Bounded-model equilibrium reconstruction for a tokamak digital twin
Anulum Institute / SCPN Fusion Core
Challenge
Tokamak digital twins combine MHD equilibrium solvers, transport models, and control loops into a single predictive system. Each component makes assumptions — axisymmetry, closure models, synthetic diagnostic calibrations — but these assumptions are rarely surfaced as part of the claim. A reconstruction can look validated while only being closure-checked against idealised synthetic data.
Approach
Fusion Studio treats each equilibrium reconstruction as a bounded-model claim by construction. The reconstruction pipeline emits a studio.equilibrium-reconstruction.v1 evidence bundle that records the validity domain, the evidence kind (measured), the empirical level, and the exact synthetic equilibria against which the closure was checked. The claim boundary is rendered verbatim so a reader cannot mistake a synthetic-closure check for an experimental validation.
Results
The pipeline produces reproducible reconstruction bundles for a range of plasma shapes and current profiles. Every bundle carries the closure assumption set and the synthetic-equilibria digest; the publication that cites the bundle states explicitly that the result is bounded-model and does not yet claim experimental validation.