Nonlinear Theory for Plasma Wakefield Produced by Elliptical Beams via Data-Driven Physical Modeling
Abstract/Description/Artist Statement
Plasma-based acceleration is widely regarded as a highly promising candidate technology for next-generation linear colliders and light sources. In the blowout regime of plasma wakefield acceleration, a plasma wave wake is excited by an intense particle beam. However, there has been no theory that predicts the asymmetric nonlinear wakefield driven by elliptical beam drivers. In this work, we focus on developing a nonlinear theory for plasma wakefield produced by elliptical electron beams via data-driven physical modeling. Machine learning has been applied in predicting physical quantities at substantially reduced computational cost compared to conventional simulations, while maintaining a high level of predictive accuracy. In particular, emerging physics-informed machine learning methods enable the direct incorporation of physical laws into learning frameworks, yielding interpretable and highly accurate surrogate models for complex physical systems. Here, we employ physics-informed machine learning models to approximate analytically inaccessible components in the blowout theory, enabling the discovery of the equations that govern the wakefield. We present encouraging preliminary results obtained using implementations based on PyTorch and Matlab.
Faculty Advisor/Mentor
Qianqian Su
Faculty Advisor/Mentor Email
qsu@odu.edu
Faculty Advisor/Mentor Department
Electrical & Computer Engineering
College/School Affiliation
School of Data Science
Student Level Group
Graduate/Professional
Presentation Type
Poster
Nonlinear Theory for Plasma Wakefield Produced by Elliptical Beams via Data-Driven Physical Modeling
Plasma-based acceleration is widely regarded as a highly promising candidate technology for next-generation linear colliders and light sources. In the blowout regime of plasma wakefield acceleration, a plasma wave wake is excited by an intense particle beam. However, there has been no theory that predicts the asymmetric nonlinear wakefield driven by elliptical beam drivers. In this work, we focus on developing a nonlinear theory for plasma wakefield produced by elliptical electron beams via data-driven physical modeling. Machine learning has been applied in predicting physical quantities at substantially reduced computational cost compared to conventional simulations, while maintaining a high level of predictive accuracy. In particular, emerging physics-informed machine learning methods enable the direct incorporation of physical laws into learning frameworks, yielding interpretable and highly accurate surrogate models for complex physical systems. Here, we employ physics-informed machine learning models to approximate analytically inaccessible components in the blowout theory, enabling the discovery of the equations that govern the wakefield. We present encouraging preliminary results obtained using implementations based on PyTorch and Matlab.