Machine Learning on FPGAs for Real-Time Processing for ATLAS Liquid Argon Calorimeter

• Lauri Laatu

The ATLAS experiment at the Large Hadron Collider (LHC) is operated at CERN and measures proton–proton collisions at multi-TeV energies with a repetition frequency of 40 MHz. Within the phase-II upgrade of the LHC, the readout electronics of the liquid-argon (LAr) calorimeters of ATLAS are being prepared for high luminosity operation expecting a pileup of up to 200 simultaneous proton–proton interactions. Moreover, the calorimeter signals of up to 25 subsequent collisions are overlapping, which increases the difficulty of energy reconstruction by the calorimeter detector. Real-time processing of digitized pulses sampled at 40 MHz is performed using field-programmable gate arrays (FPGAs). To cope with the signal pileup, new machine learning approaches based on recurrent neural networks outperform the optimal signal filter currently used, both in assignment of the reconstructed energy to the correct proton bunch crossing and in energy resolution. The improvements concern in particular energies derived from overlapping pulses.

Search for very-high-energy gamma ray and neutrino emission from microquasars with HESS and ANTARES/KM3NeT

• Sebastien Le stum (CPPM)

Microquasars are binary systems with a compact object (black hole, neutron star) and a star. They exhibit periods of intense luminosity known as "outbursts". These phenomena are still poorly understood at high energy. Observations of gamma photons and neutrinos from these sources are necessary. In order to select these periods, multi-wavelenght data are collected. From a list of known microquasars, HESS and ANTARES archival data are analysed to compute upper limits on gamma photons and neutrino fluxes, respectively. Finally, these results are compared to existing models. In order to improve the triggering of new observations, an automation of real-time data analysis in KM3NeT/ORCA is performed and applied to microquasars.

Towards precision cosmology with void lensing

• Renan Isquierdo Boschetti (CPPM)

The next generation of Large-Scale Structure (LSS) surveys will provide an incredible amount of cosmological data, which consists of mapping out almost the entire volume available in the observable Universe. In order to be able to optimize the amount of information one is able to extract from this data, it is necessary to study cosmological analysis using simulations. Two promising probes of LSS information are cosmic voids - underdense regions between filaments and clusters - and weak lensing - the distortion in the shape of background galaxies due to inhomogeneities in LSS. The combination of both is highly sensitive to modifications in the standard model of cosmology, such as modified gravity, non-standard neutrino masses and time-varying dark energy. In this work we show advances in the measurement and interpretation of the signature of weak-lensing signal by voids.