Agni Bethani, Christophe Delaere, Luigi Favaro, Andrea Giammanco, Vincent Lemaitre, Fabio Maltoni

With the 2012 discovery of the Higgs boson at the Large Hadron Collider, LHC, the Standard Model of particle physics has been completed, emerging as a most successful description of matter at the smallest distance scales. But as is always the case, the observation of this particle has also heralded the dawn of a new era in the field: particle physics is now turning to the mysteries posed by the presence of dark matter in the universe, as well as the very existence of the Higgs. The upcoming run of the LHC at 13 TeV will probe possible answers to both issues, providing detailed measurements of the properties of the Higgs and extending significantly the sensitivity to new phenomena.

Since the LHC is the only accelerator currently exploring the energy frontier, it is imperative that the analyses of the collected data use the most powerful possible techniques. In recent years several analyses have utilized multi-variate analysis techniques, obtaining higher sensitivity; yet there is ample room for further improvement. With our program we will import and specialize the most powerful advanced statistical learning techniques to data analyses at the LHC, with the objective of maximizing the chance of new physics discoveries.

We have been part of AMVA4NewPhysics, a network of European institutions whose goal is to foster the development and exploitation of Advanced Multi-Variate Analysis for New Physics searches. The network offered between 2015 and 2019 extensive training in both physics and advanced analysis techniques to graduate students, focusing on providing them with the know-how and the experience to boost their career prospects in and outside academia. The network develops ties with non-academic partners for the creation of interdisciplinary software tools, allowing a successful knowledge transfer in both directions. The network studies innovative techniques and identifies their suitability to problems encountered in searches for new physics at the LHC and detailed studies of the Higgs boson sector.

External collaborators: University of Oxford, INFN, University of Padova, Université Blaise Pascal, LIP, IASA, CERN, UCI, EPFL, B12 Consulting, SDG Consulting, Yandex, MathWorks.

Jérôme de Favereau, Christophe Delaere, Pavel Demin, Luigi Favaro, Andrea Giammanco, Vincent Lemaitre

Observability of new phenomenological models in High Energy experiments is delicate to evaluate, due to the complexity of the related detectors, DAQ chain and software. Delphes is a new framework for fast simulation of a general purpose experiment. The simulation includes a tracking system, a magnetic field, calorimetry and a muon system, and possible very forward detectors arranged along the beamline. The framework is interfaced to standard file format from event generators and outputs observable analysis data objects. The simulation takes into account the detector resolutions, usual reconstruction algorithms for complex objects (FastJet) and a simplified trigger emulation. Detection of very forward scattered particles relies on the transport in beamlines with the Hector software.

Luigi Favaro, Andrea Giammanco, Maxime Lagrange, Zahraa Zaher

We are among the founders of MODE (Machine-learning Optimized Design of Experiments, https://mode-collaboration.github.io/), a multi-disciplinary consortium of European and American physicists and computer scientists who target the use of differentiable programming in design optimization of detectors for particle physics applications, extending from fundamental research at accelerators, in space, and in nuclear physics and neutrino facilities, to industrial applications employing the technology of radiation detection.
We also participate to the very closely related activities of the "Codesign" work package in the EUCAIF network.
We aim to develop a modular, customizable, and scalable, fully differentiable pipeline for the end-to-end optimization of articulated objective functions that model in full the true goals of experimental particle physics endeavours, to ensure optimal detector performance, analysis potential, and cost-effectiveness.
The main goal of our activities is to develop an architecture that can be adapted to the above use cases but will also be customizable to any other experimental endeavour employing particle detection at its core. We welcome suggestions, as well as interest in joining our effort, by researchers focusing on use cases for which this technology can be of benefit.

External collaborators: See updated list here: https://mode-collaboration.github.io/ For EUCAIF, see: https://eucaif.org/.