Choose timezone
Your profile timezone:
Special EPE Seminar: Sicong Lu
→
US/Pacific
42/R-032 (CERN)
Description
Title: Search for Higgsinos using Neural Networks and ASIC Verification
Abstract: The first part of this talk will present a search for Higgsinos with compressed-mass spectra using neural network (NN) classifiers. Higgsinos near the Electroweak scale remains an attractive natural solution to the hierarchy problem while also providing a dark matter candidate in the form of the lightest neutralino. The decays of pair-produced Higgsino-like chargino/neutralino can be identified through the unique signatures of large missing transverse energy and a slightly displaced soft pion track. Our approach uses a two-level neural network model for signal region definition, a 3-fold cross-validation to avoid overfitting, and a semi-data-driven weight correction to address Monte Carlo (MC) mis-modeling. The corrected MC prediction shows good agreement with Data in the two-level NN-based orthogonal validation regions. The expected exclusion limits suggest that the NN analysis improves the sensitivity over the traditional cut-based analysis, by an amount equivalent to a substantial increase in the dataset size, ranging from 30% to 130% depending on the chargino and neutralino masses.
The second part will shift focus to the Autonomous Monitor And Control (AMAC) ASIC, designed by the Penn ATLAS team for the ITk-Strip Detector in the upcoming HL-LHC upgrade. Given its crucial role in real-time detector monitoring and as the primary safeguard for damage control in the detector module, its reliability in a high-radiation environment is of utmost importance. I will discuss quantitative grading during AMAC prototype probing, as well as simulation design verification for the AMAC production against Single Event Effects (SEE). The comprehensive evaluation process has allowed us to identify and rectify critical design vulnerabilities, contributing to the production of high-yield, robust ASICs critical for the success of the upgraded ATLAS detector.
Bio: Sicong Lu is a Ph.D. Candidate in the Department of Physics and Astronomy at the University of Pennsylvania, working with the Penn ATLAS Group. Sicong completed his Bachelor of Arts in Physics and Mathematics at the University of California, Berkeley, where he contributed to the ATLAS Detector's sensitivity improvement in ttH→γγ analysis using boosted decision trees. At Penn, his research focuses on the application of neural networks to search for Higgsinos in compressed mass spectra at the LHC. He also contributed to the testing and simulation of AMAC ASICs in the ITk modules of the ATLAS Upgrade for HL-LHC.
