Neural gyrokinetics to predict confinement — not just equilibrium.
The AI-Native S.M.A.R.T. Generator Master Blueprint — eight layers (L0→L7), one control stack, wired to both machines. Telemetry rises in microseconds; control descends the same path.
Category: B · physics · Plugs into: L3 · Horizon: NOAK · Status: on the roadmap — not yet built
What it is
Confinement (τ_E) is set by turbulent transport, which today's equilibrium PINN does not capture. A neural surrogate of reduced gyrokinetics (TGLF / QuaLiKiz-class) predicts turbulent fluxes fast enough for control.
The method
Train a network on gyrokinetic databases to emit heat and particle fluxes as a function of the gradients; couple it to the transport twin for real-time τ_E prediction.
Why it matters
It closes the gap between "we solved the equilibrium" and "we predict the confinement" — the quantity that actually sets the operating point. Plugs into L3 as a transport module in the twin.