Graph-based Techniques for Orbit Classification: Early Results

An important step in the quest for low-cost fusion power is the ability to perform and analyze experiments in prototype fusion reactors. An automated analysis and interpretation software toolkit allows physicists to quickly analyze their experiments and plan new ones. In this preliminary report, they consider the analysis of Poincare plots that contain the orbits generated by the particles in a fusion reactor. They describe how they can use graph-based methods to extract features from orbits. These features are then used to evaluate the accuracy of machine learning algorithms and their response on orbits with few points