Machine Learning Algorithms for the Detection of Plasmoids in Multiple-X-Line Collisionless Reconnection Regions

Correctly identifying structures in multiple-X-line reconnection regions is crucial for understanding the physics of the coupling of the microscale to the macroscale, such as the potential role that the plasmoid instability plays in reconnection dynamics and energy transfer. One specific area of research where this is important is the study of naturally occurring reconnection regions in Earth's magnetotail via analysis of in-situ data from spacecraft. A limitation of this data is that spacecraft can only sample a single point in space for each timestep, and trace a 1D path through the plasma. This limitation makes detection and identification of dynamic plasma structures difficult, especially if the plasma is sampled by only a single spacecraft. Techniques such as identifying structures by eye and by fitting to mathematical models are commonly and effectively used, but neither is suitable for the detection of large numbers of structures which are stretched or warped from their idealized shape. Previous work tackling this methodological problem used simple hand-tuned algorithms for detection and classification (Bergstedt et al. 2020). This work develops a more nuanced and robust detection algorithm which utilizes a set of simulated 'spacecraft' trajectories through 2D PIC simulations of reconnection to train a machine learning model to identify regions of data corresponding to plasmoids. The results from a simple binary classifier based on a 1D Convolutional Neural Network (CNN) architecture are presented and evaluated. Potential applications of the classifier are discussed.