Finite-time recovery and forward-loss defects in a shearing magnetofluid

This preprint presents a focused archival reanalysis of finite-time recovery in an eight-component, self-gravitating magnetofluid. Reversing momentum and background rotation while retracing the physical shearing wavevectors gives an exact reversal identity for the retained ideal operator. The analysis distinguishes that continuous-time identity from numerical recovery under a finite timestep. The ideal evidence contains 48 return cases across two background settings and two timestep refinements. Six targeted full-shear branches were repeated as complete forward/return trips at timestep 1/8960. Their maximum terminal and mirrored-history relative errors are 8.67 × 10⁻¹⁵ and 1.48 × 10⁻¹⁴, normalized by each branch's initial perturbation norm. These near-roundoff results apply to that refined subset. A separate factorial comparison examines 72 returns after resistive, cooling, or combined forward evolution, with 24 inherited ideal controls that overlap the original ideal panel. Every return uses the ideal inverse. At the prescribed rates and duration, cooling-only family-median defects span 0.259-0.472, compared with 0.0391-0.0511 for resistivity alone. This ordering is specific to the tested parameters and protocol. An exact flow-map relation expresses the defect through nonideal tendencies transformed by the ideal inverse, conditional on admissible backward flows. A state-return defect is not dissipated energy or evidence of fundamental information loss. Numerical recovery at fixed cutoff does not establish spatial convergence, which remains unresolved. The deposit supplies the paper, detailed supplement, editable sources, figures, compact evidence, and saved-data reproduction scripts. The evidence is reused from the parent study (https://doi.org/10.5281/zenodo.23633307); no new trajectories are integrated in this reanalysis, and companion papers are not independent replications. OpenAI Codex assisted with source inspection, mathematical discussion, saved-data reduction, plotting code, and manuscript preparation. Jake Campbell directed the research and accepts responsibility for the manuscript.