What this diagnostic establishes

All 18 exact-trajectory controls passed, covering three physical models and six motion profiles. The independent Cartesian and local-navigation calculations agree to about 1.8 × 10⁻¹⁵ m/s² in required force. That is software agreement, not the precision or accuracy of a real instrument.

Stationary controls correctly retain indistinguishable force magnitudes. Some moving controls produce different magnitudes under different models; the largest incorrect-model RMS mismatch in this matrix is about 0.030 m/s².

This is an important limit on interpretation: with exact calibration and an exact receiver trajectory, acceleration and coordinate geometry alone can separate some generated cases. It does not show how much of the real-data result comes from those inputs. The full gyro/acceleration comparison cannot be divided into percentages by subtracting these diagnostic scores.

The two changing body attitudes give the same magnitude residual, so this calculation does not borrow a heading or fused orientation. It still depends on the candidate coordinate metric, the stated background term, apparent gravity and acceleration calibration. A mismatch of this size is not a detection threshold or evidence that actual calibration errors are smaller.

No recordings were read, no model was fitted to observed data, and the existing 53-stretch findings remain unchanged. The fixed background rate in these controls is a declared software input, not a measured solar/sidereal rate.

Complete control matrix · Receiver-derivative groundwork · Remaining gate.