domain/

Package allocation divides a fixed alert budget across detectors of unequal quality.

Package calibration implements calibration and combination (§10).

Package cellgrid implements the 168-cell timing representation of §7.1 — the construction the paper argues is defective — as an ablation detector for evaluation hypothesis E9.

Package cooccurrence implements Detector III, population co-occurrence (§8).

Package derive infers structure inside field values and registers a coarser field beside the original.

Package event implements the event model of §5.1, equation (3).

Package marginal implements Detector IV, the population-scope marginal outlier (§9).

Package novelty implements Detector I, per-entity categorical novelty (§6).

Package noveltyrate asks whether an entity is producing MORE first-ever values than it normally does, rather than how improbable any single first-ever value was.

Package objective states what an alerting configuration is *for*, as a quantity that can be maximised.

Package pairing is Detector III's per-entity replacement: is this combination of values novel for THIS entity, against its own history?

Package registry implements the field registry of §5.1.

Package statistics provides the interval estimators and paired tests the evaluation tables require.

Package timing implements Detector II, periodic timing (§7.2).

Package volume implements the volume half of Detector II (§7.4).