Scalar calibration projection controls
72/72 software controls passed in 3.85 seconds. No original recordings, observed fits or native integrations were used.
The matrix compares stationary, eastbound, changing-direction and climbing reference motion under three truth models, all three candidates and two body-orientation histories. All seven calibration columns remain in the rank calculation.
All 72 outcomes include ranks, local projection, feasibility and actual nonlinear residual at feasible linear corrections. These are mathematical controls, not measured device errors or calibrated evidence.
| Motion | Cases | Rank range | Largest uncorrected RMS | Largest projected RMS | Feasible linear steps |
|---|---|---|---|---|---|
| stationary | 18 | 4–6 | 8.234253e-16 | 8.108177e-16 | 18 |
| eastbound | 18 | 6–6 | 0.03006832 | 1.119002e-15 | 18 |
| changing direction | 18 | 7–7 | 0.04140383 | 0.00433956 | 6 |
| climbing | 18 | 7–7 | 0.0206112 | 0.0007693376 | 6 |
RMS values are in m/s². Maxima can belong to different cases; compare individual rows for matched changes.
A local projection is unbounded and first order. Its removed difference need not be attainable with feasible nonlinear calibration; its remaining difference does not rule out other nonlinear solutions. No model verdict follows.
The main 53-stretch results and unresolved horizontal-only sensitivity outcome are unchanged.
Frozen plan and limits · Calibration ambiguity · Measurement contributions.