Negative heat diffusivity leaves the MAST-U reference inadmissible
We ran the approved trapped gyro-Landau fluid (TGLF) diagnostic to explain why the MAST Upgrade reference rejects a face the candidate surrogate serves. The matched scenario, solver, inputs, guard and every face were preserved.
The diagnostic reproduces the saved profiles and final reference mask exactly. The uncovered face has valid prepared inputs but negative raw electron heat diffusivity at the final state. Negative heat diffusivity also rejects it at other accepted states. The guard therefore retains its physics fallback.
| evidence | result |
|---|---|
| bounded diagnostic and saved-output audit | the diagnostic completes in 517 seconds within its 30-minute bound, with 40 finite converged steps, exact historical temperature/density/poloidal-flux-profile and final-mask equality, and 171 passing saved-output integrity checksC-193 |
| uncovered face and rejection history | final face 4 at normalized radius 0.16 has valid inputs but raw electron heat diffusivity -0.02792124888 m²/s; negative heat diffusivity rejects that face at 8 of 40 accepted states; reference rejection is 323/1040 accepted-state face records (31.0577%)C-194 |
| formal candidate temperature criterion | remains not evaluable |
A read-only audit verifies sources, inputs, full actual configuration, accounting, trajectory provenance and completed offline artifact checksums. It reconstructs worker-input policies and the raw guard for every recorded face and state. It uses the same report logic, so it checks integrity rather than independently validating the physics. Exact historical profile equality is a separate saved-array comparison. Missing historical dependency bytes remain explicit.
The diagnostic identifies the guard cause. It does not establish the physical or numerical origin of the signed heat coefficients: direct heat-flux and mode-resolved solver outputs were not saved. A physically justified reference remedy needs owner review before another comparison. The separate feedback-sampling proposal remains unapproved for fitting.
K-030 stays open; the current model and guard remain selected. No face, tolerance or physical contract changed. This diagnostic approval is consumed. Tracking remains offline and nothing has been synchronized, replaced or published.
What’s next. Review reference treatment for the signed transport evidence, then prepare any additional bounded collection or remedy for its own review.
Technical details → Verification › reference diagnostic result
Decisions → OQ-31 · OQ-32