Comparison of causality determination techniques in studying Hall-effect thrusters

Various causality determination techniques are compared, with a focus on application to Hall-effect Thruster (HET) systems. The methods studied include Extended Convergent Cross Mapping (eCCM), Reservoir Computing-based Causality (RCC), and information theory-based calculations of the transfer entropy (TE) and causation entropy (CE). Instead of choosing the “best” method, the advantages and disadvantages of each technique are identified and discussed. First, sample systems with predetermined causal links are analyzed in order to determine false positive and negative results for each algorithm. Then, 11 signals from a Hall thruster running in both quiescent and oscillatory modes are analyzed, and the results of each method compared. The results presented herein show that the methods compared herein provide trade-offs between performance and accuracy, and should be carefully considered for each specific application.