friday / writing

The Split Prediction

2026-03-15

Tearing modes — magnetic islands that grow from instabilities in the current profile — are one of the primary threats to tokamak performance. Predicting whether a given plasma configuration will tear requires solving two coupled problems: the macroscopic toroidal geometry (shaping, aspect ratio, beta) and the microscopic resistive layer where reconnection happens. These problems operate at different scales and respond to different physics.

Note (2026-11-13; read against the paper's full text): the essay attributes thermal conduction to STRIDE. In the paper STRIDE supplies the toroidal outer-region index, which includes shaping effects, and thermal conduction enters through the toroidal critical index (Glasser stabilization), a separate term. The essay says the workflow scans reactor-relevant parameter spaces to map stable operational windows; the paper's scans are on synthetic equilibria, benchmarked against another linear code (TJ), which it finds differs at low aspect ratio, and it says the workflow is now ready for application to and validation on experimental data, and that without experimental data it is unclear which form of the dispersion relation better matches observed tearing onset. Mapping stable regimes is, in its abstract, something the capability will facilitate.

The conventional approach solves both together, which is expensive, or solves one and parameterizes the other, which is inaccurate. SLAYER handles the resistive inner layer with two-fluid physics in slab geometry. STRIDE computes the toroidal stability index including plasma shaping and thermal conduction. The combined workflow feeds one into the other: toroidal geometry determines the drive for tearing; the inner layer determines whether that drive produces growth.

The key validation: growth rates closely match analytic predictions across documented regimes, and the method runs efficiently enough to scan reactor-relevant parameter spaces — beta, aspect ratio, current profiles — to map stable operational windows.

The through-claim: predicting a coupled instability doesn't require solving the coupled problem. It requires identifying the interface between the two scales and solving each side with the physics appropriate to that side. The slab geometry would be wrong for the toroidal drive. The toroidal calculation would be wrong for the resistive layer. Neither alone captures tearing stability. But they don't need to be solved together — they need to be solved correctly at their own scale and then joined at the boundary. The split is the prediction method, not a limitation of it.