Neural network eigenmode classification in the MHD spectroscopy code Legolas
Date:
Seminar at the Centre for mathematical Plasma Astrophysics, KU Leuven, Belgium.
Abstract. For a given one-dimensional magnetohydrodynamic equilibrium, the spectroscopy code Legolas computes the linear eigenoscillations for this state. Depending on the specifics of the equilibrium, multiple types of modes may be present and an automated classification (or first guess) may be desirable. To classify the Legolas output, we turn to both supervised and unsupervised machine learning algorithms. While similar, these two different machine learning techniques have distinct applications. Here I present an introduction to both techniques, some preliminary results of how we applied them to Legolas and offer some perspectives for the future.