Hacking for Science, Jonas Beck at the 2024 SBI Hackathon

At the end of March 2024 the mackelab welcomed a group of 30 interdisciplinary researchers to hack on the SBI toolbox.

SBI is a toolbox for [Simulation-based inference (SBI) that was originally developed by the mackelab in 2020, but thanks to the eas of using it, has garnered a large community of users that also adds features and maintains it. SBI is a powerful approach for statistical inference when the underlying likelihood function is intractable or difficult to evaluate. The key idea behind SBI is to use computer simulations of the model to perform inference, rather than relying on analytical expressions for the likelihood. It is used to tune the parameters of powerful and complex simulator models to align them with real-world observations in fields such as particle physics, computational neuroscience, evolutionary biology, climate science, and cosmology.

Thanks to generous support from the Tübingen AI Center, over 30 PhDs, postdocs and professors from across Germany and even parts of Europe could participate on site. Since I had previously worked with and contributed to the toolbox myself, I also took part in the already third installment of the SBI hackathon.

The goal of the hackathon was to improve several aspects of the toolbox, as well as to make it more flexible with respect to recently developed methods. After a brief kickoff meeting, we assigned ourselves to issues and began closing them throughout the day / week. Thanks to the shared office space the university provided us with, the answer to any coding or toolbox related question were just a room a way. This lead to a very lively and collaborative atmosphere where often several people worked on tackling complex tasks. Apart from coding, several participants gave presentations on their, SBI related, research, which provoked long and interesting discussions immediately after and in the following days. Additionally, the hackathon provided a great opportunity to seamlessly socialize and network with colleagues, as well as bouncing interesting ideas of one another. Both scientifically and personally the hackathon was really stimulating. I had many great conversations and was able to gain a much better understanding of the SBI toolbox.It was amazing to be able to connect and collaborate in person with colleagues from all over and I am thrilled to be part of this lovely and inspiring group of people. While the week went by in a flash, it has left a lasting impression on me and I am already looking forward to the next hackathon.

Thanks to the mackelab for hosting me and thanks to the Tübingen AI Center, for enabling this to happen on site. 

For their perspective you can also read their article here.

Share article


Jonas Beck is interested in probabilistic inference in mechanistic models of neuroscience. More specifically, he applies probabilistic numerical methods and simulation-based inference to fit Hodgkin-Huxley models to observed (experimental) data as part of the PIMMS Network Project of the Cluster of Excellence - Machine Learning for Science.


Data Science

Research Group

Neuronal Modeling