Periodic Reporting for period 2 - INSIGHTS (International Training Network for Statistics in High Energy Physics and Society)
Reporting period: 2019-09-01 to 2022-02-28
Statistical data analysis is a central tool in all types of scientific research, especially in High Energy Physics (HEP). To carry out an HEP analysis capable of discovering new phenomena one needs multivariate methods that exploit recent innovations from Machine Learning together with accurate parametric models and advanced computational methods. Tremendous progress has been made in the development and application of novel statistical methods in HEP in recent years, and these have proved decisive in many important discoveries. The aim of INSIGHTS has been to extend this progress and thus substantially increase the potential of facilities like the LHC to achieve their scientific goals.
The development and application of modern statistical methods cannot be decoupled from software tools. Through training events and secondments with industrial partners, INSIGHTS provided training in modern software engineering practice. Much of the resulting software has made publicly available and as a result is of high value to HEP and other research and industrial fields.
Secondments with INSIGHTS partners were the primary mechanism through which ESRs could exchange ideas between their primary research field and other areas. These included placements of ESRs with partners in fields such as climate science (CICERO, Oslo), volcanology (INGV, Naples), 3D modelling (DigitalComoedia), business and finance (KPMG, Amsterdam and FISCAL, Reading), and high-performance computing (C2PAP, Munich).
The ESRs have developed novel statistical methods, implemented these in software and applied them to research projects. All the methods were implemented in software, much of which is in use by the research labs or companies in which the work was done and/or has been made publicly available. Dissemination of the results has been through peer-reviewed journals, archive and conference contributions, as well as internal notes of experimental collaborations (e.g. ATLAS, CMS). The 13 ESRs made significant contributions to 70 papers and technical reports.
Of the 13 ESRs, 12 were enrolled in PhD programmes. As of August 2022, 5 PhDs have been awarded, 4 are expected by the end of the year, and 3 more are continuing towards completion in PhD programmes designed to go beyond the end of INSIGHTS.
ESRs have taken part in numerous outreach activities including the INSIGHTS blog and social media, guiding visitors at Particle Physics Labs (CERN, MPI, Nikhef), and participating in science festivals and in events with schools and the broader public.
• L. Brenner, P. Verschuuren, G. Cowan, et al., Comparison of unfolding methods using RooFitUnfold, International Journal of Modern Physics A, Vol. 35, No. 24, 2050145 (2020); e-print arXiv:1910.14654
• Tommaso Dorigo, Hevjing Yarar et al., RanBox: Anomaly Detection in the Copula Space, presentation at the 10th International Conference on New Frontiers in Physics (ICNFP 2021).
• Lukas Layer et al., “Clustering of experimental seismo-acoustic events using Self-Organizing Map (SOM)” Frontiers in Earth Science.
• A. Golovatiuk, G. De Lellis, A. Ustyuzhanin, Deep learning for Directional Dark Matter search, Journal of Physics: Conference Series, vol. 1525, 012108 (2020).
• V. Hafych, P. Eller, O. Schulz, A. Caldwell, Parallelizing MCMC Sampling via Space Partitioning, Statistics & Computing 32,56 (2022); e-print arXiv:2008.03098
• ATLAS Collaboration (R. Balasubramanian), ATLAS-CONF-2021-053, Combined measurements of Higgs boson, production and decay using up to 139 fb−1 of proton–proton collision data at √s =13 TeV collected with the ATLAS experiment (2021).
• N. Simpson et al., Neos-Neural End-to-End Optimised Summary Statistics, to appear in ACAT 2021 proceedings, IOP Conference series; arXiv:2203.05570 [physics.data-an] (2022).
• Victor Ananyev, Alexander Lincoln Read, Approximating the mode of the non-central chi-squared distribution, arXiv:2106.12267 [math.CA] (2021)
• Sitong An, Lorenzo Moneta, C++ Code Generation for Fast Inference of Deep Learning Models in ROOT/TMVA, contribution to ICHEP 2021, EPJ Web Conf., 251 (2021) 03040.
• ATLAS Collaboration (Serena Palazzo), Measurements of top-quark pair single- and double-differential cross-sections in the all-hadronic channel in pp collisions at √s = 13 TeV using the ATLAS detector, Journal of High Energy Physics volume 2021, 33 (2021)
The impact of the INSIGHTS ITN is most directly felt by the ESRs themselves, who begun working as physicists and data scientists in universities (Naples, Verona), and companies (Mercedes-Benz, logically.ai Hoverture, PlantingSpace). Those who are currently finishing their PhDs will soon follow along similar paths.
Most training events of the INSIGHTS ITN were open to a broad audience, attracting several tens to more than 100 participants, and in this way the impact of INSIGHTS extended far beyond the network. Training events supported by INSIGHTS continue beyond the end of the programme, including the Pan-European Advanced School of Statistics created by INSIGHTS partner DESY. INSIGHTS partners from fields such as climate science, volcanology and finance hosted secondments that gave ESRs training and experience in industry and led to mutually beneficial knowledge transfer.
Socio-economic impact of INSIGHTS also includes outreach through social media, science festivals and events with school groups that have helped raise the aspirations of a younger generation to enter into scientific fields.