This integration experiment uses the frozen NOAA IBTrACS snapshot and storm
track 2023251N16334 (MARGOT). It exercises one chain rather than separate
language slices:
- each of 91 samples is recorded as source-qualified evidence;
- recording alone is checked not to mutate the asserted position;
- the runner explicitly accepts the sample through
move_at; - sampled entry/exit and a six-hour departure monitor execute;
- a guarded normative containment constraint is evaluated;
- a hypothetical move executes on a clone; and
- the final asserted/evidence graph and SHACL shape are parsed and validated with RDFLib, pySHACL, and the GEOS function adapter.
Run the frozen case with:
python -m pulse_spatial.experiments.end_to_end `
--output-json experiments/end-to-end/results/ibtracs-four-mode-2026-07-19.json `
--output-markdown experiments/end-to-end/results/ibtracs-four-mode-2026-07-19.md `
--require-all-checksThe executed trace contains 90 accepted moves, 91 evidence records, three sampled entry/exit events, one duration-qualified event, one state transition, two normative violations before guard deactivation, an isolated scenario, and exact internal/SHACL agreement over 926 data triples and 6 shape triples. This is an integration case, not a productivity or industrial-deployment study.