Generative, energy-based models for diagnostic reconstruction and analysis

Energy-based models (EBMs) provide a powerful and flexible way of learning relationships in data by constructing an energy surface. We extend EBMs to laboratory plasma physics, a domain characterized by highly nonlinear phenomena studied using incomplete diagnostic information. These diagnostics can be unreliable or difficult to analyze. In addition, the possible configuration space of a plasma device is sufficiently large that it cannot be efficiently searched using conventional analysis techniques. EBMs provide a way to address these issues. At the Large Plasma Device (LAPD), a CNN- and attention-based EBM is trained on a set of randomly generated machine conditions and their corresponding diagnostic time series. We demonstrate diagnostic reconstruction using this EBM and also show that including additional diagnostics improves reconstruction error and generation quality. Symmetries in the data can be found by directly evaluating the energy surface, potentially leading to a new line of inquiry using learned models. In addition, this multimodal EBM is able to unconditionally reproduce all distributional modes, suggesting future potential in anomaly detection on the LAPD. Fundamentally, this work demonstrates the flexibility and efficacy of EBM-based generative modeling of laboratory plasma data, and showcases practical use of EBMs in the physical sciences.