Local Level Lab

Airborne measurements. Open methods.

Saved horizontal correction audit

All twelve saved endpoints were inspected with three affine formulations each: 36 linear solves, zero native predictions and zero model refits. The failed parent qualification remains unchanged.

Endpoint Original fit passed Rank / columns Original probe max step Scaled probe max step Ridge probe max step
sphere_rotating-bias1_unsubtracted-horizontal-zero True 23/26 0.001 0.001 5.53e-07
sphere_rotating-bias1_unsubtracted-horizontal-opposed True 23/26 0.001 0.001 3.99e-05
sphere_rotating-bias1_profiled_removal-horizontal-zero True 24/27 0.001 0.001 5.51e-06
sphere_rotating-bias1_profiled_removal-horizontal-opposed True 24/27 0.001 0.001 1.18e-05
sphere_still-bias1_unsubtracted-horizontal-zero True 24/26 0.001 0.001 0.000115
sphere_still-bias1_unsubtracted-horizontal-opposed True 23/26 0.001 0.001 1.71e-05
sphere_still-bias1_profiled_removal-horizontal-zero True 24/27 0.001 0.001 4.24e-05
sphere_still-bias1_profiled_removal-horizontal-opposed True 24/27 0.001 0.001 7.92e-06
flat_still-bias1_unsubtracted-horizontal-zero False 23/26 0.001 0.001 2.1e-06
flat_still-bias1_unsubtracted-horizontal-opposed False 23/26 0.001 0.001 1.52e-06
flat_still-bias1_profiled_removal-horizontal-zero False 24/27 0.001 0.001 2.06e-06
flat_still-bias1_profiled_removal-horizontal-opposed False 24/27 0.001 0.001 1.06e-06

These are linear predictions at saved endpoints. They cannot establish a fresh nonlinear improvement, change stopping labels or qualify observed-data fitting. Ridge regularization changes the diagnostic correction objective; it does not change any physical-model score.

Scope and call-allowance correction · Interpretation · Next analyses.