Finite-grid event responses to dynamics interventions in stochastic simulators

A bounded simulator audit of two declared dynamics substitutions. Rotating an Ornstein–Uhlenbeck diffusion covariance preserves its trace and eigenvalues and reduces event-axis noise variance from 0.22 to 0.04. Archived finite-grid counts show that the drop in observing B first is carried by fewer boundary hits before the horizon. A nonlinear drift-gain example gives start- and horizon-dependent contrasts, with three of nine simultaneous intervals including zero. No biological mechanism is claimed, and no external DOI is assigned.