Abstract Details
Abstracts
Author: Dario Panici
Requested Type: Consider for Invited
Submitted: 2023-03-24 19:15:01
Co-authors: R. Conlin, D. Dudt, P. Kim, E. Rodriguez, K. Unalmis, E. Kolemen
Contact Info:
Princeton University
710 Hibben Magie Road
Princeton, New Jersey 08540-6523
United States
Abstract Text:
Stellarator optimization is the search for attractive stellarators in the huge and varied phase space of all possible 3D magnetic configurations. New ways of attacking the problem can yield novel insights and better optimized stellarators. We present innovative approaches to stellarator equilibrium and optimization in the DESC code suite that are shown to yield excellent results over conventional methods. DESC’s automatic differentiation yields exact derivatives of objective functions, and also enables DESC to be the first stellarator optimization code to use only a single equilibrium solution at each iteration, which reduces the computation time by three orders of magnitude in tests compared to STELLOPT and enables exploration over a higher-dimensional parameter space. Further, a novel quasi-isodynamic (QI) metric has been developed and implemented in the code that can represent the full QI parameter space of interest. Boundary conditions constraining the near-axis behavior to match near-axis theory have been implemented in DESC and shown to result in much better optimized configurations. Constraining the Poincaré section instead of the last closed flux surface (LCFS) is also implemented and shown to recover LCFS solutions while requiring just a fraction of the variables to represent the boundary shape as compared to the LCFS. DESC has also been connected to the GX code to allow for turbulence optimization. Constrained optimization techniques implemented in DESC, such as the augmented Lagrangian approach, in tandem with these new capabilities allow for configurations with improved quasi-symmetry, quasi-isodynamicity, and turbulent transport properties to be found.
[1] Dudt, D. & Kolemen, E. Physics of Plasmas (2020).
[2] Panici, D. et al. JPP (Accepted) (2023).
[3] Conlin, R. et al. JPP (In Review) (2023).
[4] Dudt, D. et al. JPP (Accepted) (2023).
Comments: