Quantum Graph Neural Networks for High Energy Physics Analysis at the LHC

The Large Hadron Collider (LHC) built by CERN is the world’s largest and the most powerful particle accelerator, which generates about 1 billion parti- cle collisions per second. Jet tagging, which is the process of determining the elementary particle that produced a jet, is of key interest at CERN and to particle physics, in general, to distinguish useful signals from background pro- cesses and other less interesting signals. Several machine learning approaches have been employed to handle the vast amounts of data generated by LHC. Due to the sparse, irregular and heterogeneous nature of the data produced at the LHC, Graph Neural Networks (GNNs) are well-suited and widely studied for jet tagging. In fact, GNNs have also produced state-of-the-art AUC for jet tagging. Since quantum computing seems to offer a promising solution to efficiently handle large data with inherent parallelism, the goal of this project is to utilize quantum capabilities to improve GNNs for jet tagging. We shall focus on both hybrid quantum-classical and fully quantum approaches and carry out a detailed analysis to understand the pros and cons of both methods.