Too much data? Too many cores? TTree analysis is tedious? Come and enjoy an intro plus discussion of RDataFrame, ROOT's current-generation way of writing super-efficient analyses! It moves all the intricacies into ROOT's responsibility, leaving you simply with a declarative formulation of the filters and computations that make up your analysis.
The objective of this BOF is to help defining strategies to further improve ROOT's components dedicated to data analysis, with emphasis on parallel declarative analysis provided with RDataFrame in the context of forthcoming new accelerators, HENP detectors and existing detector upgrades.
The session can be roughly organised in two steps:
1) A summary is given of the declarative analysis tools provided by ROOT and their main functionality, their interplay with data IO, parallelism and planned future evolution of the Framework (new histograms, new RForest IO backend, new graphics).
2) The ROOT Team engages to gather for feedback about the status of the analysis tools offered by ROOT for parallelised declarative data analysis. Possible synergies are explored of RDataFrame with existing and future experiments' data models. The spotlight is on ATLAS' xAOD, CMS's NanoAOD as well as Alice and LHCb Run3 analysis format and workflows. Potential demonstrators and proof of concepts developed by the ROOT team in collaboration with experiments are identified.
The areas that the discussion can also cover are interactive data analysis running on future architectures, distributed analysis and complements to PROOF (scalability, existing and future technologies, requirements for data IO, error handling and debugging), programming models and languages.
Sign up at this Doodle poll here - limited places!