Position-based motion software passes its first qualification
All ten identities passed: 24 analytic paths and 2,304 noisy realizations. The supplied-error calculations covered all nine trend coefficients in 252–256 of 256 realizations per family. Their exact one-sided 99% coverage lower bounds are at least 0.955, above the frozen 0.95 requirement. The arbitrary-correlation bound covered every realization.
Ignoring correlation is materially different. In correlated families, the independence calculation covered only 103–127 of 256 realizations. A smoother position-derived path therefore cannot justify treating its errors as independent. At the largest simulated noise, point estimates also produced false turn flags.
The analytic paths check the changing geographic metric, both hemispheres, longitude wrapping, nonuniform epochs, turning and climb. Maximum course error was 1.81e-10°/s and maximum speed error 8.22e-10 m/s. These are software agreement figures, not real receiver accuracy.
The run took 18.5 seconds and reconciled all 2,328 reserved assessments. Every generated input and interval is retained. No receiver file, gyroscope observation or protected confirmation group was read. The existing ESA selection still admits zero intervals.
Next is a separate, bounded receiver-only comparison on development recording 187. It will retain unsupported windows and distinguish uncertainty in a fitted trend from unknown instantaneous aircraft motion. The real receiver covariance and its temporal behavior remain uncalibrated.