🤖 AI Summary
To address the high computational cost and difficulty of parameter optimization in stellarator design, this work proposes a deep learning–based inverse design method. Specifically, it introduces Mixture Density Networks (MDNs) for the first time to generate fusion device configurations, leveraging probabilistic modeling to handle solution non-uniqueness inherent in multi-constraint inverse problems. The method directly synthesizes high-aspect-ratio magnetically confined configurations satisfying multiple physics-driven constraints—including low elongation, high rotational transform, finite plasma beta, and excellent fast-ion confinement. Integrated with a simplified physics model and a differentiable stellarator parameterization framework, the approach consistently produces configurations that balance engineering feasibility and plasma physics performance. It achieves target performance metrics while substantially reducing computational overhead compared to conventional simulation-driven design, thereby enhancing both design efficiency and scalability.
📝 Abstract
The design of fusion devices is typically based on computationally expensive simulations. This can be alleviated using high aspect ratio models that employ a reduced number of free parameters, especially in the case of stellarator optimization where non-axisymmetric magnetic fields with a large parameter space are optimized to satisfy certain performance criteria. However, optimization is still required to find configurations with properties such as low elongation, high rotational transform, finite plasma beta, and good fast particle confinement. In this work, we train a machine learning model to construct configurations with favorable confinement properties by finding a solution to the inverse design problem, that is, obtaining a set of model input parameters for given desired properties. Since the solution of the inverse problem is non-unique, a probabilistic approach, based on mixture density networks, is used. It is shown that optimized configurations can be generated reliably using this method.