Close both published RM3 PTO Extension free-decay gaps #618

Resolution: unchanged published-input physical gates

PR #664 archives the exact pinned MATLAB HDF5 and both unchanged 30 s, 0.1 s RM3_PTO_Extension body/PTO histories and adds an independent public Python physical implicit-added-mass gate. Both float and spar runs use the published no-wave inputs and 5 m extension, with no MATLAB motion, force, or solver-step histories driving Python. All 301 samples of both bodies' six position and velocity axes, PTO stroke and speed, and zero force and power pass the inherited original-input limits. The pinned source workflow regenerated both histories and checked the source data and hydrodynamic inputs. PR #665 records both as independent_motion; RM3 remains 21/21 and the total accepted-baseline count remains 59/63. The two original inputs are checked in #624.

The largest active float heave/stroke and speed gaps are 85.849 mm and 202.600 mm/s, below inherited 90 mm and 210 mm/s limits. The spar gaps are 14.877 mm and 6.522 mm/s, below 18 mm and 8 mm/s. The tightest margin is 7.400 mm/s, or 3.52% of the float-speed limit. These thresholds predate this physical-path audit but were not chosen blind to all earlier observed errors. The formal passes do not establish pointwise reproduction or source convergence.

Earlier diagnosis and retained caveat

The earlier source-compatible delayed-mass and refined-step comparisons showed float/spar active-heave differences of 47.1/6.84 mm and speed differences of 86.0/1.87 mm/s at the published step; these are not the qualifying physical-default results above. An opt-in source-ode4 route reproduces the published numerical trajectory, while source-only FixedStep refinement removes over 99% of the float and over 99.99% of the spar squared gap to a refined trace. The published MATLAB trajectories remain step-sensitive. The repository's separate simu.dt=0.00078125 s corrected baselines and full-duration physical gates remain archived as supporting evidence.

Evidence: published physical fixture and provenance, portable unchanged-input gate, parity ledger, earlier physical-step diagnostic, FixedStep control, source-ode4 comparison.