Session B01: HEP Detectors and Software

Machine learning and computing tools for LHC Run3 and HL-LHC: challenges and opportunities

 At the dawn of LHC Run 3, machine learning enjoys an enormous success on many fronts of the experimental high energy physics program: event reconstruction, interpretation, background estimation, and many more. 

At the same time, the ambitious high luminosity program of the LHC that will develop over the next two decades poses new computing and physics challenges to the HEP community. 

In this talk I will summarize some applications of machine learning developed to tackle current and future challenges, with a particular focus on innovative analysis techniques and data quality monitoring at the CMS experiment.   

Search for Exotic Higgs Decays at CMS and the CMS High Granularity Calorimeter upgrade for HL-LHC

Although the 125 GeV Higgs scalar displays spin, parity, and fermionic and bosonic couplings consistent with those predicted by the Standard Model (SM), constraints on its branching ratio to invisible or non-SM final states are only at the 20-30% level. Direct searches for Higgs decays to non-SM final states offer further insights into the structure of the Higgs sector, specifically whether it consists of the single doublet of the Standard Model, or multiple doublets as proposed by many theories that extend the Standard Model. In this talk, I will present recent results on searches using data collected by the Compact Muon Solenoid (CMS) detector for Higgs decays to non-SM final states, focusing on decays that proceed via new light Higgs states.

The CMS HGCal is a sampling calorimeter constructed of alternating layers of active material and stainless steel absorber. In the regions of highest expected total ionizing dose, silicon is used as active material, while elsewhere plastic scintillator tiles coupled to silicon photomultipliers (SiPMs) are used. It features unprecedented transverse and longitudinal segmentation for both electromagnetic and hadronic compartments. It will replace the current endcap electromagnetic and hadronic calorimeters in the high luminosity LHC starting in 2027. The latest results for the HGCal software and hardware developments and tests will also be discussed.    

Study of NOvA TestBeam Detector Response for Protons

Name: Bishnu Acharya

Study of NOvA TestBeam Detector Response for Protons 

The NOvA (NuMI Off-Axis ??e Appearance) Experiment is a long-baseline accelerator neutrino experiment in Fermilab. It consists of two functionally identical detectors, a 300 ton Near Detector (ND) at Fermilab, 1 km from the beam source, and a 14kton Far Detector (FD) in Ash River, Minnesota, 810 km’s away from the source, both placed 14 mrad off-axis with respect to the neutrino beam. The primary goals of this experiment are to determine the neutrino mass ordering, perform precision measurements of the neutrino mixing parameters, including investigation of possible CP violation in the lepton sector, through the study of (anti)neutrino flavor oscillations.

Because systematic uncertainties have a strong impact on measurement of the oscillation parameters, the NOvA Test Beam program was established to measure detector response of charged particles ( electrons, protons, pions and kaons) to reduce some of the dominant uncertainties. The NOvA Test Beam Detector was a scaled down 30 ton detector equipped with both FD and ND technologies. NOvA has historically seen Data/MC discrepancy in the proton dE/dx distribution. In this talk, I will discuss how I am exploring this discrepancy using TestBeam protons including some preliminary results.   

Electron Energy Response Measurement in the NOvA Test Beam Detector

The NuMI Off-Axis Electron Neutrino Appearance (NOvA) experiment is a long-baseline accelerator-based neutrino experiment hosted by Fermilab. A 300 ton Near Detector and a 14 kton Far Detector, functionally-identical detectors located 809 km apart and placed 14 mrad off-axis to Fermilab’s NuMI (Neutrino at the Main Injector) neutrino beam, are designed to study and describe neutrino properties through their flavor oscillations. NOvA’s primary physics goals are to determine the neutrino mass ordering, investigate CP violation in the lepton sector, and conduct precision measurements of the neutrino mixing parameters. In order to help further NOvA's physics reach, the NOvA Test Beam program was operated to measure the charged particles produced in the neutrino interactions. The Test Beam Detector, a scaled down 30-ton detector equipped with both Far Detector and Near Detector technologies, was employed to study the electrons, protons, pions, muons, and kaons for an improved understanding of the largest systematic uncertainties that impact NOvA’s analyses. These charged particles were identified and momentum-selected within a range of 0.3 - 2.0 GeV/c by a tertiary beamline at Fermilab and analyzing them is necessary to characterize the detector response and calibration. In this talk, I will discuss the tagging of electrons and positrons with the beamline along with the preliminary results on the energy response obtained from the detector measurements of their electromagnetic activities.   

"Freight Train" production model on the NOvA experiment and NOvA efforts at Argonne Leadership Computing Facility

A large obstacle for any experiment is how to store, move, and process a large amount of data effectively. On the NuMI Off-axis nu_e Appearance (NOvA) experiment this is a rather substantial obstacle. The traditional method of grabbing a file is by categorizing all the files we care about by event-type and then going through each tape and gathering all files of that type. This is not ideal as it leads to multiple readings of the same tape. The Freight Train model of production at NOvA aims to address this inefficiency and will be presented in this talk. Additionally, NOvA has implemented a scheme to run a pre-reconstruction filter (neural network-based) to remove the most obvious cosmic rays from our large recorded cosmic ray sample to reduce processing needs. This step is processed at the Argonne Leadership Computing Facility (ALCF), leveraging the additional GPU and CPU resources. This talk will present NOvA's freight train production model and the pre-reconstruction implementation at ALCF.