SciML-Driven Capillary Optimization for Emittance Preservation in Active Plasma Lenses

Laser wakefield acceleration (LWFA) allows for the intense acceleration of electron beams across short distances (9 GeV energy gain, 1.3 meters). However, as electrons reach relativistic speeds, they outrun the laser beam that produces the plasma wakefield responsible for acceleration. For this reason, LWFA cannot accelerate electron beams over extended spatial and temporal distances. Hence, the production of higher energy beams requires the chaining of multiple acceleration stages together, utilizing active plasma lenses (APLs) in order to maintain the alignment of additional laser pulses and the accelerating electron beam. Within the APL, inconsistencies in the radial temperature gradient within the capillary induce severe transverse aberrations and lead to increased emittance in the electron beam. This project seeks to identify an optimized set of capillary parameters, notably gas species and pressure, radius and plasma density, which minimizes emittance, focusing on the combined minimization of Joule heating and Coulomb scattering. The computational process of this project will utilize PLUTO for magnetohydrodynamics simulations (modeling production of azimuthal magnetic field in APLs) and an Particle-in-Cell software for electron beam optics (electron movement through azimuthal magnetic field, while taking into account the plasma wakefield and general plasma background) in order to establish a synthetic dataset, baseline emittance and final optimized emittance. The generated synthetic dataset will be used in conjunction with governing partial differential equations to train a physics-informed neural network for surrogate modeling and machine learning optimization processes. Minimizing emittance in APLs preserves beam quality across stages, enabling higher luminosity in compact linear colliders and advanced high-energy physics applications.