RooFit has seen very active developments in recent years, focusing on performance optimizations, GPU support, automatic differentiation,
fixing bugs, and addressing user requests. Now we think it would be a good time for a workshop with RooFit experts in the experiment communities to discuss what can be done on the experiment side, so that all users can benefit from the new features and improvements in RooFit.
The list of topics we would like to discuss includes:
- New RooFit functionality to speed up your analysis on the CPU or using the GPU
- Automatic differentiation in RooFit
- Serializing models in other formats (e.g. JSON)
- Improved minimization algorithms for RooFit (e.g. work related to Minuit 2)
- RooFits connection to Machine Learning (e.g. for simulation-based inference)
- Evolution of higher-level interfaces for statistical analysis (like RooStats)
- The integration of the newest RooFit features with experiment frameworks
On this occasion, we would also like to hear from the experiments how RooFit is typically used right now, where the performance bottlenecks and usability shortcomings are, and how they think RooFit should evolve in the next few years to serve the experiments best.