Phase 3 Entanglement/Tomography IBM Execution Package¶
Date: 2026-05-20
Decision¶
- Campaign: reduced-Pauli entanglement/tomography follow-up.
- Backend preflight target:
ibm_marrakesh. - Live preflight artefact:
data/phase3_entanglement_tomography/entanglement_tomography_live_ibm_marrakesh_2026-05-20T001956Z.json. - Completed execution artefact:
data/phase3_entanglement_tomography/entanglement_tomography_live_ibm_marrakesh_2026-05-20T004334Z.json. - Status:
completed. - Hardware submission performed:
True. - Job IDs:
d86g7h1789is738vkreg,d86ggpis46sc73f6v170.
Live Preflight Summary¶
| Field | Value |
|---|---|
| Main circuits | 162 |
| Readout circuits | 4 |
| Total circuits | 166 |
| Main shots | 2048 |
| Readout shots | 8192 |
| Estimated QPU minutes | 1.5217 |
| Budget ceiling minutes | 25.0 |
| Selected physical qubits | [1, 2, 3, 4] |
| Mean readout error | 0.005584716796875 |
| Backend pending jobs | 0 |
| Maximum transpiled depth | 388 |
| Maximum basis-expansion ratio | 1.0718232044198894 |
The preflight passed the preregistered guards: the circuit count remains below
the DLA+FIM ceiling, estimated QPU time remains below the budget ceiling, and
measurement-basis expansion remains below the 1.20 depth-ratio guard.
Execution Commands¶
Live readiness only:
Approved QPU submission:
python scripts/phase3_entanglement_tomography_ibm.py --backend ibm_marrakesh --submit --confirm-budget
Post-run raw-count analysis:
python scripts/analyse_phase3_entanglement_tomography.py \
data/phase3_entanglement_tomography/entanglement_tomography_live_<backend>_<timestamp>.json
First-Pass Analysis¶
Post-run analysis produced:
- JSON summary:
data/phase3_entanglement_tomography/entanglement_tomography_summary_2026-05-20.json - Observable rows:
data/phase3_entanglement_tomography/entanglement_tomography_rows_2026-05-20.csv - Manifest:
docs/campaigns/phase3_entanglement_tomography_manifest_2026-05-20.md
Result snapshot:
| Metric | Value |
|---|---|
| Observable rows | 54 |
| Mean absolute deviation from exact reference | 0.12989296537986128 |
| Maximum absolute deviation from exact reference | 0.5560906424788263 |
| Rows SHA256 | 3d18308d60fe32827bae7517f18fd71690240b105779287408c4749cb0e7dc72 |
Paper Spin¶
Working title:
Reduced-Pauli entanglement checks for DLA-sector leakage mechanisms on IBM Heron hardware
Core result class:
- compare measured reduced-Pauli correlators against exact classical references for the promoted DLA parity and FIM-pair circuits;
- report whether the leakage/retention mechanism has a measurable entanglement-structure companion;
- include a null or negative result as a bounded measurement-cost and mechanism-separation result.
Blocked claims:
- quantum advantage;
- scalable tomography;
- backend-general entanglement dynamics;
- full-state reconstruction;
- claims on unmeasured subsystems, unmeasured depths, or unmeasured backends.
Paper-facing artefacts:
- Plan:
docs/campaigns/phase3_entanglement_tomography_paper_plan_2026-05-20.md - Canonical LaTeX source:
paper/submissions/submission_005_phase3_reduced_pauli_entanglement/phase3_entanglement_tomography.tex - Built PDF:
paper/submissions/submission_005_phase3_reduced_pauli_entanglement/phase3_entanglement_tomography.pdf - Historical Markdown scaffold:
paper/submissions/submission_005_phase3_reduced_pauli_entanglement/phase3_entanglement_tomography_short_paper.md
Completed Extensions¶
The initial execution package was extended on the same date with second-backend
replication, pinned-layout full correlated readout calibration, and a
preregistered five-channel ZNE stress test on ibm_fez.
Extension artefacts:
data/phase3_entanglement_tomography/entanglement_tomography_live_ibm_fez_2026-05-20T014536Z.jsondata/phase3_entanglement_tomography/entanglement_tomography_live_ibm_fez_2026-05-20T020452Z.jsondata/phase3_entanglement_tomography/entanglement_tomography_live_ibm_fez_2026-05-20T023204Z.jsondata/phase3_entanglement_tomography/entanglement_tomography_live_ibm_fez_2026-05-20T023600Z.jsondata/phase3_entanglement_tomography/entanglement_zne_summary_2026-05-20_ibm_fez_zne.jsondocs/campaigns/phase3_entanglement_zne_manifest_2026-05-20_ibm_fez_zne.md
Extension jobs:
- backend replication and full-readout calibration:
d86hdk8p0eas73dkv9eg,d86hedp789is738vm7mg - ZNE subset and 16-state readout calibration:
d86hs6qs46sc73f70h90,d86hsltg7okc73el4lg0
ZNE snapshot:
| Metric | Value |
|---|---|
| ZNE scale rows | 15 |
| ZNE channels | 5 |
| Scale-1 mean absolute deviation | 0.4196274906488838 |
| Linear-ZNE mean absolute deviation | 0.44124749932943935 |
| Full-readout-mitigated linear-ZNE mean absolute deviation | 0.4468173734963394 |
| Quadratic-ZNE mean absolute deviation | 0.4477931807530503 |
The ZNE extension is interpreted as a mechanism stress test only. The simple global-folding lane does not erase the four dominant DLA transverse deviations; the FIM control channel behaves differently and improves under linear extrapolation.
Boundary¶
This package records a completed, approved QPU submission and first-pass reduced-Pauli analysis. It supports only the bounded claim that measured small-system reduced-Pauli correlators deviate from exact references under the specified backend, layout, circuit family, shots, repetitions, and calibration window. It does not support scalable tomography, quantum advantage, or backend-general entanglement dynamics.