issues in qr_update f32f960
@argriffing argriffing Merge pull request #4801 from danielsjensen1/trust-region-fixes
BUG: Fix TypeError in scipy.optimize._trust-region.py when disp=True.
3249faf
@argriffing argriffing Merge pull request #4803 from ewmoore/qrup_fixes
Issues with relaxed strides in QR updating routines
d4fbcce
@alexbrc alexbrc TST: increase scipy.optimize test coverage for disp=True 7db6786
@alexbrc alexbrc PEP8: add blank line 0cd978a
Commits on May 05, 2015
@alexbrc alexbrc MAINT: use an informed initial guess for cauchy fit 8e2077a
@aeklant aeklant ENH: `scipy.stats` now returns namedtuples.
See issue #4192
cd31033
@aeklant aeklant MAINT: moved namedtuple definition to within the function that uses it. b787803
@aeklant aeklant TST: added a generic function for namedtuples attribute verification. 3e5ea71
@aeklant aeklant STY: removed extra blank lines (PEP8) 3e6bfdd
@aeklant aeklant TST: Enhanced generic test for namedtuples to handle masked arrays. 3d060c2
@aeklant aeklant ENH: namedtuples change made to `stats.py` made.
See #4192
e80361f
@aeklant aeklant ENH: namedtuples for `mstats`, `morestats` and `_binned_statistic`. c8ea1f5
@rgommers rgommers Merge pull request #4773 from rgommers/0160-notes
DOC: update 0.16.0 release notes.
a1c0772
@rgommers rgommers Merge pull request #4784 from andreas-h/ppoly_axis
ENH: add an axis attribute to PPoly and friends
6c23655
@rgommers rgommers Merge pull request #4775 from ev-br/funm_psd_doc
DOC: linalg: add funm_psd as a docstring example
511d92d
@rgommers rgommers Merge pull request #4675 from cwrowley/fix-lsim
MAINT: fix signal.lsim
2af6024
@rgommers rgommers Merge pull request #4766 from argriffing/lsim-benchmark
BENCH: add signal.lsim benchmark
8644798
@alexbrc alexbrc TST: add another cauchy fit test b6f5af1
@e-q e-q ENH: scipy.sigal - Addition of CSD, coherence; Enhancement of welch
I have adapted the matplotlib.mlab code for use in scipy.signal. Some
reworking of the matplotlib code was needed to make functions work for
multidimensional arrays, as the current scipy.signal.welch function allows.
Things were also adjusted to keep the same welch API.

Advantages of this code change include:
- A speedup of about 8x in welch
- A speedup of about 6x in csd compared to extending the current
  algorithm to cover the cross spectral density.
- Straightforward extensibility to add spectrogram functionality

There are probably still rough edges, but this seems to be working. Some
tests are included. (Known test failure : test_complex_32 in
test_spectral.TestWelch; effectively correct, but the tolerances are too
tight.)
5f77ab9
@e-q e-q Tests, docstrings, private functions, whitespace fe2dfc8
@e-q e-q Integer division fixes, style fixes 643551d
@e-q e-q Clean up private functions, PEP8 style variable names a5bdeaa
@e-q e-q Empty array fixes and tests for axis!=-1 242f449
@e-q e-q Clean up unncessary work in _fft_helper ba73160
@e-q e-q Ease up tolarances on hardcoded data comparison tests 337afd2
@e-q e-q Clean up trailing whitespaces ee74038
@e-q e-q Return correctly broadcast shape for empty input 0ce5017
@e-q e-q Make window index last whether axis kwarg is -1 or not 73a59da
@e-q e-q Fix inadvertant integer division in calculation of window times 5034987
@e-q e-q First pass at benchmarks
Running on master:
[100.00%] ··· Running signal.CalculateWindowedFFT.time_welch 785.73ms

Now:
[100.00%] ··· Running signal.CalculateWindowedFFT.time_welch 66.92ms

[ci skip]
1f8bd40
@e-q e-q Remove singleton dimension in a python2.6 compatible way 2a5dbad
@e-q e-q PEP8 doesn'nt like lambdas in detrend_func defs
[ci skip]
52e1b6f
@e-q e-q Fix conversion to onesided for odd nfft at last frequency point
See matplotlib/matplotlib#4324 for description of the problem. The
previous welch algorithm suffered from the same bug, so the unit tests
had to be updated/corrected as well.
3fdb71b
@e-q e-q DOC: Examples, references 9241f1d
@Newman101 Newman101 PEP8ify codata.py 4bb5465
@rgommers rgommers Merge pull request #4799 from WarrenWeckesser/mahalanobis
BUG: spatial: Fix a couple edge cases for the Mahalanobis metric in pdist and cdist.
3b713a3
@jakevdp jakevdp Merge pull request #3116 from perimosocordiae/patch-1
sparse: enhancements for DIA format
16654c2
@ewmoore ewmoore MAINT: remove repeated divisions from lfilter main loop
This gives ~3x speed up to general lfilter calls.
7de60f5
@alexbrc alexbrc MAINT: simplify the suppressed_stdout decorator 723e95e
@e-q e-q Fix typo in benchmark imports 6289934
Commits on May 06, 2015
@aeklant aeklant STY: fix typos and naming consensus details. 6676ee8
@andreas-h andreas-h MAINT: update contributor name mapping c030ad5
@rgommers rgommers Merge pull request #4814 from andreas-h/gh-4774
MAINT: update contributor name mapping
02f1eb6
@maniteja123 maniteja123 BENCH: Add distributions to stats benchmarks
minor nitpicks cleanup in the stats benchmarks
ea141f8
@maniteja123 maniteja123 BENCH: Benchmarks for stats module
Some style nitpicks and changes to fisher benchmarks
af2f992
@ewmoore ewmoore ENH: add a FIR filter path to lfilter.
Using np.convolve is much faster than using signal.convolve here.
9f8c679
@ewmoore ewmoore Merge pull request #4684 from ewmoore/he_bug
BUG: various errors in weight calculations in orthogonal.py
fc5f348
@aeklant aeklant TST: fixed a change omission. 214445d
Anne Archibald Added many missing special functions to docs ea88824
@rgommers rgommers Merge pull request #4709 from aeklant/issue_#4192
ENH: `scipy.stats` now returns namedtuples.
ccedf73
@ewmoore ewmoore BUG: Relaxed strides cleanup in decomp_update.pyx.in
With relaxed stride checking, the stride of a singleton dimension
cannot be used for any purpose.

Closes gh-4736.
21483e7
@rgommers rgommers Merge pull request #4682 from e-q/csd2
ENH: scipy.signal - Addition of CSD, coherence; Enhancement of welch
7443c29
@befelix befelix ENH: Split signal.lti into subclasses, part of #2912
new classes: ss, tf, zpk
* backwards compatible
* Remove redundant information in lti
* Make internal transformations more transparent
f1ea8f7
@befelix befelix Fix typo in tf class docstring 6c5d068
@befelix befelix Rename conversion functions to to_tf(), to_ss() and to_zpk() 60e2ba0
@befelix befelix Enforce array structure for setter inputs in lti subclasses 9f3725c
@befelix befelix Handle instance conversion in a cleaner way 4b179d7
@befelix befelix Change copy method to be private bc46ad1
@befelix befelix Add lsim2 methods to lti class fdd29d0
@befelix befelix Fix typo in lti.step2 docstring 4bae6d6
@befelix befelix Rename classes 39b5333
@befelix befelix Fix line break a54bd29
@befelix befelix Small change to the way copies are handled 0b699bd
@befelix befelix Remove unecessary **kwargs 9cf0245
@befelix befelix Zeros and Poles should be at least 1d 9db3144
@befelix befelix Fix docstrings 6e53478
@befelix befelix Fix typos in docstrings b7e29f8
@befelix befelix Docstrings, thanks endolith 072bc05
@befelix befelix Add subclasses to documentation f0bc8de
@befelix befelix Large documentation update daa2a34
@befelix befelix To_* methods make copies if system is already in desired form 0e351fc
@befelix befelix Add test for subclasses 97e82ec
@befelix befelix Small typo 637c699
@befelix befelix Make sure simulation functions don't use multiple system conversions 40fcd3a
@ev-br ev-br Merge pull request #4806 from argriffing/cauchy-fit-iqr
MAINT: use an informed initial guess for cauchy fit
2df0168
@ev-br ev-br Merge pull request #4787 from argriffing/rice-cdf
Add rice cdf and ppf.
83f96b0
@rgommers rgommers Merge pull request #4783 from aarchiba/special_docfix
DOC: added many missing special functions to docs
8eff9fd
@ewmoore ewmoore BUG/TST: Fail on scalar inputs in QR update routines fd1d6a8
@befelix befelix Revert "Add lsim2 methods to lti class"
This reverts commit fdd29d0.
d0afc2d
@rgommers rgommers BLD: fix Bento build for sgemv fix and non-Accelerate usage on OS X
[ci skip]
bf067d0
@rgommers rgommers BLD: install cython_blas/lapack.pxd in Bento build.
[ci skip]
5b27d65
@rgommers rgommers Merge pull request #4820 from rgommers/bento-sgemv
BLD: update Bento build for sgemv fix and install cython blas/lapack pxd's
5b341aa
@rgommers rgommers Merge pull request #4628 from ewmoore/lfilter_speed
ENH: speed up signal.lfilter and add a convolution path for FIR filters
abf7644
Commits on May 07, 2015
@befelix befelix Make lti tests less verbose 176f38e
Anne Archibald Make refguide_check smarter about false positives c42574c
Anne Archibald ENH: replace exception list with a bogus call 93a6992
Anne Archibald ENH avoid builtin name 'all' e93e492
Anne Archibald BUG fix deletion from inside for loop 17b6f7b
@ewmoore ewmoore DOC: add csd and coherence to __init__.py
[ci skip]
7d0721a
@ev-br ev-br Merge pull request #4827 from ewmoore/csd_doc
DOC: add csd and coherence to __init__.py
51c8efe
@ev-br ev-br MAINT: restore the --check_docs (non-deprecated objs only) 2d168fa
@ev-br ev-br Merge pull request #4795 from aarchiba/refguide_check
Make refguide_check smarter about false positives
477f1b3
@rgommers rgommers Merge pull request #4828 from ev-br/pr/4795
MAINT: restore the --check_docs option of refguide_check.py
48ec4b9
@rgommers rgommers Merge pull request #4714 from maniteja123/stats-bench
BENCH: Add benchmarks for stats module
4b13b95
@rgommers rgommers REL: update release scripts before 0.16.x branching.
Note that we're not building Python 2.6/3.2 binaries anymore, for the
next release we do build 2.7, 3.3 and 3.4 binaries.

The paths to BLAS/LAPACK/ATLAS are made consistent with where those
libs are put by https://github.com/numpy/numpy-vendor

[ci skip]
f194c00
Commits on May 08, 2015
@e-q e-q ENH: scipy.signal - Addition of spectrogram function
Given #4682, a spectrogram function is effectively free; it's just welch
without the averaging. I've added one test to make sure that the average of
the spectrogram segments agrees with the output of welch. Should more
tests be written, given that the only code difference between welch and
spectrogram is the averaging?
621ba73
@e-q e-q DOC: scipy.signal Remove uneccesary mentions of Hz in spectral.py
Also, adds ``versionadded`` lines, and fixes a typo in the coherence
docstring.

[ci skip]
c3a4832
@rgommers rgommers Merge pull request #4823 from e-q/spectrogram
ENH: scipy.signal - Addition of spectrogram function
dc313cb
Commits on May 09, 2015
@befelix befelix Final docstring tweaks f623a14
@rgommers rgommers BLD: fix issue in linalg ``*lange`` wrappers for g77 builds.
Fixes the first issue identified in gh-4829.
464174c
@rgommers rgommers Merge pull request #4576 from befelix/lti_subclasses
ENH: Split signal.lti into subclasses, part of #2912
f1e7689
@rgommers rgommers TST: fix test failures in scipy.special with mingw32 due to test prec…
…ision.

Closes gh-4840.
7509169
@rgommers rgommers DOC: update site.cfg.example. Mostly taken over from Numpy
[ci skip]
444a95b
@pv pv Merge pull request #4842 from rgommers/update-sitecfg
DOC: update site.cfg.example.  Mostly taken over from Numpy
6ea0796
@pv pv Merge pull request #4706 from ev-br/specfun_ub
MAINT: special: avoid out-of-bounds access in specfun
2c1a428
@pv pv Merge pull request #4797 from perimosocordiae/patch-7
BUG: spatial: numpoints not updated for incremental Voronoi
b304760
@pv pv Merge pull request #4701 from ev-br/bpoly_anti
ENH: interpolate: add Bpoly.antiderivatives / integrals
15434ac
@rgommers rgommers Merge pull request #4810 from Newman101/PEP8Codata
MAINT: PEP8ify codata.py
808013c
@WarrenWeckesser WarrenWeckesser BUG: signal: Make spectrogram's return values order match the docs an…
…d test.

Closes gh-4844
cf1f43f
Commits on May 10, 2015
@sturlamolden sturlamolden BUG: fix prefetch statement for MSVC 0bbaa32
@sturlamolden sturlamolden MAINT: remove GNU vector extensions in distance computations f5fb519
@sturlamolden sturlamolden BUG: initialize array with correct types 0588eff
@sturlamolden sturlamolden MAINT: remove NotImplementedError guard in test_ckdtree_parallel fcdf071
@rgommers rgommers Merge pull request #4845 from WarrenWeckesser/specgram-return
BUG: signal: Make spectrogram's return values order match the docs and test.
4edcae3
@rgommers rgommers Merge pull request #4812 from ewmoore/qr_audit
BUG: Relaxed strides cleanup in decomp_update.pyx.in
df38d00
@maniteja123 maniteja123 DOC:Fix error in ode docstring example
Fix print formatting error in ode docstring example. fixes #4772
7175e14
@rgommers rgommers Merge pull request #4841 from rgommers/special-failures
TST: fix test failures in scipy.special with mingw32 due to test precision
b19ebe1
@rgommers rgommers Merge pull request #4833 from rgommers/fix-lange-wrappers
BLD: fix issue in linalg ``*lange`` wrappers for g77 builds.
c3dcc41
@rgommers rgommers Merge pull request #4451 from juliantaylor/vect-distance2
ENH: vectorize distance loops with gcc
aa2f836
@rgommers rgommers Merge pull request #4374 from sturlamolden/ckdtree-query-fix-memleak
ENH: Enhancements to spatial.cKDTree
bea41ad
@rgommers rgommers Merge pull request #4849 from maniteja123/issue4772
DOC:Fix error in ode docstring example
1a8db27
@rgommers rgommers REL: set version to 0.17.0-dev0, add 0.17.0 release notes. 145e011
@ev-br ev-br CI: set pipefail to propagate the test error code 4b5c894
@ev-br ev-br CI: make validate_runtests_log executable af741e2
Commits on May 11, 2015
@WarrenWeckesser WarrenWeckesser MAINT: signal: Add matplotlib import to the lombscargle example. a4d185c
@rgommers rgommers Merge pull request #4851 from WarrenWeckesser/lombscargle-plot
MAINT: signal: Add matplotlib import to the lombscargle example.
dabfb45
@ev-br ev-br CI: make validate_runtests_log detect test failures 18d079c
@ev-br ev-br Merge pull request #4852 from ev-br/ci_build_log
CI build log: try harder to detect test failures
e6407b6
@sturlamolden sturlamolden BUG: fix typo causing memleak
tree_buffer misspelled as tree_vector in destructor
71832db
Commits on May 12, 2015
@rgommers rgommers Merge pull request #4857 from sturlamolden/fix
BUG: fix typo causing memleak
fb23da7
Commits on May 13, 2015
@FRidh FRidh DOC: Add example to scipy.signal.hilbert
Add example showing how to determine the amplitude envelope and
instantaneous frequency of a signal.
279aa59
Commits on May 14, 2015
@viknat viknat Fixed description of linkage function
The documentation says that linkage returns a 4 by n-1 matrix, but it actually has n-1 rows  and 4 columns. The rest of the paragraph uses the correct dimensions.

```
In [6]:
from scipy.cluster.hierarchy import linkage
from scipy.spatial.distance import pdist, squareform
import numpy as np
 
n=11
X = np.array(range(22)).reshape(n,2)
dist_matrix = pdist(X)
linkage(dist_matrix).shape

Out[6]:
(10, 4)
5b8a8ea
@peteryin21 peteryin21 Fixed small typo d7e4f57
Commits on May 15, 2015
@argriffing argriffing Merge pull request #4879 from peteryin21/master
DOC: Fixed small typo
3bb02c0
Commits on May 17, 2015
@cgohlke cgohlke BLD: fix MSVC compile errors.
Fix fatal error C1083: Cannot open include file: 'xmmininrin.h'
Fix error C2036: 'void *' : unknown size

(forward port of #4871, merged in 12690b0)
c574d7c
Commits on May 18, 2015
@andyfaff andyfaff DOC: differential_evolution, corrected maximum number of function eva…
…lulations
17327c6
@Eric89GXL Eric89GXL FIX: Silence warnings
(cherry picked from commit 2c37b1f)
02ea478
@rgommers rgommers MAINT: minor cleanups in ltisys and ellip_harm tests.
(cherry picked from commit 2af1e0f)
2cad7a6
@rgommers rgommers Merge branch 'forward-port-gh-4870' into master.
Cherry-pick of commits in #4870
to master.
c1fa391
@argriffing argriffing Merge pull request #4805 from argriffing/test-optimize-disp
TST: increase scipy.optimize test coverage for disp=True
3610c9e
Commits on May 19, 2015
@stefanv stefanv DOC: Fix spelling of Von Hann's surname 3e38dca
@rgommers rgommers Merge pull request #4891 from stefanv/hann
DOC: Fix spelling of Von Hann's surname
4cbf30e
@ewmoore ewmoore Merge pull request #4874 from FRidh/hilbert
DOC: Add example to scipy.signal.hilbert
5f436f7
Commits on May 20, 2015
@dlax dlax Merge pull request #4883 from andyfaff/de_doc
DOC: differential_evolution, corrected maximum number of function evaluations
989405a
Commits on May 21, 2015
@WarrenWeckesser WarrenWeckesser MAINT: signal: Remove unused imports. b836855
@WarrenWeckesser WarrenWeckesser MAINT: io: Remove unused variables. d3fcbe7
@WarrenWeckesser WarrenWeckesser MAINT: io: Remove unused import in netcdf.py. 1dc2c61
@rgommers rgommers Merge pull request #4898 from WarrenWeckesser/maint
MAINT: A bit of pyflakes-driven cleanup.
1985c47
Commits on May 22, 2015
@argriffing argriffing Merge pull request #4877 from viknat/fix-linkage-docs
DOC: Fixed description of linkage function
1b59e21