LSFM Interpretation of Regime Transitions in HL-3 Disruption

Status: Unreviewed preprint (Zenodo). Independent, unconfirmed research; not peer-reviewed.

Recent disruption prediction on the HL-3 tokamak (Yang et al., Nuclear Fusion 2025) revealed systematic drift across five operational regimes, necessitating Predict-First Neural Network (PFNN) architectures for reliable performance. We propose an interpretation using the Log-Scaling Flow Metric (LSFM), χ(λ) = [2 ln(λ)] −1 , proposed as a measure of deviation from recursive scaling at λ * = e ≈ 2.718. We hypothesize that lower-disruption regimes may correspond to χ near 0.5. We list six predictions that could be tested, including real-time χ monitoring for disruption avoidance. If confirmed, this could inform ITER commissioning under non-stationary conditions.

Note: a separate re-analysis of the ITPA H-mode database found that χ's correlation with confinement quality (H98) mostly comes from both quantities sharing τ_E (in preparation).