Numerical Methods for Identifying Integrability With Applications in Plasma Confinement
Plasma confinement physics offers a wide variety of problems for scientific computation, especially as the plasmas approach fusion temperatures and pressures. Algorithms are required for finding plasma equilibria, quantifying confinement, assessing stability, and many more tasks. Moreover, algorithms must be fast and reliable if they are to be used for design and optimization. Here, we present multiple projects for efficiently computing for stellarators, a type of magnetic confinement device. Much of the thesis is concerned with the integrability of magnetic fields, an important quality for confining particles and sustaining temperatures and pressures. We present two algorithms for measuring, classifying, and parameterizing integrability. First is a kernel-based method for assessing the integrability of a whole plasma volume, and second is an extrapolation-based method for classifying trajectories and parameterizing invariant tori. In the final chapter of the thesis, we consider the problem of quickly computing plasma equilibria themselves. We show that the near-axis expansion, a recently popularized method for equilibrium solving, is ill-posed. Then we regularize this problem, showing that high order equilibria can be found at times significantly faster than full 3D PDE solvers. Finally, we show that we can compute high-order corrections to flux coordinates using the extrapolated series, connecting with the theme of computing integrable structures.