Sitemap

March 2025 Project Updates

NumFOCUS
8 min readMar 27, 2025

Sponsored Projects

Bokeh

Bokeh just released version 3.7. Highlights were described in a blog post and their brand new Bluesky account.

Blosc

Spyder

  • Spyder 6.0.5 was released, with an important fix to ensure Spyder works on restricted Windows environments, patches to the new run architecture and an option to not include project files when searching in the Switcher
  • The first alpha was released for Spyder 6.1, the next feature version. It comes with many significant improvements, including the long-awaited multi-cursor support for the Editor. It also has better support for PyQt6 and a significantly enhanced PYTHONPATH manager with the option of whether to prepend or append paths to sys.path.

CuPy

CuPy v13.4.0 is now available! This release includes:

  • NVIDIA CUDA 12.8 & Blackwell architecture support
  • AMD ROCm 6 compatibility
  • Python 3.13 binary packages
  • DLPack v1 support

Check out the release notes for details: https://github.com/cupy/cupy/releases/tag/v13.4.0

rOpenSci

rOpenSci Champions Program 2025 In Spanish: Apply before April 30th!

We have great news: The call for applications to be part of the new cohort of our 2025 Program is now open!

And for the first time it will be in Spanish!

Our program seeks to identify, recognize and reward people who are leaders in an open science community, research software engineering and the R programming community.

This year’s program is focused on people from Latin America and for the first time will be conducted entirely in Spanish.

The main goal is to foster sustainable research software as a pillar of Open Science in Latin America through capacity and community building.

Find out more in our call for applications, open until Wednesday, April 30, 2025.

Better documentation for R-universe!

Thanks to funding by Google Season of Docs, we were able to start a new comprehensive documentation project for all users and developers of R-universe.

We established a central place where we collect the various sources of information and describe examples and use cases, using popular authoring tools to support collective maintenance.

Read more in our blog post, and read the documentation website.

Events

### Coworking

Read all about coworking!

And remember, you can always cowork independently on work related to R, work on packages that tend to be neglected, or work on what ever you need to get done!

MDAnalysis

Affiliated Project Updates

FESTIM

Alpha release of FESTIM 2.0!

  • Complete rewrite of FESTIM using FEniCSx (instead of legacy-FEniCS)
  • Multi-speices transport
  • Advanced interface condition handling
  • Many more exciting features!

New conda package for FESTIM facilitating installation for users

https://festim.readthedocs.io/en/latest/

Skforecast

🚀 Skforecast 0.15.0 has just been released! 🎉

This release is all about enhancing probabilistic forecasting in Skforecast, giving you more powerful tools to assess and manage uncertainty in your forecasts.

🎯 Conformal framework for Probabilistic forecasting

You can now generate prediction intervals using the conformal prediction split method, offering a robust approach to quantify the uncertainty of your forecasts.

📊 Binned residuals extended to more Forecasters

Enjoy adaptive prediction intervals with residuals conditioned on predicted values, now available in more Forecasters

🔍 Conformal calibration of prediction intervals

Introducing our new calibrator, which levergaes conformal prediction to adjust prediction intervals generated by other techniques, such as quantile regression or bootstrapped residuals, to achieve the desired coverage probability.

With this release, we’re taking the first step toward expanding the probabilistic capabilities of Skforecast. Our goal is to provide even more robust and versatile tools for forecasting uncertainty — stay tuned for more to come!

🔗 Dive into the full details: https://skforecast.org/0.15.0/releases/releases

🔗 Skforecast docs: https://skforecast.org/

👤 Joaquin Amat Rodrigo

👤 Javier Escobar Ortiz

Happy forecasting! 📈

SPHinXsys

The following are the major updates of SPHinXsys project.

  1. We have released the newest version of the library version 1.1 (https://github.com/Xiangyu-Hu/SPHinXsys/releases/tag/v1.1-sycl) in which heterogeneous parallelism, based on SYCL and DPC++, is introduced with numerical examples on thermo-fluid simulations.
  2. New physics, including generalized non-hourglass formulation for ULSPH solid dynamics (https://github.com/Xiangyu-Hu/SPHinXsys/pull/651), multi-resolution solid dynamics (https://github.com/Xiangyu-Hu/SPHinXsys/pull/634), standard k-ε turbulence model (https://github.com/Xiangyu-Hu/SPHinXsys/pull/724), have been added to the library.
  3. A brand new version SPHinXsys Blog website (xiangyu-hu.github.io/SPHinXsys/) is constructed by using Jekyll, a static content generator.
  4. The Github repository of SPHinXsys has now obtained 355 stars and 252 public forks.

Crystal

We released version 1.15. Two highlights from the latest version:

  • New event loop implementation for UNIX based systems. The main benefits of this new implementation are: improved performance, one less external dependency (libevent), and more control over Crystal’s runtime which is a foundation for other upcoming enhancements, like the new execution context runtime that’s at the center of the Crystal’s revamped multi-threading support. Read more about these aspects in the event loop RFC and the execution context RFC.
  • Windows support keeps growing: you can now opt to use MinGW or MSYS2 as alternative compiler toolchains, and since 1.15 you can use Crystal on Windows on ARM processors (tier 3 support for now).

SBI

I would like to share info about an important upcoming event — a new instance of the SBI Hackathon.
This event will take place next week March 17–21, in Tübingen. We will have 27 participants from across Europe from a diverse set of research backgrounds all coming to Tübingen to work on improving the SBI package. Additionally, we have 15 participants joining remotely from all across Europe, from Japan and from the US. This is a great progress in making SBI a community driven project.

Additionally, there will be a new SBI release coming this week in preparation of the hackathon. The release will make available several important new features and improvements and will bring everyone joining the hackathon on the same page.

Lastly, SBI is participating in the Google Summer of Code under the NumFOCUS umbrella and we received a lot of interest for our proposed projects already and have some promising applications.

Optuna

We released Optuna v4.2 with the following updates:

  • OptunaStorageProxyService: Storage caching system for improved throughput in heavily parallelized optimization
  • Constrained optimization support for GPSampler
  • c-TPE: A new constrained optimization algorithm for TPE
  • Benchmark problems in OptunaHub!We currently have bbob and its variants, dtlz, wfg, zdt, nasbench201, hpobench and hpolib.
  • SMACSampler added in OptunaHub
  • It uses SMAC3 developed by AutoML.org as the backend, which implements a random forest-based Bayesian optimization.

GeomScale

🔬 GeomScale is an open-source research and development project focused on cutting-edge algorithms at the intersection of data science, optimization, geometry, and statistical computing.
📰 Latest news
📢 Google Summer of Code 2025

We are thrilled to share that GeomScale has been accepted as a mentoring organization for the 2025 Google Summer of Code 🎉.

This year we propose several topics for software development internships with a strong algorithmic and mathematical flavor.

We are searching for contributors with strong programming skills in at least one of the following languages: C++, Python, R and solid background in at least one of the following topics: (randomized) algorithms, linear algebra, computational geometry, statistics, optimization.

Stipends are paid by Google and interns can work from anywhere in the world.

There are three types of projects:

  • small ~90 hours
  • medium ~175 hours
  • large ~350 hours

Moreover, everybody can apply, no need to be enrolled as a student.

Proposed topics are available here: https://lnkd.in/dFjPduFM

If interested check out our project page and get in touch with mentors as soon as possible!

pvlib

On March 19, we released version 0.12.0 based on contributions from 18 people! Among many small bug fixes and improvements, the package distribution files have been made much smaller by excluding the tests and test files from the wheel, resulting in faster and more compact installations. This release also removes several functions that were previously deprecated.

More details about this release (and others) are found here: https://pvlib-python.readthedocs.io/en/stable/whatsnew.html

We’re also happy that NumFOCUS has been selected as one of the 185 mentoring organizations for Google Summer of Code (GSoC) 2025, and look forward to having some keen young individuals join us in making pvlib even better — hopefully not just for one summer, but also beyond! We have three mentors on standby and four projects suggestions covering different skill levels and time commitments, but we’re open to other good ideas too.

Our GSoC 2025 details are found here: https://github.com/pvlib/pvlib-python/wiki/GSoC-2025-Projects

Mesa

Mesa is proud to announced it was selected for Google Summer of Code (GSoC) for the second year!

We have several great ideas that will further enhance Mesa, but are excited for new ideas we did not think of.

See the whole list on our GSoC 2025 Wiki Page

However, our top three ideas are:

Front End Upgrade

Summary: Mesa recently transitioned to a new Solara-based visualization system that enables interactive, browser-based model exploration. While the core functionality is in place, there are several opportunities to enhance its robustness, performance, and user experience. This project aims to stabilize and extend Mesa’s visualization capabilities, making them more powerful and user-friendly.

Mesa-LLM

Summary: This project aims to integrate large language models (LLMs) as decision-making agents into the Mesa agent-based modeling (ABM) framework. This project will enable more sophisticated, language-driven agent behaviors, allowing researchers to model scenarios involving communication, negotiation, and decision-making influenced by natural language.

Mesa-Frames Upgrade

Summary: As ABMs are simulations and often have phase transitions (periods of rapid change to new stable states), being able to go back in time and replay key results would be a great addition to Mesa. Critically, no computation would be needed as the results are stored.

GSoC proposals open March 24th and closes April 8th and based on the discussions so far we are looking forward to some exciting proposals.

Join us on Matrix

Join us on GitHub

Thank you!

--

--

NumFOCUS
NumFOCUS

Written by NumFOCUS

Our open source scientific software projects are changing the world. Learn more on our website: https://numfocus.org

No responses yet