10:00–10:30 am ERC 401
We study the application of deep learning techniques to facilitate the analysis and classification of simulations of ions accelerated at shocks. Simulations of perpendicular and parallel shocks in collisionless plasmas were run using dHybridR (Haggerty et al. 2019) to capture the trajectories of ions encountering a shock. Ions were classified as thermal, supra-thermal, or non-thermal, depending on the energy they achieved and the acceleration regime they fell under. These classifications were used to train deep learning models to predict a particle's acceleration outcome with high accuracy, using only time series of the local magnetic field they experienced. An autoencoder was also tested, for which a time series of magnetic field information for a particle was used to recover momentum time series in the perpendicular shock. By studying these models, further insight is also given into the process of injection into shock acceleration. This study sets the groundwork for future advances in machine learning applications to acceleration at shocks.
Advisor: Damiano Caprioli
