Magnetohydrodynamic eigenfunction classification with a Neural Network
Kuczynski, M. D. 1Borchardt, M. 1Kleiber, R. 1Koenies, A. 1Nuehrenberg, C.1
扫码查看
点击上方二维码区域，可以放大扫码查看
作者信息
1. Max Planck Inst Plasma Phys
折叠
Abstract
We present a Fourier-decomposition-based approach aided by a Neural Network for the classification of the eigenfunctions of an operator appearing in ideal magnetohydrodynamics. The Neural Network is trained on individual Fourier modes, which enhances the robustness of the classification. In our tests, the algorithm correctly classified 93.5% of the data and returned the remaining 6.5% for manual classification. The probability of misidentifying the eigenfunctions is estimated as 0.03%. The discussion is kept quite general allowing for potential applications in other fields. (c) 2021 Elsevier B.V. All rights reserved.
NSTL
Elsevier