Ruprecht Karls Universität Heidelberg


Master Arbeit

Track Reconstruction with Graph Neural Networks for the ATLAS Trigger System

Our group explores the feasibility of applying modern machine learning algorithms such as Graph NeuralNetworks (GNN) deployed on FPGAs for online track reconstruction within the ATLAS online server farm for the High-Luminosity LHC (HL-HLC) upgrade. GNNs are a powerful class of geometric deep learning methods for modelling spatial dependencies via message passing over graphs. They are well-suited for track reconstruction tasks by learning on an expressive structured graph representation of hit data. A
considerable speed-up over CPU-based execution is possible on FPGAs.
We can offer several topics for theses:

  1. focusing on performance simulation aspects and model optimizations
  2. limiting the input data to e.g. the pixel detector only and studying the performance if the GNN only delivers track seeds or even only hit triplets
  3. focusing on model translation and hardware deployment
  4. studying the robustness of the models with respect to detector deformations

Kontakt: André Schöning, Sebastian Dittmeier

Veröffentlicht am: 2023-03-21
EDV Abteilung