Authors: Hameedullah Farooki (NJIT), Sung-Jun Noh (LANL), Youra Shin (NJIT), Hyomin Kim (NJIT), Haimin Wang (NJIT), Jason T. L. Wang (NJIT), Yasser Abduallah (NJIT), Qiang Hu (UAH), Yu Chen (UAH)

The solar wind is full of small-scale magnetic flux ropes (SMFRs), plasma structures having magnetic field configurations resembling those of coronal mass ejections, but plasma properties more akin to the properties of the background solar wind. The origin and geoeffectiveness of these SMFRs are still uncertain. In order to better understand SMFRs using additional data from spacecraft, it is necessary to efficiently detect SMFRs in large numbers. Currently, the most thorough and systematic way to detect SMFRs is the Grad-Shafranov (GS)-based algorithm, which is an exhaustive search method. We develop an optimized version of the GS algorithm using GPU processing, physical reduction of the search space, and deep learning. We also apply this methodology to larger volumes of higher cadence measurements of SMFRs and report on the statistical findings.