Mailman 3 February 2022 - SciPy-Dev

optimize.linprog: make HiGHS the default method and deprecate old methods
by Matt Haberland Feb. 2, 2022

Feb. 2, 2022

Hi team, The HiGHS methods have been part of `scipy.optimize.linprog` since 1.6.0, and they have demonstrated that they are (much) faster and more robust than `method='simplex'`, `method='revised simplex'`, and `method='interior-point'`. I think it's time to deprecate the old methods and make `method='highs'` the default. Please join the discussion about this proposed change in gh-15512 <https://github.com/scipy/scipy/pull/15512> . Thanks! Matt -- Matt Haberland Assistant Professor BioResource and Agricultural Engineering 08A-3K, Cal Poly

1 0

Contributor experience and community meetings
by Melissa Mendonça Feb. 2, 2022

Feb. 2, 2022

Hello, all! As some of you may know, recently we have been awarded a joint grant for NumPy, SciPy, matplotlib and pandas, aiming to improve contributor experience and hopefully explore and execute actions that help sustain and improve our communities. You can read the full grant proposal here [1]. As part of this grant, I will be dedicating some time to SciPy. Currently, I am an active contributor to NumPy but have only started with tiny contributions to SciPy. One action that we think might be beneficial to the projects is setting up community meetings, in this case picking up on the work Pamphile Roy was doing last year. Of course, the maintainers have the final word here, but I'd like to offer help with organizing, archiving notes, and whatever else you all feel might be helpful :) For this, we must first decide on a time and date. If you can, please answer this poll with your preferred times: https://whenisgood.net/mfd5qhz We can re-use the hackmd that Pamphile set up for the notes: https://hackmd.io/pyhudrC5TgSwdJtpPldHgQ Feel free to add to this hackmd anything you think might be worth discussing to the agenda. Depending on the poll results we can figure out frequency, but initially I would be happy to organize this every two weeks, alternating time zones if we feel this would include more people. Please let me know if there are any questions, concerns or suggestions - I am looking forward to getting more involved and helping in any way I can. Cheers, Melissa [1] https://figshare.com/articles/online_resource/Advancing_an_inclusive_cultur…

3 5

PR: add fisher and nichols methods to combine_pvalues in SciPy.stats
by Fabian Kloosterman Feb. 2, 2022

Feb. 2, 2022

Hi, https://github.com/scipy/scipy/pull/15503 is a PR that adds two new methods for combining p-values from multiple indepedent tests. See also issue https://github.com/scipy/scipy/issues/15481. I have added a test for both methods as well as citations. Thanks for taking a look! -Fabian

1 0

resolving trouble updating PROPACK submodule
by Ralf Gommers Feb. 1, 2022

Feb. 1, 2022

Hi all, It looks like the combination of renaming `master` to `main` in the PROPACK repo, adding a commit there, and changing some config in scipy/.gitmodules is making it quite hard to update that submodule. If you have trouble, try this: git submodule deinit -f . rm -rf scipy/sparse/linalg/_propack/PROPACK/rm -rf .git/modules/scipy/sparse/linalg/_propack/PROPACK/ git submodule update --init --recursive Cheers, Ralf

1 0