Reduced-Pauli entanglement checks for DLA-sector leakage mechanisms on IBM Heron hardware

This preprint reports a cost-bounded reduced-Pauli entanglement and tomography check for promoted Phase 3 DLA-parity and Fisher-information-modified Kuramoto--XY circuit families. The purpose is to distinguish parity-sector leakage structure from layout, prepared-state, and readout artefacts in small superconducting-hardware experiments. The archived IBM execution used physical qubits 1, 2, 3, 4 on ibmₘarrakesh, 162 main circuits, four readout-calibration circuits, 2048 shots per main circuit, and 8192 shots per readout circuit. A preregistered reducer produced 54 observable rows; the manuscript reports measured-minus-exact deviations, identifies the largest transverse two-qubit correlator deviations, and compares readout-mitigated replay and same-day ibmfez replication. The record includes manuscript source, bibliography, and PDF. The claim boundary is hardware-observation and artefact-reduction evidence for these named circuit families, not a backend-independent universal promotion.