Preregistered Tests of Local-Noise Predictions for GHZ Coherence Under Matched Qubit Environments on IBM Kingston

DOI: 10.5281/zenodo.22283967Record set: EC-MECH campaign (EC-MECH-001 pilot + EC-MECH-002)Related to: EC-STORAGE-001, DOI 10.5281/zenodo.21299091 Research question: Can the coherence lifetime of an entangled multi-qubit state be predicted from the measured coherence losses of its individual qubits? Plain language summary When several qubits are entangled together, how fast does the entangled state lose its coherence compared with the qubits measured one at a time? The textbook expectation, if each qubit's noise is its own private business, is that the losses simply add up. This record tests that expectation directly on IBM hardware, and reports two experiments: one that returned ABSTAIN under its own preregistered rules rather than supporting a scientific conclusion, and a redesigned successor that did produce one. The first experiment failed for an instructive reason. Its quality gate asked whether the measured decay curves looked like clean exponentials, when the question that actually mattered was whether the decay rate had been pinned down precisely. Those are different things, and for shallow decays they come apart badly. Five measurements whose rates were known to within 4–6% were discarded because their curves were too flat for the goodness-of-fit statistic to work with. Under the frozen rules that cascaded into an ABSTAIN on every downstream question. The verdict stands unamended. Diagnosing that failure exposed a second and more serious problem: the single-qubit reference measurements had been performed in a different noise environment from the entangled measurements they were being compared against. The redesigned experiment fixed both problems and added a dedicated probe of the environment mismatch itself. The result: once the environments were matched, the discrepancies during plain idling became substantially smaller — especially for the 4- and 6-qubit states — but the preregistered precision was still insufficient to establish additivity. Under a standard error-suppression pulse sequence, additivity was rejected at all three sizes. And the dedicated probe confirmed the environment mismatch was real, not merely a theoretical worry. Total hardware cost: 241 seconds across both experiments. Background: an unsupported claim, entered into the record EC-STORAGE-001 (DOI 10.5281/zenodo.21299091) recorded a pre-registration miss — a bare-arm coherence witness crossing at 9.0 µs against a registered 10–30 µs band — and explained it by asserting that the payload resided in weight-4 stabilizer correlations "whose coherences decay at the sum of constituent rates." That explanation does not close numerically against data in the same deposit. The selected qubits had reported T₂ of 178–368 µs. A sum-of-rates model over four such qubits predicts a joint coherence time of roughly 45–60 µs; the measured value was 8.81 ± 0.30 µs. The stated model over-predicts by a factor of roughly 5–7. The claim was therefore a hypothesis written in the grammar of a derivation. It is entered in the adjudication ledger as UNSUPPORTED, and the EC-MECH campaign was constructed to either repair it or retract it. This deposit does not resolve it in EC-STORAGE-001's favour, and readers of that record should treat the mechanism sentence as withdrawn pending a direct test on the encrypted-cloning encoding itself. What we did Both experiments ran on ibm_kingston (156-qubit IBM Heron processor) on a connected 6-qubit chain, physical qubits [14, 15, 19, 35, 34, 33], selected by a frozen policy from the same-day calibration snapshot with no manual override. Neither experiment uses, requires, or reproduces the encrypted-cloning protocol. They test the underlying physics assumption in isolation, using GHZ states and idle delays only. No proprietary components are involved, and the deposited code is fully self-contained. EC-MECH-001 (pilot) — rate additivity Nested GHZ states on the first k qubits (k = 2, 4, 6) were idled for τ ∈ [0, 45] µs and their weight-k coherence read out via parity oscillation. In parallel, all six qubits were prepared in |+⟩ and idled simultaneously to obtain per-qubit in-situ dephasing rates. The frozen predicate compared the fitted GHZ decay rate Γ_k against the sum Σ Γᵢ of the measured single-qubit rates. 320 circuits, 1024 shots each, one job, 93 s QPU. EC-MECH-002 — coherence-function additivity The successor abandons fitted rates entirely. For any family, define c(τ) = C(τ) / C(0) χ(τ) = −ln c(τ) Under independent local phase noise, the GHZ phase is the sum of the local phases, so the coherence factorises exactly: χ_S(τ) = Σ_{i∈S} χ_i(τ) This identity assumes nothing about decay shape — exponential, Gaussian, stretched, and non-Markovian decays all satisfy it. The scientific object is the residual Δ_S(τ) = χ_S(τ) − Σ χᵢ(τ), adjudicated through the scale-free ratio r_S(τ) = Δ_S(τ) / Σ χᵢ(τ) as a preregistered equivalence test with margin |r| ≤ 0.15, using simultaneous 95% confidence intervals (Bonferroni, n = 8, z = 2.734). CI wholly inside the margin → ACCEPT; wholly outside → REJECT; overlapping the boundary → ABSTAIN. Two design repairs distinguish it from the pilot: No R² gate, no fitted rate, no assumed decay law. Matched noise environments. Every non-target chain qubit is pinned in |0⟩ — including the GHZ spectators at k < 6 — rather than left in |+⟩. This matters because a GHZ block is immune to intra-block ZZ coupling: |0…0⟩ and |1…1⟩ are both +1 eigenstates of Z_iZ_j, so the relative phase carrying the coherence is untouched. In the pilot, the single-qubit reference was exposed to neighbour-state-dependent dephasing consistent with this mechanism, while GHZ symmetry cancels intra-block static ZZ — biasing the prediction high. Single-qubit controls use a two-colour scheme on the chain (targets {14, 19, 34}, then {15, 35, 33}) so every target has all chain neighbours pinned. τ = 0 is oversampled at 4096 shots because it is the shared normaliser and its uncertainty enters every χ, inducing covariance that is carried explicitly in the analysis. Design parameters (frozen). Normalisation point τ = 0 at 4096 shots, never adjudicated. Eight informative τ points at 1024 shots each: 12, 18, 20, 22, 25, 28, 35, 45 µs The grid is pilot-informed and deliberately non-uniform: the low end starts at 12 µs because the D_min = 0.10 denominator gate would exclude earlier times once neighbour pinning reduces the local χ, and five of the eight points are clustered in the 18–28 µs window to resolve structure in Δ(τ) there. See the evidence ceiling for what this costs. Arms: bare (plain delay) and dd (symmetric XY4, one cycle). Four phase points per sweep, spanning one full period of the weight-k oscillation. Runtime dynamical decoupling and twirling disabled, so the arms are defined solely by the circuits. Diagnostic family diagPlus on the bare arm at τ ∈ {12, 25, 45} µs — all three exact members of the primary grid, so no interpolation is performed. Circuit execution order randomised under frozen seed 20260902. 376 circuits, 520,192 shots, one job, 148 s QPU against a preregistered estimate of ~139 s (6.1% error). Driver provenance. The EC-MECH-002 driver was hardened after the EC-MECH-001 pilot and before the EC-MECH-002 freeze. The hardening bound submission to the freeze manifest and to the binding layout report (removing hand-entered qubit chains), replaced diagnostic interpolation with exact grid indexing, added randomised circuit execution order under a frozen seed, and added a drift-robustness case to the offline validator. All of it predates the freeze; the deposited digest covers the hardened files, and no change was made after data was seen. Results EC-MECH-001 — ABSTAIN (frozen, unamended) All six weight predicates returned ABSTAIN. Cause: five single-qubit component fits failed the frozen R² ≥ 0.90 gate (bare q15 = 0.883, q33 = 0.873; DD q15 = 0.893, q19 = 0.871, q33 = 0.859) despite relative rate uncertainties of 3.7–5.5%. Because q15 participates from k = 2 onward, the failure cascaded into every weight. All 18 fits in the run had σ(Γ)/Γ ≤ 0.064. R² measures the fraction of variance in log C explained by the line; when a decay is shallow the true variance is small and ordinary scatter consumes a large share of it. R² was the wrong gate. The pilot did establish, as observation rather than verdict, that in-situ dephasing under simultaneous idling ran ~2.4–4.3× faster than the reported T₂ on every qubit (e.g. q35: 67.4 µs in situ against 292.9 µs reported). EC-MECH-002 — primary adjudication Arm k usable τ median r χ²/dof p Verdict bare 2 7 −0.217 2.40 0.0185 ABSTAIN bare 4 8 −0.014 0.62 0.7625 ABSTAIN bare 6 8 −0.087 1.16 0.3166 ABSTAIN dd 2 7 −0.396 5.46 <0.0001 REJECT dd 4 8 −0.213 5.36 <0.0001 REJECT dd 6 8 −0.206 6.90 <0.0001 REJECT Negative r means the GHZ state retains coherence better than the independently measured single-qubit coherences predict. The bare ABSTAINs are not a proof of independence. The equivalence predicate was built precisely so that "we could not reject zero" cannot be reported as "we demonstrated independence." At k = 4 and k = 6 the residual is small (−0.014, −0.087) and the confidence intervals straddle the ±0.15 boundary; the correct statement is that a positive equivalence claim was not supported at the preregistered precision. k = 2 warrants extra caution in both arms. It is the only weight that lost a τ point to the denominator gate, and because r = χ_S/D − 1 with the smallest denominator, it amplifies any bias in D more than the other weights. Its values (−0.217 bare, −0.396 DD) are the most extreme on the board and the least reliable. ZZ diagnostic (preregistered as diagnostic, excluded from the predicate) A dedicated family reproduced the pilot's all-|+⟩ environment to test whether intra-chain ZZ coupling really was contaminating the reference. The preregistered qualitative prediction was excess χ > 0 for every qubit, larger at two-neighbour interior sites than at one-neighbour endpoints. Qubit chain neighbours mean excess χ (|+⟩ vs |0⟩ neighbours) q14 1 +0.0935 q33 1 +0.1269 q34 2 +0.1623 q35 2 +0.1718 q19 2 +0.1998 q15 2 +0.2105 Both parts confirmed. Every qubit shows positive excess, and the separation is complete — the smallest interior value (+0.1623) exceeds the largest endpoint value (+0.1269) with no overlap. Interior/endpoint ratio 1.69, against a naive neighbour-count prediction of 2.00. Significance 1. A construct-validity failure was caught by the pre-registration, not by hindsight. The pilot's ABSTAIN was a correctly-functioning gate firing on a badly-chosen criterion. Had the gate been calibrated differently, the run would have returned a confident REJECT of rate additivity that we now believe was largely an artifact of the reference environment. The frozen predicate prevented a wrong result from entering the record; it did not prevent a wrong design, and the difference is the lesson. 2. Neighbour-state-dependent dephasing excess, with the predicted spatial ordering. The diagnostic strongly supports the preregistered ZZ-contamination mechanism: all six sites showed positive excess dephasing with |+⟩ rather than |0⟩ neighbours, and every two-neighbour interior site exceeded both one-neighbour endpoints, with an interior/endpoint mean ratio of 1.69 against a naive neighbour-count prediction of 2.00. The diagnostic varied the neighbour state; it did not independently switch the physical ZZ coupling on and off. It therefore does not isolate ZZ from every possible neighbour-state-dependent crosstalk mechanism. The practical consequence holds regardless of which mechanism dominates: the all-|+⟩ product reference was not a matched control in this experiment, and such a reference can be invalid on a coupled chain unless neighbour-state-dependent effects are controlled. The correction is cheap — pin non-targets in the computational basis. 3. Under plain idling, the residual is much smaller than the pilot suggested. The k = 4 bare point estimate was near zero (median r = −0.014, χ²/dof = 0.62, p = 0.76), but its simultaneous confidence intervals did not satisfy the preregistered equivalence criterion. The result remains ABSTAIN: it indicates substantially less non-additivity than the pilot did, but it does not establish independence. The equivalence predicate exists precisely so that "we could not reject zero" cannot be reported as "we demonstrated independence," and that discipline is applied here to our own most favourable-looking number. 4. The non-additivity is sequence-dependent, and its mechanism is undetermined. This is the most important caveat in the record and it is stated prominently rather than buried. The justified statement is narrow: non-additivity was detected under the implemented XY4 arm at all three weights, while under bare evolution practical additivity was neither established nor rejected. The residual is sequence-dependent. We do not determine its origin. Candidate mechanisms include coherent pulse error, DD-filtered correlated or non-Markovian noise, interaction terms in the toggling frame, and other sequence-dependent effects. Dynamical decoupling changes the noise filter function, so correlated noise can legitimately manifest differently under XY4 than under bare idling. The absence of a comparable bare-arm rejection therefore does not isolate pulse error from DD-filtered noise correlation. One observation bears on the pulse-error hypothesis without settling it. The χ-additivity identity is derived for stochastic local phase noise; coherent pulse error is neither stochastic nor local, and the identity carries no guarantee for it. GHZ_k and |+⟩ also have different symmetry under the XY4 pulse group — for a 0.05 rad over-rotation, |+⟩ is exactly invariant under X errors and takes 6.3×10⁻⁴ infidelity from Y errors, while GHZ₆ takes 3.7×10⁻³ from both. This makes coherent pulse error a live hypothesis, not a demonstrated cause. One documented candidate deserves naming explicitly, because it is testable and because the design used here is vulnerable to it. Simultaneous π pulses on both ends of a coupled edge do not reverse the ZZ interaction. Since XZX = −Z, the two sign flips cancel: (X⊗X)(Z⊗Z)(X⊗X) = (XZX)⊗(XZX) = (−Z)⊗(−Z) = Z⊗Z [verified numerically] and identically for Y. Static ZZ therefore survives the simultaneous XY4 arm used in this experiment. Staggering the pulse timing between the two colours of a two-colourable topology creates intervals in which only one end of an edge is toggled: (X⊗I)(Z⊗Z)(X⊗I) = (XZX)⊗Z = −Z⊗Z [verified numerically] allowing the coupling to average toward zero. Crosstalk-robust sequences built on this principle are published and report at least a 3× improvement in fidelity decay rate on IBM hardware. Residual static ZZ under a non-robust sequence is therefore a concrete, documented candidate for the observed residual — still a candidate, not a demonstrated cause. Having just demonstrated why mechanism claims require dedicated controls, this record declines to make a third one. 5. A crosstalk-robust DD arm is the natural next experiment. Repeating this design with a staggered, crosstalk-robust XY4 as a third arm — using the same two-colouring already present for the localA/localB controls, since the chain is two-colourable by construction — would test whether the residual is driven by static ZZ surviving the simultaneous sequence. Interpretation should remain conservative: substantial collapse of the residual under the staggered sequence would support static-ZZ sensitivity of the simultaneous sequence as a contributor; it would not establish ZZ as the only mechanism. A cycle-count series (1, 2, 4 cycles) remains available as secondary characterisation, but it is not a clean discriminator, because both coherent pulse error and DD-filtered correlated noise can vary with cycle count as the filter function changes. Relation to prior work No novelty claim is made. Every physical ingredient in this record is published, and several precedents sit close to this experiment. This positioning rests on a targeted search rather than a systematic review, and no "first to" claim should be read into it. Layer 1 — known: GHZ coherence scaling diagnoses noise correlation. Ozaeta and McMahon measured GHZ coherence for N = 1…8 on IBM ibmqx5, fitted exponential decays, and found the decay rate approximately linear in N, consistent with effectively uncorrelated qubit noise (arXiv:1712.07080; Quantum Sci. Technol., 2019). Monz et al. observed approximately N² scaling in a 14-qubit trapped-ion register, consistent with correlated Gaussian phase noise (arXiv:1009.6126; Phys. Rev. Lett., 2011). That GHZ scaling can depart from independent-qubit expectations in either direction is long established. Layer 2 — known: spectator state alters single-qubit coherence via residual ZZ. Tripathi et al. prepared a main qubit in |+⟩, varied spectator states among |0⟩, |1⟩, and |+⟩, and observed spectator-state-dependent Ramsey behaviour attributable to ZZ coupling on IBM superconducting processors, with the differences largely removed by DD applied to spectators (arXiv:2108.04530; Phys. Rev. Applied 18, 024068, 2022). The physical effect underlying this record's diagnostic is prior art and is not claimed as new. Layer 3 — related Heron result, using different observables. Cerezo García compared three-qubit GHZ states against |+++⟩ product states under idle evolution on three Heron-generation processors (ibm_torino, ibm_fez, ibm_boston), reporting the product state decaying 2.4–13.8× faster than the GHZ metric (Zenodo 10.5281/zenodo.18274447, January 2026), with an extended version adding dynamical decoupling (Zenodo 10.5281/zenodo.18284373). Presented as a preprint / technical note; journal peer-review status not verified. Different devices from ibm_kingston used here. The comparison is relevant but not equivalent. That work's GHZ quantity is reported as P(000) + P(111), a computational-subspace population metric, whereas the product state is rotated back and read out phase-sensitively. A fully phase-dephased mixture ½|000⟩⟨000| + ½|111⟩⟨111| retains essentially full population inside that subspace, so P(000) + P(111) is not a direct measure of the |000⟩–|111⟩ off-diagonal coherence. EC-MECH measures that coherence explicitly, through parity-oscillation amplitude. A matched-observable replication would be required before the two results could be placed on the same axis. Layer 3b — known: simultaneous-pulse DD does not decouple static ZZ. Hickman, Wu, and Quiroz introduced a crosstalk-robust DD protocol that modifies pulse timing on two-colourable topologies specifically to suppress static ZZ under bounded control, reporting at least a 3× improvement in fidelity decay rate on IBM devices (Phys. Rev. Applied 25, 064041, 2026). The XY4 arm used here is the non-robust, simultaneous-pulse variant, which is a documented candidate mechanism for the DD-arm residual reported above (see Significance §4). Layer 4 — what this record adds. A preregistered demonstration that an all-|+⟩ product reference can be a construct-invalid comparator for GHZ coherence unless neighbour environments are matched, followed by a pointwise, decay-law-free χ-additivity test under matched environments, and a separately preregistered diagnostic of the mismatch itself. The specific combination is: matched |0⟩ neighbours in the primary control; a separately frozen all-|+⟩ diagnostic family excluded from the predicate; the one-neighbour versus two-neighbour spatial ordering as a preregistered qualitative prediction; and the whole thing connected prospectively to a documented failed first experiment. A hypothesis this raises, stated as such. Published GHZ-versus-product comparisons on this hardware family use an all-|+⟩ reference of the kind the diagnostic here shows carries neighbour-state-dependent excess dephasing at every site, while the matched-spectator bare arm at k = 4 lands near additivity (median r = −0.014). The present result motivates testing whether reference-environment effects and observable choice contribute to previously reported GHZ-versus-product performance gaps on fixed-coupled hardware. This record does not claim that any published result is an artifact: the protocol here was not a replication, and differed in weights, devices, and — importantly — in the GHZ observable itself. It identifies control variables that a matched-reference, matched-observable replication could test directly. What is NOT claimed Not that noise independence is violated on ibm_kingston. The DD-arm rejection is sequence-dependent and its mechanism is undetermined. See §4. Not that the DD residual is caused by coherent pulse error. That is one live hypothesis among several and is not isolated by this experiment. Not that intra-block ZZ is the unique cause of the diagnostic excess. The diagnostic varied neighbour state, not the coupling itself. Not that coherence-function additivity holds under plain idling. The bare arm ABSTAINed; a positive equivalence claim was not supported. Not that EC-STORAGE-001's weight-4 mechanism is confirmed or repaired. GHZ_k coherence is not the encrypted-cloning payload, which is a weight-4 stabilizer correlation inside a different encoding. Any transfer is an inference, not a measurement. Not a device-comparison result. One device, one chain, one calibration epoch. Not a novel physical phenomenon. See Relation to prior work. Not anything bearing on the no-cloning theorem. No operation performed here copies an unknown state, and nothing in this record weakens, tests, or circumvents no-cloning. The connection to EC-STORAGE-001 is motivational: quantum information is protected by delocalising it into multi-qubit correlations because it cannot be copied, and this record concerns how such correlations decay. Not a characterisation of dynamical decoupling in general. A single XY4 cycle spanning up to 45 µs is coarse decoupling. Pre-registration audit Both experiments were frozen before execution: a SHA-256 digest covering the analysis code, the predicate document, the tolerance bands, the ABSTAIN gates, the qubit-selection policy, and the stopping rules. The digest binds the decision function, not the outcome. EC-MECH-001 predicate digest c89e316ca8f5a1ab98443f5da901a80999a50b3eee5b86192da06cbd14fe4aa7 (frozen 2026-09-03T00:40:37Z; job dacc5flnj4cs73ace6fg) EC-MECH-002 predicate digest d22feceecf93aadec4a09301fabbc73242dc13690c3fae7d16b2bfd4302e3795 (frozen 2026-09-03T02:50:44Z; job dace2pm42tqs73as21o0) ec_mech_002.py — f86fab426058a871451275df2d3f55f0a4f371ea706333578fdfe9de6fb4410b ec_mech_002_preregistration.md — 9bb4d0747876f7e5e7556f23867bc1c0d507cec2e2035e2b77c7f0d9c5641856 freeze manifest — 41eabdff568dff5aeb19e59fa2173cce58dd267b6be919b73bc530d9b48d4813 git commit d697020 — Freeze EC-MECH-002 coherence-additivity experiment Binding layout. The qubit chain was selected post-freeze by the frozen selector on the first successful layout search: 536 chains enumerated, 522 qualified, [14, 15, 19, 35, 34, 33] selected with no manual override. The selection ran against the same-day calibration snapshot deposited with this record. Misses and errors recorded: EC-MECH-001's R² ≥ 0.90 admissibility gate was inappropriate for the shallow-decay regime the experiment operated in. This was diagnosable from the calibration snapshot before execution — reported T₂ of 217–477 µs against a 45 µs window implies ~9–12% total excursion — and was not diagnosed. The pre-submission power analysis modelled binomial shot noise only and therefore predicted a 100% fit success rate for precisely the fits that failed. EC-MECH-001's product-family reference did not share the GHZ observable's noise environment (the ZZ asymmetry). This is a construct-validity error, not a statistical one. The EC-STORAGE-001 weight-4 sum-of-rates explanation is entered as UNSUPPORTED (see Background). EC-MECH-002 is pilot-informed, not blind. Its τ grid, informative region, and expected effect size were chosen using EC-MECH-001 observations. It is a confirmatory test of a pilot-generated hypothesis and is weaker on that axis than a fully blind pre-registration. Declared in the frozen document, not added after. No gate, margin, τ grid, or tolerance band was altered after data was seen. The EC-MECH-001 ABSTAIN is deposited as ABSTAIN. Evidence ceiling Single session, single calibration epoch, single 6-qubit chain, single device. No readout-error mitigation. The τ = 0 normalisation removes τ-independent contrast loss but not τ-dependent readout drift. The three qubits within each two-colour control family share circuits; their marginals are treated as independent when they are only approximately so. The DD arm carries ~4 extra single-qubit pulses per qubit per cycle, so its wall-clock idle exceeds the bare arm's by ~0.1–0.2 µs at the longest τ. Not compensated. Pinned |0⟩ neighbours contribute a static ZZ frequency shift, assumed non-dephasing. Not independently verified here. The ZZ diagnostic samples only three of the eight primary τ points and is therefore a coarser mechanism probe, although all three coincide exactly with primary-grid τ values and the analyser performs no interpolation. k = 2 has the smallest denominator and correspondingly the largest sensitivity to bias in D. Uncertainties are within-session fit uncertainties. No session-to-session error bars exist. The informative τ grid is non-uniform and clustered. Five of the eight points lie within 18–28 µs, a window covering 30% of the 12–45 µs span. The eight Bonferroni-corrected tests are therefore not eight independent probes of the τ range; adjacent points interrogate nearly the same physics. Bonferroni remains conservative in this situation and cannot manufacture a false REJECT, but the secondary χ² statistic takes dof = n_usable, which assumes independence beyond the explicitly modelled shared-C(0) term. Any slow systematic drift correlating neighbouring τ would inflate χ². Randomised execution order under seed 20260902 is the mitigation, and it decorrelates drift from τ, but the χ² dof should still be read as an upper bound on the effective number of independent constraints. The grid's lower bound of 12 µs was chosen from pilot observations. Behaviour of Δ(τ) below 12 µs is unmeasured in this record. Files in this deposit File Contents ec_mech_001.py Pilot driver: freeze / validate / layout / estimate / submit / analyze ec_mech_001_preregistration.md Frozen predicate, gates, stopping rules, ceiling ec_mech_001_diagnose.py Read-only post-hoc diagnostic; does not amend the verdict ec_mech_002.py Successor driver, χ-additivity predicate with covariance handling ec_mech_002_preregistration.md Frozen predicate, equivalence margin, gates, ceiling EC-MECH-001_freeze_*.json Pilot predicate digest and manifest EC-MECH-002_freeze_*.json Successor predicate digest and manifest EC-MECH-00*_job_*.json Submission metadata, circuit plan, calibration snapshot EC-MECH-00*_results_*.json Full coherence curves, covariances, verdicts, diagnostic ibm_kingston_cal_*.json Same-day calibration snapshots at submission time SHA256SUMS Manifest over all deposited files Both drivers include an offline --validate mode requiring no network and no QPU: noiseless circuit algebra plus, for EC-MECH-002, an end-to-end statistical round-trip against synthetic data with known ground truth (independent noise → mean r = −0.0053, n = 200; GHZ protected 2× → −0.4985, n = 200; GHZ degraded 1.5× → +0.4727, n = 192; drift robustness check max |r| = 0.1036). Verification and reuse Two paths, deliberately distinguished. Reanalysis of the deposited campaign does not require QPU time and does not create a new freeze. Offline verification of the deposited record: sha256sum ec_mech_002.py ec_mech_002_preregistration.md # match the digests above python ec_mech_002.py --validate # no network, no QPU New execution of the same design (a new experiment, not a reproduction of this one): the driver binds submission to the freeze manifest and to the binding layout report; the qubit chain is taken from that report rather than typed in, so the selection policy cannot be bypassed by hand. Consult --help in the deposited driver for the exact invocation. Generating a fresh freeze and submitting new hardware jobs produces a new record and must not be presented as a reproduction of dacc5flnj4cs73ace6fg or dace2pm42tqs73as21o0. Both job IDs are recorded for independent verification against IBM Quantum records. References Quantum-Clarity LLC, EC-STORAGE-001: Encrypted-clone storage lifetime on ibm_kingston, Zenodo (2026). DOI 10.5281/zenodo.21299091. A. Ozaeta and P. L. McMahon, Decoherence of up to 8-qubit entangled states in a 16-qubit superconducting quantum processor, arXiv:1712.07080; Quantum Sci. Technol. (2019). T. Monz et al., 14-qubit entanglement: creation and coherence, arXiv:1009.6126; Phys. Rev. Lett. (2011). A. Cerezo García, Comparative Coherence Decay of GHZ and Product States Under Idle Noise in IBM Heron Quantum Processors, Zenodo (2026). DOI 10.5281/zenodo.18274447. Extended analysis with dynamical decoupling: DOI 10.5281/zenodo.18284373. V. Tripathi, H. Chen, M. Khezri, K.-W. Yip, E. M. Levenson-Falk, and D. A. Lidar, Suppression of crosstalk in superconducting qubits using dynamical decoupling, arXiv:2108.04530; Phys. Rev. Applied 18, 024068 (2022). E. Hickman, X. Wu, and G. Quiroz, Crosstalk-robust dynamical decoupling for bipartite-topology quantum processors, Phys. Rev. Applied 25, 064041 (2026). DOI 10.1103/r75m-b2d9. Qiskit contributors, Qiskit: An open-source framework for quantum computing (qiskit.org). Experiments executed via Qiskit and Qiskit IBM Runtime. Acknowledgments and disclosure Executed on IBM Quantum services (ibm_kingston). The views expressed are the author's and do not reflect the official policy or position of IBM or the IBM Quantum team. This record contains no proprietary QuantaCore™ components; the deposited code is fully self-contained and uses only public Qiskit interfaces. QUANTACORE™ and EIGENSPECTRUM™ are trademarks of Quantum-Clarity LLC; patent(s) pending. License: CC BY 4.0 Citation @dataset{quantum_clarity_ecmech_2026, author = {{Quantum-Clarity LLC}}, title = {{Preregistered Tests of Local-Noise Predictions for GHZ Coherence Under Matched Qubit Environments on IBM Kingston}}, year = 2026, month = sep, publisher = {Zenodo}, doi = {10.5281/zenodo.22283967}, url = {https://doi.org/10.5281/zenodo.22283967} }