HSF Working Groups Mandate

v1.0, 2018-10-11

The mandate for the new HSF Working Groups was widely publicised to the HEP software and computing community, developed with input from experiments and others and was agreed on in the open HSF meeting of 2018-10-11.


The mandates are based around the CWP, A Roadmap for HEP Software and Computing R&D for the 2020s and the plans laid out therein.

Three convenors will be appointed for each group. Nominations for convenorships are open for HEP experiments and members of the HSF community to make. To properly reflect the full diversity of the community, the HSF Coordinators will propose convenors from the nominated candidates, which will then be made public and open for comment before being finalised.

Appointments as convenors are for one year, renewable.

Group Activities

The specific activities of each group are based on the HSF Roadmap and should ensure that:

To help build a community for collaboration each group should:

The group convenors will organise sessions for their activity areas and possible joint sessions as part of annual HSF workshops.

Areas of Study and Interest

The HSF Roadmap is the primary document laying out the working domains of each group, but a short summary is provided for reference below.

Detector Simulation

The Detector Simulation Working Group will consider approaches to making detector simulation faster and more accurate for HEP experiments. Simulation that produces a traditional digitised detector output, as well as more radical approaches to generate reconstruction or analysis level outputs directly are within the remit of the group. Improvements to physics models, faster particle transport options and using machine learning and other approximate techniques can be considered. The adaption of simulation techniques to modern computing hardware (covering, e.g., multithreading, vectorisation and different accelerator technologies) are an important area of work.

Reconstruction and Software Triggers

The Reconstruction and Software Triggers Working Group will consider approaches and solutions to common challenges across HEP in the area of event reconstruction and software triggering (e.g., algorithms and data structures designed for online and offline processing of raw detector data). The working group targets challenges identified during the CWP process as well as new ones arising as the state of R&D advances. This forum should foster collaboration on design and implementation challenges, the adoption of common approaches, and raise awareness of existing solutions known to the community. Topics of interest include the evaluation of new foundational libraries; techniques for track and calorimetric-object reconstruction and identification; pattern recognition and clustering; new approaches, including applications of machine learning techniques and real-time analysis techniques; the effective use of modern computing hardware through threading, vectorisation and use of accelerator technologies; and the application of profiling and quality assurance toolkits.

Data Analysis

The Data Analysis Working Group considers developments and approaches that will make the analysis of HEP data towards final physics results faster and more scalable in the future. One new approach to be considered is the development of expressive functional interfaces for analysis. The meshing of HEP specific tools with those found in the larger data science community is another area of interest, where database approaches or formats optimised for high-speed processing may be profitable. The development of test beds for high performance analysis clusters, that may speed-up turn-around and efficiency is a major area of R&D. The group should help coordinate activities and communication here between experiments and IT experts. The integration of such systems into the end-to-end data flow, with caching and distribution policies, are important. Optimising the actual content of analysis formats is a boundary condition to discuss with physics groups, but can bring major gains.