Reconstruction Gating in a Paired Non-Equilibrium Field System

This dataset demonstrates that reconstruction gating causally improves reconstruction efficiency without increasing geometric fragmentation, under a strictly non-circular paired design. Gated and ungated systems are evaluated under identical exogenous inputs and shared stochastic noise realizations to ensure direct comparability. Gate activation is determined exclusively from a reference (control) trajectory, enforcing strict non-circularity. Activation thresholds are selected automatically based on quantiles of a reconstruction score computed solely from the reference system. Results show that reconstruction gating produces a robust and statistically significant increase in reconstruction efficiency while simultaneously reducing geometric fragmentation. Injection-matched supplementary experiments demonstrate that fragmentation is primarily injection-dominated, whereas reconstruction efficiency remains strongly gate-dependent, establishing a clear causal separation between structural and informational effects. A secondary audit metric, the mean reconstructed mass, indicates that gating regulates reconstructed mass rather than amplifying it, preventing uncontrolled accumulation while preserving superior reconstruction efficiency.All simulation code, analyses, figures, and cryptographic hashes of exogenous inputs are included to ensure full reproducibility and independent auditability.