Authors:

Anna G. Klupshas

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

Silicon-based sensors are crucial in modern particle physics experiments, enabling precise tracking of charged particles in high- energy collisions. As particle accelerators like the Large Hadron Collider (LHC) at CERN advance into higher luminosity regimes, the development, testing, and optimization of such sensors become increasingly important for accurate data collection and new physics discoveries. My research focuses on two complementary efforts to improve silicon pixel detector performance for next-generation experiments. First, we investigate regression optimization algorithms intended for deployment on SmartPixel sensor hardware. This machine learning approach is designed to predict spatial and angular variables of particle tracks from charge deposition data. Using Mixture Density Networks (MDNs) trained on large simulated datasets, the model learns statistical correlations between deposited charge cluster patterns and track parameters. Ongoing efforts involve enhancing prediction accuracy through the introduction of regularizers in the loss function, dynamic scheduling of hyperparameters, and Pareto front analyses to balance trade-offs between physical and statistical performance. In parallel, we contribute to the quality control testing of MaPSA (Macro Pixel Sub-Assemblies) sensors developed for the High-Luminosity LHC upgrade. Using a semi-automated probe station and a Linux-based testing environment, we perform critical validation procedures, including leakage current, pixel response, and masking tests to evaluate sensor performance. Each sensor is assigned a quality grade based on these results, determining its suitability for deployment. Keywords: Particle Physics; Machine Learning; Detection Systems; Quality Control

Source:

Purdue University / 2025

Topics:

No topics listed

Co-authors:

Anna G. Klupshas