Mailman 3 December 2021 - SciPy-Dev

NumPy 1.22. 0rc2 Release
by Charles R Harris Dec. 10, 2021

Dec. 10, 2021

Hi All, On behalf of the NumPy team, I'm pleased to announce the release of NumPy 1.22.0rc1. NumPy 1.22.0rc1 is a big release featuring the work of 151 contributers spread over 589 pull requests. There have been many improvements, highlights are: - Annotations of the main namespace are essentially complete. Upstream is a moving target, so there will likely be further improvements, but the major work is done. This is probably the most user visible enhancement in this release. - A preliminary version of the proposed Array-API is provided. This is a step in creating a standard collection of functions that can be used across applications such as CuPy and JAX. - NumPy now has a DLPack backend. DLPack provides a common interchange format for array (tensor) data. - New methods for `quantile`, `percentile`, and related functions. Thenew methods provide a complete set of the methods commonly found in the literature. - A new configurable allocator for use by downstream projects. These are in addition to the ongoing work to provide SIMD support for commonly used functions, improvements to F2PY, and better documentation. The Python versions supported in this release are 3.8-3.10, Python 3.7 has been dropped. Note that 32 bit wheels are only provided for Python 3.8 and 3.9 on Windows, all other wheels are 64 bits on account of Ubuntu, Fedora, and other Linux distributions dropping 32 bit support. All 64 bit wheels are also linked with 64 bit integer OpenBLAS, which should fix the occasional problems encountered by folks using truly huge arrays. Wheels can be downloaded from PyPI <https://pypi.org/project/numpy/1.22.0rc2/>; source archives, release notes, and wheel hashes are available on Github <https://github.com/numpy/numpy/releases/tag/v1.22.0rc2>. Linux users will need pip >= 0.19.3 in order to install the manylinux2014 wheels. A recent version of pip is needed to install the universal2 macos wheels. *Contributors* A total of 151 people contributed to this release. People with a "+" by their names contributed a patch for the first time. - @DWesl - @Illviljan - @h-vetinari - @yan-wyb + - Aaron Meurer - Abel Aoun + - Adrian Gao + - Ahmet Can Solak + - Ajay DS + - Alban Colley + - Alberto Rubiales + - Alessia Marcolini + - Amit Kumar + - Andrei Batomunkuev + - Andrew Watson + - Anirudh Dagar + - Ankit Dwivedi + - Antony Lee - Arfy Slowy + - Arryan Singh + - Arun Palaniappen + - Arushi Sharma + - Bas van Beek - Brent Brewington + - Carl Johnsen + - Carl Michal + - Charles Harris - Chiara Marmo - Chris Fu (傅立业) + - Christoph Buchner + - Christoph Reiter + - Chunlin Fang - Clément Robert + - Constanza Fierro - Damien Caliste - Daniel Ching - David Badnar + - David Cortes + - David Okpare + - Derek Huang + - Developer-Ecosystem-Engineering + - Dima Pasechnik - Dimitri Papadopoulos + - Dmitriy Fishman + - Eero Vaher + - Elias Koromilas + - Eliaz Bobadilla + - Elisha Hollander + - Eric Wieser - Eskild Eriksen + - Evan Miller + - Fayas Noushad + - Gagandeep Singh + - Ganesh Kathiresan - Ghiles Meddour + - Greg Lucas - Gregory R. Lee - Guo Shuai + - Gwyn Ciesla + - Hameer Abbasi - Hector Martin + - Henry Schreiner + - Himanshu + - Hood Chatham + - Hugo Defois + - Hugo van Kemenade - I-Shen Leong + - Imen Rajhi + - Irina Maria Mocan + - Irit Katriel + - Isuru Fernando - Jakob Jakobson - Jerry Morrison + - Jessi J Zhao + - Joe Marshall + - Johan von Forstner + - Jonas I. Liechti + - Jonathan Reichelt Gjertsen + - Joshua Himmens + - Jérome Eertmans - Jérôme Kieffer + - KIU Shueng Chuan + - Kenichi Maehashi - Kenny Huynh + - Kent R. Spillner + - Kevin Granados + - Kevin Modzelewski + - Kevin Sheppard - Lalit Musmade + - Malik Idrees Hasan Khan + - Marco Aurelio da Costa + - Margret Pax + - Mars Lee + - Marten van Kerkwijk - Matthew Barber + - Matthew Brett - Matthias Bussonnier - Matthieu Dartiailh - Matti Picus - Melissa Weber Mendonça - Michael McCann + - Mike Jarvis + - Mike McCann + - Mike Toews - Mukulika Pahari - Nick Pope + - Nick Wogan + - Niels Dunnewind + - Niko Savola + - Nikola Forró - Niyas Sait + - Pamphile ROY - Paul Ganssle + - Pauli Virtanen - Pearu Peterson - Peter Hawkins + - Peter Tillema + - Prathmesh Shirsat + - Raghuveer Devulapalli - Ralf Gommers - Robert Kern - Rohit Goswami + - Ronan Lamy - Ross Barnowski - Roy Jacobson + - Samyak S Sarnayak + - Sayantika Banik + - Sayed Adel - Sebastian Berg - Sebastian Schleehauf + - Serge Guelton - Shriraj Hegde + - Shubham Gupta + - Sista Seetaram + - Stefan van der Walt - Stephannie Jimenez Gacha + - Tania Allard - Theodoros Nikolaou + - Thomas J. Fan - Thomas Li + - Tim Hoffmann - Tom Tan + - Tyler Reddy - Vijay Arora + - Vinith Kishore + - Warren Weckesser - Yang Hau - Yashasvi Misra - Yuval Ofek + - Zac Hatfield-Dodds - Zhang Na + Cheers, Charles Harris

1 0

Discussing a small feature contribution in scipy.signal.csd
by Anchal Gupta Dec. 7, 2021

Dec. 7, 2021

Hi everyone I work at Laser Interferometer Gravitational-Wave Observatory (LIGO) where we make spectral density estimations of various signals and noises in our system. It sometimes becomes necessary to know the uncertainty associated with each calculated point in a bin of the spectral density calculation. I, in particular, wanted to measure the exact probability density of the data in each bin before it is averaged over. When I looked at the scipy.signal.spectral.py, I saw that the information exists after the computation by _spectral_helper() but it is thrown away at the step where averaging happens. *My suggestion:*With just a few lines of code, one could add a new boolean argument getUnc (pardon my camel-case), which if false will keep everything the same as existing code but if true, the function csd will return standard deviation (in case of average='mean') or lower and upper percentile bounds (in case of average='median', percentile value can be given by another argument). Additionally, if one allows average='none', we can get the higher dimensional spectral density array for each segment. This can be used to estimate the probability density of the spread of data in each spectral bin. Now, I have already made a small package which creates this new version of csd and welch function so that we can use it in our labs. The code is present here: https://gitlab.com/anchal-physics/csdTools/-/tree/master and available at PyPi: https://pypi.org/project/csdTools/ I'm open to creating a pull request to add this feature in scipy. I think the following things need to be discussed for it though: 1) Statistical validity of estimating empirical uncertainty this way. 2) How to handle the return. At the moment, if getUnc=True, my code returns more number of objects. This can be made simpler by adding an axis to returned array (maybe). 3) Are there other dependent functions that can utilize the extended feature in some way? I'm unaware of any such functions right now. Please let me know if you would rather like me to put this in an issue with the enhancement tag. Thanks Anchal Gupta Ph.D. Candidate of Physics California Institute of Technology

3 3

Master branch now open for SciPy 1.9.0 development
by Tyler Reddy Dec. 6, 2021

Dec. 6, 2021

The maintenance/1.8.x branch has been pushed and master branch is now open for the development of SciPy 1.9.0. Best wishes, Tyler

1 0

Proposed Release Schedule for SciPy 1.8.0
by Tyler Reddy Dec. 4, 2021

Dec. 4, 2021

Hi, I'm already pretty late sending this out, and I've tried to weave some of the dates around typical US Holidays at least. I'd expect some suggestions for changes on this schedule may be likely. Two things that I need to catch up on, and that have me slightly nervous, are: 1) binary wheel support for Mac M1 architecture 2) a way to deploy the new docs theme with a release that doesn't break intersphinx/links everywhere Rough Proposal: - December 6/2021: branch 1.8.x - December 9/2021: rc1 - December 20/2021: rc2 (if needed) - January 3/2022: final release I realize this likely puts us a bit behind NumPy/our desired cadence, though not that different from Dec. 31/2020 release last year. December may sound far, but that's only 4 weekends before branching. To be fair, Python 3.10 and M1 macs combined has pushed schedules/bandwidth a bit. Best wishes, Tyler

3 6

request for feedback on cross-library design proposal
by Ralf Gommers Dec. 1, 2021

Dec. 1, 2021

(Apologies for the cross-posting, if you follow more than one of these lists, you may see this multiple times.) Hi all, I'd like to solicit the feedback of maintainers of NumPy, SciPy, scikit-learn, and scikit-image, and other interested community members on this blog post and proposed design to enable multiple array types in SciPy, scikit-learn, scikit-image and other libraries in this ecosystem: https://labs.quansight.org/blog/2021/11/pydata-extensibility-vision/. Because this design is meant to be shared across projects, it would be great to discuss it in a central place. The new https://discuss.scientific-python.org/ seems like a good venue for that. Here are two separate forum threads for this: - for the big picture: https://discuss.scientific-python.org/t/a-proposed-design-for-supporting-mu… - for the most tricky design decision (user-facing switch & default behavior): https://discuss.scientific-python.org/t/default-dispatching-behavior-for-su… Thanks in advance for any input! Cheers, Ralf

3 4