Local Level Lab

Airborne measurements. Open methods.

Receiver-error weighting checks

All 21 policy checks and 55 supporting covariance checks pass. The working policy retains the receiver's complete north/east/up covariance and tests two stated correlation assumptions.

The checks verify cross-axis terms and height direction, irregular timestamps, invalid covariance rejection, unchanged input measurements and independence from gyro or preferred-model metadata. Dense and structured calculations agree for a 27-column matrix, including a zero column. Dense requests have an explicit size cap to prevent accidental whole-record allocation.

Source and environment bindings were written before the final test run. The exact 14 source files, test report and receipt are retained through the shared reproducibility archive, under data/ilvis0-source-point-development/esa-error-policy-checks-20261023.

This verifies software for the declared weighting assumptions. It does not calibrate receiver accuracy, select an interval or fit any flight. The original zero-eligible inventory and all protected confirmation groups are unchanged.

To rerun the software checks in the matching environment:

~/venv/bin/python analysis/tests/run_numerical.py --workers 1 --threads 4 -- ~/venv/bin/python -m pytest -q analysis/tests/test_airquantumgrav_error_policy.py analysis/tests/test_ilvis0_structured_covariance.py analysis/tests/test_ilvis0_error_model.py

Working assumptions · Next analyses.