Deep Generative Machine Learning for Modeling Hadronic Interactions
Haichen Wang, Professor
Physics
Applications for Spring 2024 are closed for this project.
Reliably simulating detector response to hadrons is crucial for almost all physics programs at the Large Hadron Collider. The core component of such simulation is the modeling of hadronic interactions. Unfortunately, there is no first-principle theory guidance. The current state-of-the-art simulation tool, Geant4, exploits phenomenology-inspired parametric models, each simulating a specific range of hadron energies for some hadron flavor types. These models must be combined to simulate all hadron flavors at all energy ranges. Parameters in each model and the transition region between models must be tuned to match the experimental measurements. Those models may be updated to cope with new measurements.
Role: The project is to use ML techniques to model the hadronic interactions and to unify all parametric models. Specifically, this project involves:
1) learning hadronic interaction simulations in Geant4
2) generating hadronc interactions with Geant4
3) learning generative ML models to simulate these hadronic interactions
4) evaluating the performance of the trained ML models and writing reports
The above steps are foreseen and may be done in order. Discussing the results with researchers at LBNL and actively learning advanced Python programming and ML techniques are important parts of this work; related materials will be made available.
Qualifications: Qualifications: Applicants should be interested in particle physics. Applicants should finish classes related to Data Structure and Programming. Knowing Machine Learning Programming is a strong pluse. Knowing data analysis with Numpy and Pandas is also desired. Majoring in Physics would be great, but it is not a requirement. We encourage EECS students to apply.
Hours: 9-11 hrs
Off-Campus Research Site: Research Site: Our default mode of operation will be virtual, meeting on Zoom weekly and communicating via Slack and email. This research project will be directed by Dr. Xiangyang Ju of Lawrence Berkeley National Lab.
Mathematical and Physical Sciences