IRIS Digital Assets - FAST-HEP: Faster Analysis Software Tools for HEP
Lead Research Organisation:
University of Bristol
Department Name: Physics
Abstract
This grant has been awarded to complete a set scope of work for IRIS.
FAST-HEP (github.com/fast-hep), an open source volunteer effort started in May 2017, is a UK-led (University of Bristol & RAL) effort to tackle data analysis challenges posed by the expected large data volumes from upcoming experiments such as HL-LHC, DUNE and SKA by first exploring existing python-based data analysis tools from the wider data science community (e.g. numpy, numba, matplotlib) as well as the HEP community (awkward-array, uproot, mplhep, pyHF), and then creating high-level abstractions and glue to use these tools within particle physics research and related fields for physics analysis.
This project will focus on upgrading FAST-HEP's underlying packages and integrate new tools (e.g. boost-histogram, mplhep, pyHF) resulting in significant speed improvements. Together, these changes will make the FAST-HEP tools more useful to a wider range of analyses such as CMS L1 trigger development, DUNE & LZ data analysis, and increase compatibility with astrophysics workflows. As part of this upgrade effort we will strengthen our links to the stakeholders of these analyses by supporting their use of FAST-HEP tools. This will allow us to benchmark the improvements on a wide range of workloads.
FAST-HEP (github.com/fast-hep), an open source volunteer effort started in May 2017, is a UK-led (University of Bristol & RAL) effort to tackle data analysis challenges posed by the expected large data volumes from upcoming experiments such as HL-LHC, DUNE and SKA by first exploring existing python-based data analysis tools from the wider data science community (e.g. numpy, numba, matplotlib) as well as the HEP community (awkward-array, uproot, mplhep, pyHF), and then creating high-level abstractions and glue to use these tools within particle physics research and related fields for physics analysis.
This project will focus on upgrading FAST-HEP's underlying packages and integrate new tools (e.g. boost-histogram, mplhep, pyHF) resulting in significant speed improvements. Together, these changes will make the FAST-HEP tools more useful to a wider range of analyses such as CMS L1 trigger development, DUNE & LZ data analysis, and increase compatibility with astrophysics workflows. As part of this upgrade effort we will strengthen our links to the stakeholders of these analyses by supporting their use of FAST-HEP tools. This will allow us to benchmark the improvements on a wide range of workloads.
