2023 Sponsored Project Updates
AstroPy
February
AstroPy released 5.2 (and 5.2.1).
December
astropy v6.0.0 was released on 2023-11-27 by Simon Conseil.
Astropy welcomed C. E. Brasseur to the Astropy Coordination Committee (CoCo)
Astropy welcomed Beryl Kanali as Community Manager.
Astropy welcomed Tom Aldcroft into the new role of Maintainer Mentorship Coordinator.
OpenFHE
November
OpenFHE 1.1 was released
December
Released OpenFHE-python v0.8.2
OpenFHE is now official partners with FHERMA, a platform for Fully Homomorphic Encryption (FHE) challenges
mlpack
February
September
Bandicoot, a C++ GPU linear algebra library that has an API compatible with the Armadillo library (upon which mlpack is built), had its first stable release.
November
Released mlpack 4.2.1, a bugfix release with additional improvements to the reinforcement learning code (part of GSoC projects)
Released ensmallen 2.20.0 optimization library, which adds a new optimizer and has some bugfixes and cleanups
Released bandicoot 1.11.0 GPU linear algebra library, with some additional optimizations for submatrix operations.
Bandicoot was packaged in Homebrew for easy installation on OS X
December
Released mlpack 4.3.0, with new neural network layer support and a number of bugfixes
Released ensmallen 2.21.0, with an important bugfix to optimizer callbacks
The recording of the talk on mlpack and lightweight machine learning in C++ is now on YouTube
Spyder
January
Spyder 5.4.2 was released on Jan. 18, with support for IPython 8, various improvements to the Linux installer, and important bug fixes and performance enhancements.
February
Spyder joined NumFOCUS as a fiscally sponsored project!
April
Spyder 5.4.3 was released on April 4, improving compatibility with IPython 8 and PyZMQ 25, adding support for Mambaforge, Miniforge, and Jupyter-client 8, and fixing a serious bug with our Help pane on Linux not displaying any content.
Spyder held the first NumFOCUS-facilitated Spyder community roundtable on April 11. Many community members attended, and it sparked a lively exchange of ideas about strengths, pain points, and areas for improvement in the project.
June
Two core Spyder developers, CAM Gerlach, and Carlos Cordoba, were accepted to be part of the first class of fellows for the Github Accelerator program, the only project in the PyData ecosystem to receive that support.
July
Spyder 5.4.4 was released on July 17, with new shortcuts to switch between sequential tabs in the Editor (Ctrl/Cmd+8 and Ctrl/Cmd+9) for easier usage on macOS, improves compatibility with PySide2 and fixes several issues with our Windows installer.
November
Spyder 5.5.0 was released on Nov. 8, with improvements to the standalone installers, updater, packaging, and syntax highlighting, as well as a plethora of squashed bugs. For a full list of fixes and enhancements, see the Changelog entry.
PyBaMM
February
PyBaMM joined NumFOCUS as a fiscally sponsored project!
Some of PyBaMM’s core developers launched Ionworks Technologies to provide support for companies using open-source as well as building additional commercial software.
CalVer v23.1 was released, updating our docs to use the PyData sphinx theme and starting a user guide.
April
PyBaMM had its first Hackathon on March 29 at the University of Oxford. There were 27 attendees working on five different projects, resulting in PRs that were close to being merged.
May
Version 23.4 (and a patch 23.4.1) were released, now supporting Python 3.10 and 3.11.
PyBaMM was able to get 3 contributors for GSoC.
November
Published the release candidate for PyBaMM 23.9 (available in PyPI).
PyBaMM launched its new website, which was developed as part of one of the GSoC projects. The new website includes more detailed information about the project and the community and also hosts the docs.
December
The final release for PyBaMM 23.9 was published.
conda
April
Conda was one of thirteen projects selected to participate in the 2023 Google Season of Docs.
conda-libmamba-solver 23.3.0 was released as an update to the fast conda solver plugin.
Conda 23.3.1 & conda-build 3.24.0 was released to improved error messages, 64-bit RISC-V compatibility, faster repodata.json parsing, and experimental support for faster repodata.json downloading
Graykull 2.3.0 was released to build conda package recipes automatically from PyPI and CRAN. This version added support for Poetry version syntax and moved project metadata to pyproject.toml. Grayskull also graduated from an incubator project to a full conda organization project.
June
conda launched its new community website, conda.org!
rattler-build was released as a new rewrite of the core concepts of conda-build in Rust.
conda-lock 2.0.0 was released as a lightweight library for generating fully reproducible lock files for conda environments.
conda 23.5.0 was released with Python 3.11 support and the new conda doctor subcommand plugin for detecting corrupted installs.
conda-build 3.25.0 was released as an update to conda’s package builder.
conda-libmamba-solver 23.5.0 was released to as an update to the fast libmamba solver in conda.
grayskull 2.3.1 was released as an automatic recipe generator for the conda ecosystem.
December
2023 brought many features to conda, including:
conda 23.11.0 closing the 23.x release series
The speedy conda-libmamba-solver as the default dependency solver. This and other improvements like parallel downloads make conda run 4-5x faster.
Support for plugins, enabling customization of conda installation.
Package builders can now include desktop icons across all operating systems.
The conda documentation has been reorganized and updated.
pixi a new open-source conda-compatible package and environment manager led by the folks at prefix.dev. Implemented in Rust, using the recently introduced Rattler library as a foundation. See the announcement for details.
conda-build now supports emscripten-wasm32 and wasi-wasm32 platforms.
conda-lock, mamba, rattler-build, grayskull, quetz, rattler, condax and many other projects had new releases too.
A new recipe format for defining conda-compatible packages has been proposed and accepted. The format is much cleaner, and when implemented will enable fast parsing and building of recipes
MDAnalysis
March
Blosc was accepted as a participating organization in the 2023 GSoC season.
May
The SolvationAnalysis package was published in the Journal of Open Science Software (JOSS).
MDAnalaysis introduced two new GSoC 2023 students: Xu Hong and Egor
June
MDAnalysis welcomed Rocco Meli to the project leadership team as its newest core developer!
MDAnalysis v2.5.0 was released on conda-forge and PyPi.
September
MDAnalysis launched a series of online training workshops.
The inaugural MDAnalysis UGM (User Group Meeting) was held from September 27 to 29 in Lisbon, Portugal. It was attended by 37 participants and featured presentations and open discussions about using and developing MDAnalysis, followed by a hackathon.
November
MDAnalysis was the first featured software project in OSSci's (Open Source Science) new spotlight series.
December
More than 70 people joined live on Oct. 25 for an Intro to MDAnalysis online workshop.
Blosc
February
Python-Blosc2 2.0.0 released, adding support for user-defined filters and codecs. See more at: https://www.blosc.org/posts/python-blosc2-pipeline/
C-Blosc2 2.6.1 released, adding support for macOS universal2 binaries (arm64+x86_64 build).
May
C-Blosc2 2.9.1, with support for dynamic plugins, was released.
Python-Blosc 2.2.1, with the latest C-Blosc2 included, was released, allowing dynamic plugins to be used from Python, too.
For PyTables, a grant was funded from NumFOCUS for leveraging multi-dimensional Blosc2 super-chunks with the HDF5 enhanced direct chunk capabilities.
April
We have been very busy implementing exciting new developments:
C-Blosc2 2.7.1 was released: The Caterva library has been merged and carefully integrated into C-Blosc2 through the new b2nd interface. See more at: https://www.blosc.org/posts/blosc2-ndim-intro
C-Blosc2 2.8.0 was released: A new bytedelta filter was added, including SIMD support for Intel and ARM platforms.
Python-Blosc2 2.1.1 was released: `NDArray` was introduced, an object for handling multidimensional arrays using compression, featuring: data type handling fully compatible with NumPy, and double partitioning
June
BTUNE 1.0.0 (RC2) for Blosc2 was released. BTUNE trains a neural network on your representative datasets to find the best compression parameters for them.
October
PyTables project
PyTables 3.9.1 was release to provide optimized support for hyper-slicing in multidimensional Blosc2 chunks. This was partly funded by NumFOCUS with the grant "PyTables to leverage the HDF5 enhanced direct chunk capabilities".
Blosc project
Released C-Blosc2 2.10.4. This contains several improvements for helping the integration of C-Blosc2 in other projects.
Released Python-Blosc2 2.2.8, providing binary wheels for Python 3.12. We have also upgraded to the latest C-Blosc2 2.10.4.
Released blosc2-openhtj2k 0.1.2, an OpenHTJ2K plugin for Blosc2. This is a dynamic codec plugin for Blosc2 that allows the compression and decompression of images using the High Throughput JPEG 2000 standard. The HT version brings increased performance when compared to the traditional JPEG 2000. For details, check the HTJ2K whitepaper.
Blosc welcomed Ivan Vilata as a new member team.
December
Blosc2-Grok 0.0.1 was tagged - This is the first public release of a Blosc2 plugin for Grok, the World's Leading Open Source JPEG 2000 Codec.
b2h5py 0.2.0 was released - b2h5py is about transparently accelerating the reading of n-dimensional Blosc2 slices for h5py.
PyTables:
PyTables 3.9.2 was released -This fixes the full integration of Blosc2 into PyTables/HDF5, leading to much better performance. See more at: https://www.blosc.org/posts/pytables-b2nd-slicing/
Version 2.8.8 of numexpr was released - A fast numerical array expression evaluator for Python, NumPy, Pandas, PyTables and more
Julia
May
Julia 1.9 was released, the fastest version of Julia to date.
November
Julia version 1.9.4, the fourth patch release in the 1.9 series of releases, was released.
As a patch release, 1.9.4 contains no new features or breaking changes, only bug fixes, documentation improvements, and performance improvements. You can see the list of commits included since 1.9.3 here. We recommend that anyone currently using an earlier 1.9 release to upgrade to 1.9.4.
rOpenSci
May
An update was released where the R-universe node stack now provides data export links, which use webr to convert pkg datasets on-the-fly to JSON (via jsonlite), xlsx (via writexl), csv (via data.table), etc.
QuantEcon
November
QuantEcon launched Quantitative Economics with Google JAX, a new lecture series that introduces JAX and its techniques to solve high-dimensional economic problems. Google JAX is an open-source Python library developed by Google Research that provides a range of tools for fast linear algebra operations, automatic differentiation, and efficient parallel computing. JAX builds upon the popular NumPy library, extending it with additional features such as automated GPU and TPU support, just-in-time compilation, and efficient optimization and estimation. The lectures cover applications such as asset pricing problems, inventory dynamics, and wealth distribution dynamics.
FEniCS
November
DOLFINx v0.7.0 was released with its next-generation problem-solving environment for PDEs
NumPy
February
Mukulika Pahari and Ross Barnowski were appointed as the new NumPy documentation team leads
April
Google Season of Docs 2023
NumPy was accepted as a participating project in this year's GSoD program.
May
Google Season of Docs 2023
Mars Lee was selected as the technical writer for NumPy for the GSoD program. Read Mars’s blog post on Medium
Pamphile Roy (SciPy) and Marco Gorelli (pandas) joined the Contributor Experience Lead Team to support the work on the CZI grant “Advancing an Inclusive Culture in the Scientific Python ecosystem”.
September
NumPy 1.25.0 was released to improve the handling and promotion of dtypes, increase the execution speed, and clarify the documentation.
November
NumPy 1.26.0 was released as a continuation of the 1.25.x series that marks the transition to the Meson build system and provision of support for Cython 3.0.0.
HoloViz
September
HoloViz released a report with the results of their first user survey.
Shifted its user and contributor chats to Discord to enhance community interaction and transparency.
Organized a Panel data app contest with Anaconda.
Panel 1.0 - Released (now at 1.2) with support for Bokeh 3 and much more.
HoloViews 1.17 - Includes twin-axis support and plotting backend customization.
Datashader 0.15 - Improvements for inspection and summary reductions.
GeoViews 1.10 - Adds support for xyzservices.
Param - 2.0 Release is imminent with reactive expressions.
November
Introduced dynamic scale bars in Bokeh, typical of biomedical plots.
Added new advanced annotation tools for data labeling and interpretation.
Added HoloViews support for multi-time series subplots with separate y-axes.
HoloViews 1.17.0 brought the long-awaited twin axes support
hvPlot 0.9.0 added Xarray support to the explorer assistant along with Polars integration.
Datashader 0.15.1 and 0.15.2 were released, with support for inspecting datashaded output.
GeoViews 1.10.1 and Panel 1.2.0, 1.2.1, 1.2.2, and 1.2.3 were released, with support for Bokeh 3.2.
TARDIS
June
TARDIS started mentoring 6 students for GSoC, with 23 mentors from different physics and software backgrounds who volunteered to mentor the projects.
STARDIS became the newest member of the TARDIS family, designed to simulate stellar spectra.
November
TARDIS developed a new core team with roles for science management- a group of people each governing a specific segment of supernovae research.
TARDIS upgraded its environment by updating several major dependencies to the most recent versions within the Python ecosystem.
ITK
February
ITK 5.3.0 was released, including Dask support, extended NumPy support, accelerated performance, new segmentation, shape analysis, and more
November
ITK published a new Sphinx-based documentation portal.
ITK 5.4 Release Candidates became available for testing.
ITK migrated to Web3 infrastructure for community-based development of its testing data.
WorldWide Telescope (WWT)
November
WWT launched their new merchandise on the NUMFocus online shop
WWT launched their custom interactive explorer to zoom around the Orion Nebula.
The WWT developer documentation got a refresh, along with some bug fixes.
Matplotlib
May
Matplotlib announced support for PEP484/mypy compatible type hints.
September
Matplotlib published a release candidate for Matplotlib 3.8.0, including type hints on much of the public API for Matplotlib.
November
Matplotlib has published version 3.8.0, which had improvements to the Mathtext parsing (their internal LaTeX-like equation parser).
SymPy
September
napari released version 0.4.18, which supports textures on surfaces, much faster label painting, better display of vector fields, and many more improvements.
Zarr
Zarr-Python 2.16.0 and 2.16.1 were released
Zarr-Python developers kicked off two working groups, refactoring and benchmarking, to update the package to the latest V3 specification.
ArviZ
May
The ArviZ project expanded to announce its new member: PreliZ, a Python package aimed at helping practitioners choose prior distributions. It is built on top of SciPy and ipywidgets, among other packages, of the Python scientific stack, and includes tools for prior elicitation of 1D distributions, like the roulette method and the maxent (maximum entropy) function and also tools for predictive elicitation.
PyMC
January
PyMC released version 5.0.2 on January 24 as a maintenance and bug fix release.
April
PyMC released version 5.2.0, featuring the availability of Hilbert space approximation Gaussian processes (HSGP) that allows for dramatically faster model fitting with Gaussian processes.
CuPy
January
CuPy v11.5.0 & v12.0.0b3 were released on Jan. 19, including support for the latest CUDA 12 and more cupyx.scipy.interpolate.
April
CuPy v12 was released, bringing more NumPy/SciPy APIs optimized for GPU, including SciPy's interpolation module and ufunc methods like ‘cupy.add.at’.
SciPy
January
SciPy 1.10.0 was released, including a new dedicated datasets submodule (scipy.datasets) to be used instead of scipy.misc for dataset retrieval; several improvements to scipy.interpolate including a new scipy.interpolate.make_smoothing_spline function; scipy.stats has three new distributions, two new hypothesis tests, three new sample statistics, a class for greater control over calculations involving covariance matrices, and many other enhancements.
February
SciyPy received a 2-year grant from CZI to support SciPy with a focus on better serving biomedical applications, adding important new features, performing essential maintenance, and disseminating the work to biomedical researchers and software developers.
April
Updates from SciPy:
SciPy welcomed Jake Bowhay as a new maintainer!
Pamphile Roy joined as a Contributor Experience Lead as part of our ongoing CZI grant to improve contributor onboarding.
SciPy made it possible to submit Jupyter notebook tutorials to the SciPy documentation
MathJax
April
MathJax prepared v4.0-beta for release, updating the v4.0-alpha version with numerous bug fixes and some enhancements, including better line-breaking support in tables, updates to the new fonts, and improvements in the accessibility features.
MathJax acquired 501(c)(3) status, a long-standing goal of the project.
PyTables
February
PyTables 3.8.0 was released, including optimized paths for reading and writing in tables.