From matti.picus at gmail.com Tue May 1 01:21:13 2018 From: matti.picus at gmail.com (Matti Picus) Date: Tue, 1 May 2018 08:21:13 +0300 Subject: [Numpy-discussion] Extending ufunc signature syntax for matmul, frozen dimensions In-Reply-To: References:

Message-ID: <260099c3-500d-b7f3-a75f-c3cf8a12e8f4@gmail.com> On 01/05/18 01:45, Allan Haldane wrote: > On 04/29/2018 05:46 AM, Matti Picus wrote: >> In looking to solve issue #9028 "no way to override matmul/@ if >> __array_ufunc__ is set", it seems there is consensus around the idea of >> making matmul a true gufunc, but matmul can behave differently for >> different combinations of array and vector: >> >> (n,k),(k,m)->(n,m) >> (n,k),(k) -> (n) >> (k),(k,m)->(m) >> >> Currently there is no way to express that in the ufunc signature. The >> proposed solution to issue #9029 is to extend the meaning of a signature >> so "syntax like (n?,k),(k,m?)->(n?,m?) could mean that n and m are >> optional dimensions; if missing in the input, they're treated as 1, and >> then dropped from the output" Additionally, there is an open pull >> request #5015 "Add frozen dimensions to gufunc signatures" to allow >> signatures like '(3),(3)->(3)'. > How much harder would it be to implement multiple-dispatch for gufunc > signatures, instead of modifying the signature to include `?` ? > > There was some discussion of this last year: > > http://numpy-discussion.10968.n7.nabble.com/Changes-to-generalized-ufunc-core-dimension-checking-tp42618p42638.html > > That sounded like a clean solution to me, although I'm a bit ignorant of > the gufunc internals and the compatibility constraints. > > I assume gufuncs already have code to match the signature to the array > dims, so it sounds fairly straightforward (I say without looking at any > code) to do this in a loop over alternate signatures until one works. > > Allan > _______________________________________________ > NumPy-Discussion mailing list > NumPy-Discussion at python.org > https://mail.python.org/mailman/listinfo/numpy-discussion I will take a look at multiple-dispatch for gufuncs. The discussion also suggests using an axis kwarg when calling a gufunc for which there is PR #1108 (https://github.com/numpy/numpy/pull/11018) discussion). Matti From einstein.edison at gmail.com Tue May 1 03:06:19 2018 From: einstein.edison at gmail.com (Hameer Abbasi) Date: Tue, 1 May 2018 03:06:19 -0400 Subject: [Numpy-discussion] Extending ufunc signature syntax for matmul, frozen dimensions In-Reply-To: References:

Message-ID: I agree with Eric here. As one of the users of __array_ufunc__, I'd much rather have three separate gufuncs or a single one with axis insertion and removal. On 30/04/2018 at 23:38, Eric wrote: I think I?m -1 on this - this just makes things harder on the implementers of _array_ufunc__ who now might have to work out which signature matches. I?d prefer the solution where np.matmul is a wrapper around one of three gufuncs (or maybe just around one with axis insertion) - this is similar to how np.linalg already works. Eric On Mon, 30 Apr 2018 at 14:34 Stephan Hoyer wrote: On Sun, Apr 29, 2018 at 2:48 AM Matti Picus wrote: The proposed solution to issue #9029 is to extend the meaning of a signature so "syntax like (n?,k),(k,m?)->(n?,m?) could mean that n and m are optional dimensions; if missing in the input, they're treated as 1, and then dropped from the output" I agree that this is an elegant fix for matmul, but are there other use-cases for "optional dimensions" in gufuncs? It feels a little wrong to add gufunc features if we can only think of one function that can use them. _______________________________________________ NumPy-Discussion mailing list NumPy-Discussion at python.org https://mail.python.org/mailman/listinfo/numpy-discussion From matti.picus at gmail.com Tue May 1 03:20:57 2018 From: matti.picus at gmail.com (Matti Picus) Date: Tue, 1 May 2018 10:20:57 +0300 Subject: [Numpy-discussion] Extending ufunc signature syntax for matmul, frozen dimensions In-Reply-To: References:

Message-ID: On 01/05/18 00:38, Eric Wieser wrote: > > I think I?m -1 on this - this just makes things harder on the > implementers of |_array_ufunc__| who now might have to work out which > signature matches. I?d prefer the solution where |np.matmul| is a > wrapper around one of three gufuncs (or maybe just around one with > axis insertion) - this is similar to how np.linalg already works. > > Eric > > ? > > On Mon, 30 Apr 2018 at 14:34 Stephan Hoyer > wrote: > > On Sun, Apr 29, 2018 at 2:48 AM Matti Picus > wrote: > > The proposed solution to issue #9029 is to extend the meaning > of a signature?so "syntax like (n?,k),(k,m?)->(n?,m?) could > mean that n and m are optional dimensions; if missing in the > input, they're treated as 1, and > then dropped from the output" > > > I agree that this is an elegant fix for matmul, but are there > other use-cases for "optional dimensions" in gufuncs? > > It feels a little wrong to add gufunc features if we can only > think of one function that can use them. > _______________________________________________ > NumPy-Discussion mailing list > NumPy-Discussion at python.org > https://mail.python.org/mailman/listinfo/numpy-discussion > > > > _______________________________________________ > NumPy-Discussion mailing list > NumPy-Discussion at python.org > https://mail.python.org/mailman/listinfo/numpy-discussion I will try to prototype this solution and put it up for comment, alongside the multi-signature one. Matti From nelle.varoquaux at gmail.com Tue May 1 12:59:41 2018 From: nelle.varoquaux at gmail.com (Nelle Varoquaux) Date: Tue, 1 May 2018 09:59:41 -0700 Subject: [Numpy-discussion] summary of "office Hours" open discusison April 25 In-Reply-To: <039ab238-be1f-c83f-8a2e-8925033b102a@gmail.com> References: <039ab238-be1f-c83f-8a2e-8925033b102a@gmail.com> Message-ID: > Furher resources to consider: > - How did Jupyter organize their roadmap (ask Brian Granger)? > - How did Pandas run the project with a full time maintainer (Jeff Reback)? > - Can we copy other projects' management guidelines? > scikit-learn also has a number of full time developers. Might be worth checking out what they did. Cheers, N > > We did not set a time for another online discussion, since it was felt > that maybe near/during the sprint in May would be appropriate. > > I apologize for any misrepresentation. > > Matti Picus > > _______________________________________________ > NumPy-Discussion mailing list > NumPy-Discussion at python.org > https://mail.python.org/mailman/listinfo/numpy-discussion > -------------- next part -------------- An HTML attachment was scrubbed... URL: From shoyer at gmail.com Tue May 1 13:21:04 2018 From: shoyer at gmail.com (Stephan Hoyer) Date: Tue, 01 May 2018 17:21:04 +0000 Subject: [Numpy-discussion] summary of "office Hours" open discusison April 25 In-Reply-To: References: <039ab238-be1f-c83f-8a2e-8925033b102a@gmail.com> Message-ID: I'm happy to chat about how pandas has done things. It's worth noting that although it may *look* like Jeff Reback is a full-time maintainer (he does a lot of work!), he has actually been maintaining pandas as a side-project. Mostly the project bumbles along without a clear direction, somewhat similar to the case for NumPy for the past few years, with new contributions coming from either interested users or core developers when they have time and interest. On Tue, May 1, 2018 at 10:00 AM Nelle Varoquaux wrote: > Furher resources to consider: >> - How did Jupyter organize their roadmap (ask Brian Granger)? >> - How did Pandas run the project with a full time maintainer (Jeff >> Reback)? >> - Can we copy other projects' management guidelines? >> > > scikit-learn also has a number of full time developers. Might be worth > checking out what they did. > > Cheers, > N > > >> >> We did not set a time for another online discussion, since it was felt >> that maybe near/during the sprint in May would be appropriate. >> >> I apologize for any misrepresentation. >> >> Matti Picus >> >> _______________________________________________ >> NumPy-Discussion mailing list >> NumPy-Discussion at python.org >> https://mail.python.org/mailman/listinfo/numpy-discussion >> > _______________________________________________ > NumPy-Discussion mailing list > NumPy-Discussion at python.org > https://mail.python.org/mailman/listinfo/numpy-discussion > -------------- next part -------------- An HTML attachment was scrubbed... URL: From m.h.vankerkwijk at gmail.com Tue May 1 14:08:30 2018 From: m.h.vankerkwijk at gmail.com (Marten van Kerkwijk) Date: Tue, 1 May 2018 14:08:30 -0400 Subject: [Numpy-discussion] Extending ufunc signature syntax for matmul, frozen dimensions In-Reply-To: References:

Message-ID: Just for completeness: there are *four* gufuncs (matmat, matvec, vecmat, and vecvec). I remain torn about the best way forward. The main argument against using them inside matmul is that in order to decide which of the four to use, matmul has to have access to the `shape` of the arguments. This meants that means that `__array_ufunc__` cannot be used to override `matmul` (or `@`) for any object which does not have a shape. >From that perspective, multiple signatures is definitely a more elegant solution. An advantage of the separate solution is that they are useful independently of whether they are used internally in `matmul`; though, then again, with a multi-signature matmul, these would be trivially created as convenience functions. -- Marten From robbmcleod at gmail.com Tue May 1 16:31:41 2018 From: robbmcleod at gmail.com (Robert McLeod) Date: Tue, 1 May 2018 13:31:41 -0700 Subject: [Numpy-discussion] ANN: NumExpr 2.6.5 Message-ID: ========================== Announcing Numexpr 2.6.5 ========================== Hi everyone, This is primarily an incremental performance improvement release, especially with regards to improving import times of downstream packages (e.g. `pandas`, `tables`, `sympy`). Import times have been reduced from ~300 ms to ~100 ms through removing a `pkg_resources` import and making the `cpuinfo` import lazy. The maximum number of threads is now set at import-time, similar to `numba`, by setting an environment variable 'NUMEXPR_MAX_THREADS'. The runtime number of threads can still be reduced by calling `numexpr.set_num_threads(N)`. DEPRECATION WARNING: The variable `numexpr.is_cpu_amd_intel` has been set to a dummy value of `False`. This variable may be removed in the future. Project documentation is available at: http://numexpr.readthedocs.io/ Changes from 2.6.4 to 2.6.5 --------------------------- - The maximum thread count can now be set at import-time by setting the environment variable 'NUMEXPR_MAX_THREADS'. The default number of max threads was lowered from 4096 (which was deemed excessive) to 64. - A number of imports were removed (pkg_resources) or made lazy (cpuinfo) in order to speed load-times for downstream packages (such as `pandas`, `sympy`, and `tables`). Import time has dropped from about 330 ms to 90 ms. Thanks to Jason Sachs for pointing out the source of the slow-down. - Thanks to Alvaro Lopez Ortega for updates to benchmarks to be compatible with Python 3. - Travis and AppVeyor now fail if the test module fails or errors. - Thanks to Mahdi Ben Jelloul for a patch that removed a bug where constants in `where` calls would raise a ValueError. - Fixed a bug whereby all-constant power operations would lead to infinite recursion. -- Robert McLeod, Ph.D. robbmcleod at gmail.com robbmcleod at protonmail.com robert.mcleod at hitachi-hhtc.ca www.entropyreduction.al -------------- next part -------------- An HTML attachment was scrubbed... URL: From einstein.edison at gmail.com Wed May 2 06:24:22 2018 From: einstein.edison at gmail.com (Hameer Abbasi) Date: Wed, 2 May 2018 06:24:22 -0400 Subject: [Numpy-discussion] Extending ufunc signature syntax for matmul, frozen dimensions In-Reply-To: References:

Message-ID: There is always the option of any downstream object overriding matmul, and I fail to see which objects won't have a shape. - Hameer On 01/05/2018 at 21:08, Marten wrote: Just for completeness: there are *four* gufuncs (matmat, matvec, vecmat, and vecvec). I remain torn about the best way forward. The main argument against using them inside matmul is that in order to decide which of the four to use, matmul has to have access to the `shape` of the arguments. This meants that means that `__array_ufunc__` cannot be used to override `matmul` (or `@`) for any object which does not have a shape. >From that perspective, multiple signatures is definitely a more elegant solution. An advantage of the separate solution is that they are useful independently of whether they are used internally in `matmul`; though, then again, with a multi-signature matmul, these would be trivially created as convenience functions. -- Marten _______________________________________________ NumPy-Discussion mailing list NumPy-Discussion at python.org https://mail.python.org/mailman/listinfo/numpy-discussion -------------- next part -------------- An HTML attachment was scrubbed... URL: From m.h.vankerkwijk at gmail.com Wed May 2 11:39:41 2018 From: m.h.vankerkwijk at gmail.com (Marten van Kerkwijk) Date: Wed, 2 May 2018 11:39:41 -0400 Subject: [Numpy-discussion] Extending ufunc signature syntax for matmul, frozen dimensions In-Reply-To: References:

Message-ID: On Wed, May 2, 2018 at 6:24 AM, Hameer Abbasi wrote: > There is always the option of any downstream object overriding matmul, and I > fail to see which objects won't have a shape. - Hameer I think we should not decide too readily on what is "reasonable" to expect for a ufunc input. For instance, I'm currently writing a chained-ufunc class which uses __array_ufunc__ to help make a chain (something like `chained_ufunc = np.sin(np.multiply(Input(), Input()))`). Here, my `Input` class defines `__array_ufunc__` but definitely does not have a shape, and I would like to be able to override `np.matmul` just like every other ufunc. -- Marten From shoyer at gmail.com Wed May 2 11:47:37 2018 From: shoyer at gmail.com (Stephan Hoyer) Date: Wed, 02 May 2018 15:47:37 +0000 Subject: [Numpy-discussion] Extending ufunc signature syntax for matmul, frozen dimensions In-Reply-To: References:

Message-ID: On Wed, May 2, 2018 at 8:39 AM Marten van Kerkwijk < m.h.vankerkwijk at gmail.com> wrote: > I think we should not decide too readily on what is "reasonable" to > expect for a ufunc input. > I agree strongly with this. I can think of a couple of other use-cases off hand: - xarray.Dataset is a dict-like container of multiple arrays. Matrix-multiplication with a numpy array could make sense (just map over all the contained arrays), but xarray.Dataset itself is not an array and thus does not define shape. - tensorflow.Tensor can have a dynamic shape that is only known when computation is explicitly run, not when computation is defined in Python. The problem is even bigger for np.matmul because NumPy also wants to use the same logic for overriding @, and Python's built-in operators definitely should not have such restrictions. -------------- next part -------------- An HTML attachment was scrubbed... URL: From matti.picus at gmail.com Wed May 2 12:01:45 2018 From: matti.picus at gmail.com (Matti Picus) Date: Wed, 2 May 2018 19:01:45 +0300 Subject: [Numpy-discussion] Extending ufunc signature syntax for matmul, frozen dimensions In-Reply-To: References:

Message-ID: <25e4dbdc-b4a8-51e9-cecd-0e586bb9d805@gmail.com> On 01/05/18 21:08, Marten van Kerkwijk wrote: > Just for completeness: there are *four* gufuncs (matmat, matvec, > vecmat, and vecvec). > > I remain torn about the best way forward. The main argument against > using them inside matmul is that in order to decide which of the four > to use, matmul has to have access to the `shape` of the arguments. > This meants that means that `__array_ufunc__` cannot be used to > override `matmul` (or `@`) for any object which does not have a shape. > From that perspective, multiple signatures is definitely a more > elegant solution. > > An advantage of the separate solution is that they are useful > independently of whether they are used internally in `matmul`; though, > then again, with a multi-signature matmul, these would be trivially > created as convenience functions. > > -- Marten > _______________________________________________ > NumPy-Discussion mailing list > NumPy-Discussion at python.org > https://mail.python.org/mailman/listinfo/numpy-discussion My goal is to solve issue #9028, "no way to override matmul/@ if __array_ufunc__ is set on other". Maybe I am too focused on that, it seems shape does not come into play here. Given a call to matmul(self, other) it appears to me that the decision to commit to self.matmul or to call other.__array_ufunc__("__call__", self.matmul, ...) is independent of the shapes and needs only nin and nout. In other words, the implementation of matmul becomes (simplified): (matmul(self, other) called)-> ??? (use __array_ufunc__ and nin and nout to decide whether to defer to other's __array_ufunc__ via PyUFunc_CheckOverride which implements NEP13) -> ??????? (yes: call other.__array_ufunc__ as for any other ufunc), ??????? (no: call matmul like we currently do, no more __aray__ufunc__ testing needed) So the two avenues I am trying are 1) make matmul a gufunc and then it will automatically use the __array_ufunc__ machinery without any added changes, but this requires expanding the meaning of a signature to allow dispatch 2) generalize the __array_ufunc__ machinery to handle some kind of wrapped function, the wrapper knows about nin and nout and calls PyUFunc_CheckOverride, which would allow matmul to work unchanged and might support other functions as well. The issue of whether matmat, vecmat, matvec, and vecvec are functions, gufuncs accessible from userland, or not defined at all is secondary to the current issue of overriding matmul , we can decide that in the future. If we do create ufuncs for these variants, calling a.vecmat(other) for instance will still resolve to other's __array_ufunc__ without needing to explore other's shape. I probably misunderstood what you were driving at because I am so focused on this particular issue. Matti From m.h.vankerkwijk at gmail.com Wed May 2 12:43:44 2018 From: m.h.vankerkwijk at gmail.com (Marten van Kerkwijk) Date: Wed, 2 May 2018 12:43:44 -0400 Subject: [Numpy-discussion] Extending ufunc signature syntax for matmul, frozen dimensions In-Reply-To: <25e4dbdc-b4a8-51e9-cecd-0e586bb9d805@gmail.com> References:

<25e4dbdc-b4a8-51e9-cecd-0e586bb9d805@gmail.com> Message-ID: Hi Matti, In the original implementation of what was then __numpy_ufunc__, we had overrides for both `np.dot` and `np.matmul` that worked exactly as your option (2), but we decided in the end that those really are not true ufuncs and we should not include ufunc mimics in the mix as someone using `__array_ufunc__` should be able to count on being passed a ufunc, including all its properties. Perhaps this needs revisiting, and we should have some UFuncABC... But my own feeling remains that matmul is close enough to a (set of) gufunc that making it fit the gufunc mold is the way to go... All the best, Marten From pav at iki.fi Sat May 5 14:36:03 2018 From: pav at iki.fi (Pauli Virtanen) Date: Sat, 05 May 2018 20:36:03 +0200 Subject: [Numpy-discussion] ANN: SciPy 1.1.0 released Message-ID: <236a10a327111484ffc775d388704ebcca6500a7.camel@iki.fi> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 Hi all, On behalf of the SciPy development team I'm pleased to announce the SciPy 1.1.0 release. Sources and binary wheels can be found at https://pypi.python.org/pypi/scipy and at https://github.com/scipy/scipy/releases/tag/v1.1.0. To install with pip: pip install scipy==1.1.0 Thanks to everyone who contributed to this release! ========================= SciPy 1.1.0 Release Notes ========================= SciPy 1.1.0 is the culmination of 7 months of hard work. It contains many new features, numerous bug-fixes, improved test coverage and better documentation. There have been a number of deprecations and API changes in this release, which are documented below. All users are encouraged to upgrade to this release, as there are a large number of bug-fixes and optimizations. Before upgrading, we recommend that users check that their own code does not use deprecated SciPy functionality (to do so, run your code with ``python -Wd`` and check for ``DeprecationWarning`` s). Our development attention will now shift to bug-fix releases on the 1.1.x branch, and on adding new features on the master branch. This release requires Python 2.7 or 3.4+ and NumPy 1.8.2 or greater. This release has improved but not necessarily 100% compatibility with the `PyPy `__ Python implementation. For running on PyPy, PyPy 6.0+ and Numpy 1.15.0+ are required. New features ============ `scipy.integrate` improvements - ------------------------------ The argument ``tfirst`` has been added to the function `scipy.integrate.odeint`. This allows odeint to use the same user functions as `scipy.integrate.solve_ivp` and `scipy.integrate.ode` without the need for wrapping them in a function that swaps the first two arguments. Error messages from ``quad()`` are now clearer. `scipy.linalg` improvements - --------------------------- The function `scipy.linalg.ldl` has been added for factorization of indefinite symmetric/hermitian matrices into triangular and block diagonal matrices. Python wrappers for LAPACK ``sygst``, ``hegst`` added in `scipy.linalg.lapack`. Added `scipy.linalg.null_space`, `scipy.linalg.cdf2rdf`, `scipy.linalg.rsf2csf`. `scipy.misc` improvements - ------------------------- An electrocardiogram has been added as an example dataset for a one-dimensional signal. It can be accessed through `scipy.misc.electrocardiogram`. `scipy.ndimage` improvements - ---------------------------- The routines `scipy.ndimage.binary_opening`, and `scipy.ndimage.binary_closing` now support masks and different border values. `scipy.optimize` improvements - ----------------------------- The method ``trust-constr`` has been added to `scipy.optimize.minimize`. The method switches between two implementations depending on the problem definition. For equality constrained problems it is an implementation of a trust-region sequential quadratic programming solver and, when inequality constraints are imposed, it switches to a trust-region interior point method. Both methods are appropriate for large scale problems. Quasi-Newton options BFGS and SR1 were implemented and can be used to approximate second order derivatives for this new method. Also, finite-differences can be used to approximate either first-order or second-order derivatives. Random-to-Best/1/bin and Random-to-Best/1/exp mutation strategies were added to `scipy.optimize.differential_evolution` as ``randtobest1bin`` and ``randtobest1exp``, respectively. Note: These names were already in use but implemented a different mutation strategy. See `Backwards incompatible changes <#backwards-incompatible-changes>`__, below. The ``init`` keyword for the `scipy.optimize.differential_evolution` function can now accept an array. This array allows the user to specify the entire population. Add an ``adaptive`` option to Nelder-Mead to use step parameters adapted to the dimensionality of the problem. Minor improvements in `scipy.optimize.basinhopping`. `scipy.signal` improvements - --------------------------- Three new functions for peak finding in one-dimensional arrays were added. `scipy.signal.find_peaks` searches for peaks (local maxima) based on simple value comparison of neighbouring samples and returns those peaks whose properties match optionally specified conditions for their height, prominence, width, threshold and distance to each other. `scipy.signal.peak_prominences` and `scipy.signal.peak_widths` can directly calculate the prominences or widths of known peaks. Added ZPK versions of frequency transformations: `scipy.signal.bilinear_zpk`, `scipy.signal.lp2bp_zpk`, `scipy.signal.lp2bs_zpk`, `scipy.signal.lp2hp_zpk`, `scipy.signal.lp2lp_zpk`. Added `scipy.signal.windows.dpss`, `scipy.signal.windows.general_cosine` and `scipy.signal.windows.general_hamming`. `scipy.sparse` improvements - --------------------------- Previously, the ``reshape`` method only worked on `scipy.sparse.lil_matrix`, and in-place reshaping did not work on any matrices. Both operations are now implemented for all matrices. Handling of shapes has been made consistent with ``numpy.matrix`` throughout the `scipy.sparse` module (shape can be a tuple or splatted, negative number acts as placeholder, padding and unpadding dimensions of size 1 to ensure length-2 shape). `scipy.special` improvements - ---------------------------- Added Owen?s T function as `scipy.special.owens_t`. Accuracy improvements in ``chndtr``, ``digamma``, ``gammaincinv``, ``lambertw``, ``zetac``. `scipy.stats` improvements - -------------------------- The Moyal distribution has been added as `scipy.stats.moyal`. Added the normal inverse Gaussian distribution as `scipy.stats.norminvgauss`. Deprecated features =================== The iterative linear equation solvers in `scipy.sparse.linalg` had a sub-optimal way of how absolute tolerance is considered. The default behavior will be changed in a future Scipy release to a more standard and less surprising one. To silence deprecation warnings, set the ``atol=`` parameter explicitly. `scipy.signal.windows.slepian` is deprecated, replaced by `scipy.signal.windows.dpss`. The window functions in `scipy.signal` are now available in `scipy.signal.windows`. They will remain also available in the old location in the `scipy.signal` namespace in future Scipy versions. However, importing them from `scipy.signal.windows` is preferred, and new window functions will be added only there. Indexing sparse matrices with floating-point numbers instead of integers is deprecated. The function `scipy.stats.itemfreq` is deprecated. Backwards incompatible changes ============================== Previously, `scipy.linalg.orth` used a singular value cutoff value appropriate for double precision numbers also for single-precision input. The cutoff value is now tunable, and the default has been changed to depend on the input data precision. In previous versions of Scipy, the ``randtobest1bin`` and ``randtobest1exp`` mutation strategies in `scipy.optimize.differential_evolution` were actually implemented using the Current-to-Best/1/bin and Current-to-Best/1/exp strategies, respectively. These strategies were renamed to ``currenttobest1bin`` and ``currenttobest1exp`` and the implementations of ``randtobest1bin`` and ``randtobest1exp`` strategies were corrected. Functions in the ndimage module now always return their output array. Before this most functions only returned the output array if it had been allocated by the function, and would return ``None`` if it had been provided by the user. Distance metrics in `scipy.spatial.distance` now require non-negative weights. `scipy.special.loggamma` returns now real-valued result when the input is real-valued. Other changes ============= When building on Linux with GNU compilers, the ``.so`` Python extension files now hide all symbols except those required by Python, which can avoid problems when embedding the Python interpreter. Authors ======= * Saurabh Agarwal + * Diogo Aguiam + * Joseph Albert + * Gerrit Ansmann + * Jean-Fran?ois B + * Vahan Babayan + * Alessandro Pietro Bardelli * Christoph Baumgarten + * Felix Berkenkamp * Lilian Besson + * Aditya Bharti + * Matthew Brett * Evgeni Burovski * CJ Carey * Martin ?. Christensen + * Robert Cimrman * Vicky Close + * Peter Cock + * Philip DeBoer * Jaime Fernandez del Rio * Dieter Werthm?ller + * Tom Donoghue + * Matt Dzugan + * Lars G + * Jacques Gaudin + * Andriy Gelman + * Sean Gillies + * Dezmond Goff * Christoph Gohlke * Ralf Gommers * Uri Goren + * Deepak Kumar Gouda + * Douglas Lessa Graciosa + * Matt Haberland * David Hagen * Charles Harris * Jordan Heemskerk + * Danny Hermes + * Stephan Hoyer + * Theodore Hu + * Jean-Fran?ois B. + * Mads Jensen + * Jon Haitz Legarreta Gorro?o + * Ben Jude + * Noel Kippers + * Julius Bier Kirkegaard + * Maria Knorps + * Mikkel Kristensen + * Eric Larson * Kasper Primdal Lauritzen + * Denis Laxalde * KangWon Lee + * Jan Lehky + * Jackie Leng + * P.L. Lim + * Nikolay Mayorov * Mihai Capot? + * Max Mikhaylov + * Mark Mikofski + * Jarrod Millman * Raden Muhammad + * Paul Nation * Andrew Nelson * Nico Schl?mer * Joel Nothman * Kyle Oman + * Egor Panfilov + * Nick Papior * Anubhav Patel + * Oleksandr Pavlyk * Ilhan Polat * Robert Pollak + * Anant Prakash + * Aman Pratik * Sean Quinn + * Giftlin Rajaiah + * Tyler Reddy * Joscha Reimer * Antonio H Ribeiro + * Antonio Horta Ribeiro * Benjamin Rose + * Fabian Rost * Divakar Roy + * Scott Sievert * Leo Singer * Sourav Singh * Martino Sorbaro + * Eric Stansifer + * Martin Thoma * Phil Tooley + * Piotr Uchwat + * Paul van Mulbregt * Pauli Virtanen * Stefan van der Walt * Warren Weckesser * Florian Weimer + * Eric Wieser * Josh Wilson * Ted Ying + * Evgeny Zhurko * Z? Vin?cius * @Astrofysicus + * @awakenting + * @endolith * @FormerPhysicist + * @gaulinmp + * @hugovk * @ksemb + * @kshitij12345 + * @luzpaz + * @NKrvavica + * @rafalalgo + * @samyak0210 + * @soluwalana + * @sudheerachary + * @Tokixix + * @tttthomasssss + * @vkk800 + * @xoviat * @ziejcow + A total of 122 people contributed to this release. People with a "+" by their names contributed a patch for the first time. This list of names is automatically generated, and may not be fully complete. Issues closed for 1.1.0 - ----------------------- * `#979 `__: Allow Hermitian matrices in lobpcg (Trac #452) * `#2694 `__: Solution of iterative solvers can be less accurate than tolerance... * `#3164 `__: RectBivariateSpline usage inconsistent with other interpolation... * `#4161 `__: Missing ITMAX optional argument in scipy.optimize.nnls * `#4354 `__: signal.slepian should use definition of digital window * `#4866 `__: Shouldn't scipy.linalg.sqrtm raise an error if matrix is singular? * `#4953 `__: The dirichlet distribution unnecessarily requires strictly positive... * `#5336 `__: sqrtm on a diagonal matrix can warn "Matrix is singular and may... * `#5922 `__: Suboptimal convergence of Halley's method? * `#6036 `__: Incorrect edge case in scipy.stats.triang.pdf * `#6202 `__: Enhancement: Add LDLt factorization to scipy * `#6589 `__: sparse.random with custom rvs callable does pass on arg to subclass * `#6654 `__: Spearman's rank correlation coefficient slow with nan values... * `#6794 `__: Remove NumarrayType struct with numarray type names from ndimage * `#7136 `__: The dirichlet distribution unnecessarily rejects probabilities... * `#7169 `__: Will it be possible to add LDL' factorization for Hermitian indefinite... * `#7291 `__: fsolve docs should say it doesn't handle over- or under-determined... * `#7453 `__: binary_opening/binary_closing missing arguments * `#7500 `__: linalg.solve test failure on OS X with Accelerate * `#7555 `__: Integratig a function with singularities using the quad routine * `#7624 `__: allow setting both absolute and relative tolerance of sparse... * `#7724 `__: odeint documentation refers to t0 instead of t * `#7746 `__: False CDF values for skew normal distribution * `#7750 `__: mstats.winsorize documentation needs clarification * `#7787 `__: Documentation error in spherical Bessel, Neumann, modified spherical... * `#7836 `__: Scipy mmwrite incorrectly writes the zeros for skew-symmetric,... * `#7839 `__: sqrtm is unable to compute square root of zero matrix * `#7847 `__: solve is very slow since #6775 * `#7888 `__: Scipy 1.0.0b1 prints spurious DVODE/ZVODE/lsoda messages * `#7909 `__: bessel kv function in 0 is nan * `#7915 `__: LinearOperator's __init__ runs two times when instantiating the... * `#7958 `__: integrate.quad could use better error messages when given bad... * `#7968 `__: integrate.quad handles decreasing limits (b`__: ENH: matching return dtype for loggamma/gammaln * `#7991 `__: `lfilter` segfaults for integer inputs * `#8076 `__: "make dist" for the docs doesn't complete cleanly * `#8080 `__: Use JSON in `special/_generate_pyx.py`? * `#8127 `__: scipy.special.psi(x) very slow for some values of x * `#8145 `__: BUG: ndimage geometric_transform and zoom using deprecated NumPy... * `#8158 `__: BUG: romb print output requires correction * `#8181 `__: loadmat() raises TypeError instead of FileNotFound when reading... * `#8228 `__: bug for log1p on csr_matrix * `#8235 `__: scipy.stats multinomial pmf return nan * `#8271 `__: scipy.io.mmwrite raises type error for uint16 * `#8288 `__: Should tests be written for scipy.sparse.linalg.isolve.minres... * `#8298 `__: Broken links on scipy API web page * `#8329 `__: `_gels` fails for fat A matrix * `#8346 `__: Avoidable overflow in scipy.special.binom(n, k) * `#8371 `__: BUG: special: zetac(x) returns 0 for x < -30.8148 * `#8382 `__: collections.OrderedDict in test_mio.py * `#8492 `__: Missing documentation for `brute_force` parameter in scipy.ndimage.morphology * `#8532 `__: leastsq needlessly appends extra dimension for scalar problems * `#8544 `__: [feature request] Convert complex diagonal form to real block... * `#8561 `__: [Bug?] Example of Bland's Rule for optimize.linprog (simplex)... * `#8562 `__: CI: Appveyor builds fail because it can't import ConvexHull from... * `#8576 `__: BUG: optimize: `show_options(solver='minimize', method='Newton-CG')`... * `#8603 `__: test_roots_gegenbauer/chebyt/chebyc failures on manylinux * `#8604 `__: Test failures in scipy.sparse test_inplace_dense * `#8616 `__: special: ellpj.c code can be cleaned up a bit * `#8625 `__: scipy 1.0.1 no longer allows overwriting variables in netcdf... * `#8629 `__: gcrotmk.test_atol failure with MKL * `#8632 `__: Sigma clipping on data with the same value * `#8646 `__: scipy.special.sinpi test failures in test_zero_sign on old MSVC * `#8663 `__: linprog with method=interior-point produced incorrect answer... * `#8694 `__: linalg:TestSolve.test_all_type_size_routine_combinations fails... * `#8703 `__: Q: Does runtests.py --refguide-check need env (or other) variables... Pull requests for 1.1.0 - ----------------------- * `#6590 `__: BUG: sparse: fix custom rvs callable argument in sparse.random * `#7004 `__: ENH: scipy.linalg.eigsh cannot get all eigenvalues * `#7120 `__: ENH: implemented Owen's T function * `#7483 `__: ENH: Addition/multiplication operators for StateSpace systems * `#7566 `__: Informative exception when passing a sparse matrix * `#7592 `__: Adaptive Nelder-Mead * `#7729 `__: WIP: ENH: optimize: large-scale constrained optimization algorithms... * `#7802 `__: MRG: Add dpss window function * `#7803 `__: DOC: Add examples to spatial.distance * `#7821 `__: Add Returns section to the docstring * `#7833 `__: ENH: Performance improvements in scipy.linalg.special_matrices * `#7864 `__: MAINT: sparse: Simplify sputils.isintlike * `#7865 `__: ENH: Improved speed of copy into L, U matrices * `#7871 `__: ENH: sparse: Add 64-bit integer to sparsetools * `#7879 `__: ENH: re-enabled old sv lapack routine as defaults * `#7889 `__: DOC: Show probability density functions as math * `#7900 `__: API: Soft deprecate signal.* windows * `#7910 `__: ENH: allow `sqrtm` to compute the root of some singular matrices * `#7911 `__: MAINT: Avoid unnecessary array copies in xdist * `#7913 `__: DOC: Clarifies the meaning of `initial` of scipy.integrate.cumtrapz() * `#7916 `__: BUG: sparse.linalg: fix wrong use of __new__ in LinearOperator * `#7921 `__: BENCH: split spatial benchmark imports * `#7927 `__: ENH: added sygst/hegst routines to lapack * `#7934 `__: MAINT: add `io/_test_fortranmodule` to `.gitignore` * `#7936 `__: DOC: Fixed typo in scipy.special.roots_jacobi documentation * `#7937 `__: MAINT: special: Mark a test that fails on i686 as a known failure. * `#7941 `__: ENH: LDLt decomposition for indefinite symmetric/hermitian matrices * `#7945 `__: ENH: Implement reshape method on sparse matrices * `#7947 `__: DOC: update docs on releasing and installing/upgrading * `#7954 `__: Basin-hopping changes * `#7964 `__: BUG: test_falker not robust against numerical fuss in eigenvalues * `#7967 `__: QUADPACK Errors - human friendly errors to replace 'Invalid Input' * `#7975 `__: Make sure integrate.quad doesn't double-count singular points * `#7978 `__: TST: ensure negative weights are not allowed in distance metrics * `#7980 `__: MAINT: Truncate the warning msg about ill-conditioning * `#7981 `__: BUG: special: fix hyp2f1 behavior in certain circumstances * `#7983 `__: ENH: special: Add a real dispatch to `loggamma` * `#7989 `__: BUG: special: make `kv` return `inf` at a zero real argument * `#7990 `__: TST: special: test ufuncs in special at `nan` inputs * `#7994 `__: DOC: special: fix typo in spherical Bessel function documentation * `#7995 `__: ENH: linalg: add null_space for computing null spaces via svd * `#7999 `__: BUG: optimize: Protect _minpack calls with a lock. * `#8003 `__: MAINT: consolidate c99 compatibility * `#8004 `__: TST: special: get all `cython_special` tests running again * `#8006 `__: MAINT: Consolidate an additional _c99compat.h * `#8011 `__: Add new example of integrate.quad * `#8015 `__: DOC: special: remove `jn` from the refguide (again) * `#8018 `__: BUG - Issue with uint datatypes for array in get_index_dtype * `#8021 `__: DOC: spatial: Simplify Delaunay plotting * `#8024 `__: Documentation fix * `#8027 `__: BUG: io.matlab: fix saving unicode matrix names on py2 * `#8028 `__: BUG: special: some fixes for `lambertw` * `#8030 `__: MAINT: Bump Cython version * `#8034 `__: BUG: sparse.linalg: fix corner-case bug in expm * `#8035 `__: MAINT: special: remove complex division hack * `#8038 `__: ENH: Cythonize pyx files if pxd dependencies change * `#8042 `__: TST: stats: reduce required precision in test_fligner * `#8043 `__: TST: Use diff. values for decimal keyword for single and doubles * `#8044 `__: TST: accuracy of tests made different for singles and doubles * `#8049 `__: Unhelpful error message when calling scipy.sparse.save_npz on... * `#8052 `__: TST: spatial: add a regression test for gh-8051 * `#8059 `__: BUG: special: fix ufunc results for `nan` arguments * `#8066 `__: MAINT: special: reimplement inverses of incomplete gamma functions * `#8072 `__: Example for scipy.fftpack.ifft, https://github.com/scipy/scipy/issues/7168 * `#8073 `__: Example for ifftn, https://github.com/scipy/scipy/issues/7168 * `#8078 `__: Link to CoC in contributing.rst doc * `#8085 `__: BLD: Fix npy_isnan of integer variables in cephes * `#8088 `__: DOC: note version for which new attributes have been added to... * `#8090 `__: BUG: special: add nan check to `_legacy_cast_check` functions * `#8091 `__: Doxy Typos + trivial comment typos (2nd attempt) * `#8096 `__: TST: special: simplify `Arg` * `#8101 `__: MAINT: special: run `_generate_pyx.py` when `add_newdocs.py`... * `#8104 `__: Input checking for scipy.sparse.linalg.inverse() * `#8105 `__: DOC: special: Update the 'euler' docstring. * `#8109 `__: MAINT: fixing code comments and hyp2f1 docstring: see issues... * `#8112 `__: More trivial typos * `#8113 `__: MAINT: special: generate test data npz files in setup.py and... * `#8116 `__: DOC: add build instructions * `#8120 `__: DOC: Clean up README * `#8121 `__: DOC: Add missing colons in docstrings * `#8123 `__: BLD: update Bento build config files for recent C99 changes. * `#8124 `__: Change to avoid use of `fmod` in scipy.signal.chebwin * `#8126 `__: Added examples for mode arg in geometric_transform * `#8128 `__: relax relative tolerance parameter in TestMinumumPhase.test_hilbert * `#8129 `__: ENH: special: use rational approximation for `digamma` on `[1,... * `#8137 `__: DOC Correct matrix width * `#8141 `__: MAINT: optimize: remove unused `__main__` code in L-BSGS-B * `#8147 `__: BLD: update Bento build for removal of .npz scipy.special test... * `#8148 `__: Alias hanning as an explanatory function of hann * `#8149 `__: MAINT: special: small fixes for `digamma` * `#8159 `__: Update version classifiers * `#8164 `__: BUG: riccati solvers don't catch ill-conditioned problems sufficiently... * `#8168 `__: DOC: release note for sparse resize methods * `#8170 `__: BUG: correctly pad netCDF files with null bytes * `#8171 `__: ENH added normal inverse gaussian distribution to scipy.stats * `#8175 `__: DOC: Add example to scipy.ndimage.zoom * `#8177 `__: MAINT: diffev small speedup in ensure constraint * `#8178 `__: FIX: linalg._qz String formatter syntax error * `#8179 `__: TST: Added pdist to asv spatial benchmark suite * `#8180 `__: TST: ensure constraint test improved * `#8183 `__: 0d conj correlate * `#8186 `__: BUG: special: fix derivative of `spherical_jn(1, 0)` * `#8194 `__: Fix warning message * `#8196 `__: BUG: correctly handle inputs with nan's and ties in spearmanr * `#8198 `__: MAINT: stats.triang edge case fixes #6036 * `#8200 `__: DOC: Completed "Examples" sections of all linalg funcs * `#8201 `__: MAINT: stats.trapz edge cases * `#8204 `__: ENH: sparse.linalg/lobpcg: change .T to .T.conj() to support... * `#8206 `__: MAINT: missed triang edge case. * `#8214 `__: BUG: Fix memory corruption in linalg._decomp_update C extension * `#8222 `__: DOC: recommend scipy.integrate.solve_ivp * `#8223 `__: ENH: added Moyal distribution to scipy.stats * `#8232 `__: BUG: sparse: Use deduped data for numpy ufuncs * `#8236 `__: Fix #8235 * `#8253 `__: BUG: optimize: fix bug related with function call calculation... * `#8264 `__: ENH: Extend peak finding capabilities in scipy.signal * `#8273 `__: BUG fixed printing of convergence message in minimize_scalar... * `#8276 `__: DOC: Add notes to explain constrains on overwrite_<> * `#8279 `__: CI: fixing doctests * `#8282 `__: MAINT: weightedtau, change search for nan * `#8287 `__: Improving documentation of solve_ivp and the underlying solvers * `#8291 `__: DOC: fix non-ascii characters in docstrings which broke the doc... * `#8292 `__: CI: use numpy 1.13 for refguide check build * `#8296 `__: Fixed bug reported in issue #8181 * `#8297 `__: DOC: Examples for linalg/decomp eigvals function * `#8300 `__: MAINT: Housekeeping for minimizing the linalg compiler warnings * `#8301 `__: DOC: make public API documentation cross-link to refguide. * `#8302 `__: make sure _onenorm_matrix_power_nnm actually returns a float * `#8313 `__: Change copyright to outdated 2008-2016 to 2008-year * `#8315 `__: TST: Add tests for `scipy.sparse.linalg.isolve.minres` * `#8318 `__: ENH: odeint: Add the argument 'tfirst' to odeint. * `#8328 `__: ENH: optimize: ``trust-constr`` optimization algorithms [GSoC... * `#8330 `__: ENH: add a maxiter argument to NNLS * `#8331 `__: DOC: tweak the Moyal distribution docstring * `#8333 `__: FIX: Rewrapped ?gels and ?gels_lwork routines * `#8336 `__: MAINT: integrate: handle b < a in quad * `#8337 `__: BUG: special: Ensure zetac(1) returns inf. * `#8347 `__: BUG: Fix overflow in special.binom. Issue #8346 * `#8356 `__: DOC: Corrected Documentation Issue #7750 winsorize function * `#8358 `__: ENH: stats: Use explicit MLE formulas in lognorm.fit and expon.fit * `#8374 `__: BUG: gh7854, maxiter for l-bfgs-b closes #7854 * `#8379 `__: CI: enable gcov coverage on travis * `#8383 `__: Removed collections.OrderedDict import ignore. * `#8384 `__: TravisCI: tool pep8 is now pycodestyle * `#8387 `__: MAINT: special: remove unused specfun code for Struve functions * `#8393 `__: DOC: Replace old type names in ndimage tutorial. * `#8400 `__: Fix tolerance specification in sparse.linalg iterative solvers * `#8402 `__: MAINT: Some small cleanups in ndimage. * `#8403 `__: FIX: Make scipy.optimize.zeros run under PyPy * `#8407 `__: BUG: sparse.linalg: fix termination bugs for cg, cgs * `#8409 `__: MAINT: special: add a `.pxd` file for Cephes functions * `#8412 `__: MAINT: special: remove `cephes/protos.h` * `#8421 `__: Setting "unknown" message in OptimizeResult when calling MINPACK. * `#8423 `__: FIX: Handle unsigned integers in mmio * `#8426 `__: DOC: correct FAQ entry on Apache license compatibility. Closes... * `#8433 `__: MAINT: add `.pytest_cache` to the `.gitignore` * `#8436 `__: MAINT: scipy.sparse: less copies at transpose method * `#8437 `__: BUG: correct behavior for skew-symmetric matrices in io.mmwrite * `#8440 `__: DOC:Add examples to integrate.quadpack docstrings * `#8441 `__: BUG: sparse.linalg/gmres: deal with exact breakdown in gmres * `#8442 `__: MAINT: special: clean up Cephes header files * `#8448 `__: TST: Generalize doctest stopwords .axis( .plot( * `#8457 `__: MAINT: special: use JSON for function signatures in `_generate_pyx.py` * `#8461 `__: MAINT: Simplify return value of ndimage functions. * `#8464 `__: MAINT: Trivial typos * `#8474 `__: BUG: spatial: make qhull.pyx more pypy-friendly * `#8476 `__: TST: _lib: disable refcounting tests on PyPy * `#8479 `__: BUG: io/matlab: fix issues in matlab i/o on pypy * `#8481 `__: DOC: Example for signal.cmplx_sort * `#8482 `__: TST: integrate: use integers instead of PyCapsules to store pointers * `#8483 `__: ENH: io/netcdf: make mmap=False the default on PyPy * `#8484 `__: BUG: io/matlab: work around issue in to_writeable on PyPy * `#8488 `__: MAINT: special: add const/static specifiers where possible * `#8489 `__: BUG: ENH: use common halley's method instead of parabolic variant * `#8491 `__: DOC: fix typos * `#8496 `__: ENH: special: make Chebyshev nodes symmetric * `#8501 `__: BUG: stats: Split the integral used to compute skewnorm.cdf. * `#8502 `__: WIP: Port CircleCI to v2 * `#8507 `__: DOC: Add missing description to `brute_force` parameter. * `#8509 `__: BENCH: forgot to add nelder-mead to list of methods * `#8512 `__: MAINT: Move spline interpolation code to spline.c * `#8513 `__: TST: special: mark a slow test as xslow * `#8514 `__: CircleCI: Share data between jobs * `#8515 `__: ENH: special: improve accuracy of `zetac` for negative arguments * `#8520 `__: TST: Decrease the array sizes for two linalg tests * `#8522 `__: TST: special: restrict range of `test_besselk`/`test_besselk_int` * `#8527 `__: Documentation - example added for voronoi_plot_2d * `#8528 `__: DOC: Better, shared docstrings in ndimage * `#8533 `__: BUG: Fix PEP8 errors introduced in #8528. * `#8534 `__: ENH: Expose additional window functions * `#8538 `__: MAINT: Fix a couple mistakes in .pyf files. * `#8540 `__: ENH: interpolate: allow string aliases in make_interp_spline... * `#8541 `__: ENH: Cythonize peak_prominences * `#8542 `__: Remove numerical arguments from convolve2d / correlate2d * `#8546 `__: ENH: New arguments, documentation, and tests for ndimage.binary_opening * `#8547 `__: Giving both size and input now raises UserWarning (#7334) * `#8549 `__: DOC: stats: invweibull is also known as Frechet or type II extreme... * `#8550 `__: add cdf2rdf function * `#8551 `__: ENH: Port of most of the dd_real part of the qd high-precision... * `#8553 `__: Note in docs to address issue #3164. * `#8554 `__: ENH: stats: Use explicit MLE formulas in uniform.fit() * `#8555 `__: MAINT: adjust benchmark config * `#8557 `__: [DOC]: fix Nakagami density docstring * `#8559 `__: DOC: Fix docstring of diric(x, n) * `#8563 `__: [DOC]: fix gamma density docstring * `#8564 `__: BLD: change default Python version for doc build from 2.7 to... * `#8568 `__: BUG: Fixes Bland's Rule for pivot row/leaving variable, closes... * `#8572 `__: ENH: Add previous/next to interp1d * `#8578 `__: Example for linalg.eig() * `#8580 `__: DOC: update link to asv docs * `#8584 `__: filter_design: switch to explicit arguments, keeping None as... * `#8586 `__: DOC: stats: Add parentheses that were missing in the exponnorm... * `#8587 `__: TST: add benchmark for newton, secant, halley * `#8588 `__: DOC: special: Remove heaviside from "functions not in special"... * `#8591 `__: DOC: cdf2rdf Added version info and "See also" * `#8594 `__: ENH: Cythonize peak_widths * `#8595 `__: MAINT/ENH/BUG/TST: cdf2rdf: Address review comments made after... * `#8597 `__: DOC: add versionadded 1.1.0 for new keywords in ndimage.morphology * `#8605 `__: MAINT: special: improve implementations of `sinpi` and `cospi` * `#8607 `__: MAINT: add 2D benchmarks for convolve * `#8608 `__: FIX: Fix int check * `#8613 `__: fix typo in doc of signal.peak_widths * `#8615 `__: TST: fix failing linalg.qz float32 test by decreasing precision. * `#8617 `__: MAINT: clean up code in ellpj.c * `#8618 `__: add fsolve docs it doesn't handle over- or under-determined problems * `#8620 `__: DOC: add note on dtype attribute of aslinearoperator() argument * `#8627 `__: ENH: Add example 1D signal (ECG) to scipy.misc * `#8630 `__: ENH: Remove unnecessary copying in stats.percentileofscore * `#8631 `__: BLD: fix pdf doc build. closes gh-8076 * `#8633 `__: BUG: fix regression in `io.netcdf_file` with append mode. * `#8635 `__: MAINT: remove spurious warning from (z)vode and lsoda. Closes... * `#8636 `__: BUG: sparse.linalg/gcrotmk: avoid rounding error in termination... * `#8637 `__: For pdf build * `#8639 `__: CI: build pdf documentation on circleci * `#8640 `__: TST: fix special test that was importing `np.testing.utils` (deprecated) * `#8641 `__: BUG: optimize: fixed sparse redundancy removal bug * `#8645 `__: BUG: modified sigmaclip to avoid clipping of constant input in... * `#8647 `__: TST: sparse: skip test_inplace_dense for numpy<1.13 * `#8657 `__: Latex reduce left margins * `#8659 `__: TST: special: skip sign-of-zero test on 32-bit win32 with old... * `#8661 `__: Fix dblquad and tplquad not accepting float boundaries * `#8666 `__: DOC: fixes #8532 * `#8667 `__: BUG: optimize: fixed issue #8663 * `#8668 `__: Fix example in docstring of netcdf_file * `#8671 `__: DOC: Replace deprecated matplotlib kwarg * `#8673 `__: BUG: special: Use a stricter tolerance for the chndtr calculation. * `#8674 `__: ENH: In the Dirichlet distribution allow x_i to be 0 if alpha_i... * `#8676 `__: BUG: optimize: partial fix to linprog fails to detect infeasibility... * `#8685 `__: DOC: Add interp1d-next/previous example to tutorial * `#8687 `__: TST: netcdf: explicit mmap=True in test * `#8688 `__: BUG: signal, stats: use Python sum() instead of np.sum for summing... * `#8689 `__: TST: bump tolerances in tests * `#8690 `__: DEP: deprecate stats.itemfreq * `#8691 `__: BLD: special: fix build vs. dd_real.h package * `#8695 `__: DOC: Improve examples in signal.find_peaks with ECG signal * `#8697 `__: BUG: Fix `setup.py build install egg_info`, which did not previously... * `#8704 `__: TST: linalg: drop large size from solve() test * `#8705 `__: DOC: Describe signal.find_peaks and related functions behavior... * `#8706 `__: DOC: Specify encoding of rst file, remove an ambiguity in an... * `#8710