Anisotropic Maxwell neural operator for rapid parametric full-wave modelling of ion cyclotron resonance heating

This paper proposes the Anisotropic Maxwell Neural Operator (AMNO) for rapid parametric modeling of ion cyclotron resonance heating (ICRH) full-wave responses in the Experimental Advanced Superconducting Tokamak (EAST). The method learns within a one-parameter dielectric-field family, enabling efficient parameter sweeps and multi-case response analysis that would be computationally prohibitive with traditional full-wave solvers. The neural operator framework specifically captures anisotropic electromagnetic wave propagation in magnetized plasma.