Direct construction of optimized stellarator shapes. II. Numerical quasisymmetric solutions

TL;DR

This paper introduces a direct, optimization-free method to construct quasisymmetric stellarator magnetic fields using a differential equation, enabling faster and more efficient exploration of stellarator configurations.

Contribution

The authors develop a novel approach to generate quasisymmetric magnetic fields directly from a differential equation, bypassing traditional optimization techniques.

Findings

The method accurately produces quasisymmetric fields confirmed by VMEC and BOOZ_XFORM.

The space of solutions is parameterized by axis shape and three real parameters.

The approach enables rapid, exhaustive parameter space searches.

Abstract

Quasisymmetric stellarators are appealing intellectually and as fusion reactor candidates since the guiding center particle trajectories and neoclassical transport are isomorphic to those in a tokamak, implying good confinement. Previously, quasisymmetric magnetic fields have been identified by applying black-box optimization algorithms to minimize symmetry-breaking Fourier modes of the field strength $B$. Here instead we directly construct magnetic fields in cylindrical coordinates that are quasisymmetric to leading order in distance from the magnetic axis, without using optimization. The method involves solution of a 1-dimensional nonlinear ordinary differential equation, originally derived by Garren and Boozer [Phys. Fluids B 3, 2805 (1991)]. We demonstrate the usefulness and accuracy of this optimization-free approach by providing the results of this construction as input to the codes VMEC and BOOZ_XFORM, confirming the purity and scaling of the magnetic spectrum. The space of magnetic fields that are quasisymmetric to this order is parameterized by the magnetic axis shape along with three other real numbers, one of which reflects the on-axis toroidal current density, and another one of which is zero for stellarator symmetry. The method here could be used to generate good initial conditions for conventional optimization, and its speed enables exhaustive searches of parameter space.