Abstract Several papers on physics-informed noise modeling for superconducting qubits report gains from embedding known physics into machine-learning models, but we found little discussion in that literature of the specific mechanism by which such gains arise, or of the conditions under which embedding a known relation could instead hurt. We do not claim this gap is unaddressed anywhere in the broader machine-learning literature — bias-variance behavior of physics-informed models and the importance of multi-seed evaluation are both studied elsewhere in adjacent settings, discussed in Section 2 — only that we did not find this specific mechanism characterized for noisy-input extrapolation in the quantum-noise-modeling papers we reviewed. We derive a leading-order (small-σ) bias-variance decomposition for anchoring a regression model to an exact analytic noise relation (here, the T1-relaxation-limited dephasing rate 1/(2T1)): to leading order in the relative input-noise magnitude σ, the exact plug-in estimator has a small positive relative bias of σ² and a variance scaling with the square of the anchored quantity's absolute magnitude. We validate both predictions in a controlled single-variable Monte Carlo experiment: the simulated relative bias (0.65%) matched the leading-order prediction (0.64%) closely, and the simulated variance ratio between two test points matched its prediction to within about 1%. We then test whether this single-variable picture transfers to a more realistic two-feature extrapolation task — predicting the excess, non-relaxation-limited dephasing rate of a held-out device from noisy T1/T2 telemetry — using calibration parameters grounded in five real, independently re-verified IBM Heron/Eagle/Osprey processor snapshots, including a cross-generation transfer test (training on three Heron-class devices, T1 ≈ 137–189µs, evaluating on both a lower-coherence Osprey-class device, T1 ≈ 88µs, and a higher-coherence Eagle-class device, T1 ≈ 263µs). Run once, the isolated single-variable picture does not predict the sign of the effect in this more realistic setting: physics-anchoring loses to a plain ridge baseline on the low-T1 device. Repeating the experiment across 20 independent random seeds, with a one-sample t-test against zero, shows why: the low-T1 effect is statistically indistinguishable from zero (mean − 2.5%, 95% CI [− 10.5%, + 5.4%], t(19) = − 0.67, p = 0.51, 12 of 20 seeds favoring physics-informed), while the high-T1 benefit is statistically robust (mean + 19.7%, 95% CI [+ 16.7%, + 22.7%], t(19) = 13.7, p < 0.0001, 20 of 20 seeds favoring physics-informed). A follow-up collinearity-manipulation experiment, itself repeated across 15 seeds per condition, initially appeared non-significant when analyzed by pooling all seed-condition pairs into one regression; a corrected per-seed-slope analysis, accounting for the repeated-measures structure induced by reusing seeds across jitter levels, instead finds a small but statistically significant positive trend at both held-out devices (p = 0.021 and p = 0.011). All experiments are grounded in real, independently re-verified device numbers but built from synthetic telemetry — documented in full, equation by equation, in Section 6 — in a network-isolated environment without access to live quantum hardware; results should be read as a methodological, cautionary study on synthetic data, not a validated hardware finding, and we make no claim that any of the specific numeric effect sizes would reproduce on real devices. We report this honestly, including a citation error we found and fixed in our own earlier draft and a headline finding (the low-T1 reversal) that did not survive statistical testing, because both are more useful to a careful reader than a clean story would have been. Complete reproduction code is included with this submission.
Abstract Several papers on physics-informed noise modeling for superconducting qubits report gains from embedding known physics into machine-learning models, but we found little discussion in that literature of the specific mechanism by which such gains arise, or of the conditions under which embedding a known relation could instead hurt. We do not claim this gap is unaddressed anywhere in the broader machine-learning literature — bias-variance behavior of physics-informed models and the importance of multi-seed evaluation are both studied elsewhere in adjacent settings, discussed in Section 2 — only that we did not find this specific mechanism characterized for noisy-input extrapolation in the quantum-noise-modeling papers we reviewed. We derive a leading-order (small-σ) bias-variance decomposition for anchoring a regression model to an exact analytic noise relation (here, the T1-relaxation-limited dephasing rate 1/(2T1)): to leading order in the relative input-noise magnitude σ, the exact plug-in estimator has a small positive relative bias of σ² and a variance scaling with the square of the anchored quantity's absolute magnitude. We validate both predictions in a controlled single-variable Monte Carlo experiment: the simulated relative bias (0.65%) matched the leading-order prediction (0.64%) closely, and the simulated variance ratio between two test points matched its prediction to within about 1%. We then test whether this single-variable picture transfers to a more realistic two-feature extrapolation task — predicting the excess, non-relaxation-limited dephasing rate of a held-out device from noisy T1/T2 telemetry — using calibration parameters grounded in five real, independently re-verified IBM Heron/Eagle/Osprey processor snapshots, including a cross-generation transfer test (training on three Heron-class devices, T1 ≈ 137–189µs, evaluating on both a lower-coherence Osprey-class device, T1 ≈ 88µs, and a higher-coherence Eagle-class device, T1 ≈ 263µs). Run once, the isolated single-variable picture does not predict the sign of the effect in this more realistic setting: physics-anchoring loses to a plain ridge baseline on the low-T1 device. Repeating the experiment across 20 independent random seeds, with a one-sample t-test against zero, shows why: the low-T1 effect is statistically indistinguishable from zero (mean − 2.5%, 95% CI [− 10.5%, + 5.4%], t(19) = − 0.67, p = 0.51, 12 of 20 seeds favoring physics-informed), while the high-T1 benefit is statistically robust (mean + 19.7%, 95% CI [+ 16.7%, + 22.7%], t(19) = 13.7, p < 0.0001, 20 of 20 seeds favoring physics-informed). A follow-up collinearity-manipulation experiment, itself repeated across 15 seeds per condition, initially appeared non-significant when analyzed by pooling all seed-condition pairs into one regression; a corrected per-seed-slope analysis, accounting for the repeated-measures structure induced by reusing seeds across jitter levels, instead finds a small but statistically significant positive trend at both held-out devices (p = 0.021 and p = 0.011). All experiments are grounded in real, independently re-verified device numbers but built from synthetic telemetry — documented in full, equation by equation, in Section 6 — in a network-isolated environment without access to live quantum hardware; results should be read as a methodological, cautionary study on synthetic data, not a validated hardware finding, and we make no claim that any of the specific numeric effect sizes would reproduce on real devices. We report this honestly, including a citation error we found and fixed in our own earlier draft and a headline finding (the low-T1 reversal) that did not survive statistical testing, because both are more useful to a careful reader than a clean story would have been. Complete reproduction code is included with this submission.
Qubit layout selection determining which physical qubits a logical circuit occupies on a superconducting processor is fundamentally a noise-field sampling problem. On IBM Heron r1 and Heron r2 devices, two-qubit gate error varies by nearly one order of magnitude (7– $9\times $ ) across qubit pairs on the same chip, causing measured fidelity differences up to 0.903 from layout choice alone. We introduce the Gate Noise Sensitivity Index (GNSI), a calibration-weighted linear combination of per-qubit readout error and per-edge two-qubit gate error, derived analytically as a first-order approximation of circuit fidelity under independent, heterogeneous noise channels. For 12 backend–circuit combinations on three IBM processors (133-qubit Heron r1 and 156-qubit Heron r2), GNSI scores are computed for the full pool of 1,320 computationally enumerated candidate layouts; the top-ranked (GNSI-BEST) and bottom-ranked (GNSI-WORST) layouts are then executed on hardware. GNSI-BEST ranks first by score in all 12 instances (Hit@1 = 100% within the enumerated candidate pool) and consistently achieves higher measured fidelity than GNSI-WORST, with Spearman rank correlation ρ=0.926±0.016 and statistically significant fidelity separation in 9 of 12 paired tests. Initial hardware comparisons on ibm_fez ( n=5 ) show GNSI-BEST matching or exceeding SABRE-selected fidelity. We further identify a depth-dependent regime transition: gate-error heterogeneity appears to govern ranking in shallow circuits (depth ≤15 ), while qubit-specific T1 variation may dominate for deeper circuits (depth ≥20 ). The analytic regime boundary d∗(b) , computable from calibration data alone, shows agreement within ±2 depth layers across all three backends in initial comparisons; GNSI-Extended incorporating the decoherence term receives an initial hardware comparison on two circuits, with full validation left to future work. These results suggest that layout-dependent fidelity variation may be governed by a calibration-predictable noise field, with direct implications for compiler design and noise-adaptive quantum circuit execution.
I examine whether a noisy fixed-parameter quantum front end preserves class information from input data before variational optimization. Only the input vector x is encoded; the label is used after circuit execution to estimate training distributions and score disjoint held-out inputs. The main density-matrix grid contains 10,800 directly simulated configurations over two binary tasks, widths n∈{2,3,4} , every integer depth L=1,…,12 , five depolarizing strengths, and 30 circuit seeds. At n=4 , mean held-out mutual information has a finite-depth maximum that shifts from L=6 to 5 for Gaussian data and from L=7 to 6 for the Wisconsin Diagnostic Breast Cancer (WDBC) benchmark as p increases from 0 to 0.04. In five-fold complete-seed validation, including training-distribution mutual information with declared design variables raises held-out-accuracy R2 from 0.724 to 0.961 for Gaussian data and from 0.714 to 0.860 for WDBC. Training total variation is the stronger binary comparator in complete-seed validation; test-set total variation is reported only as a non-deployable post-outcome oracle. Paired seed-level RMSE improvements from mutual information are 0.0185 (95% bootstrap interval 0.0179–0.0190) and 0.00634 (0.00592–0.00675), and finite-shot training estimates retain most of that gain. Twenty independently seeded Gaussian realizations, twenty repeated WDBC partitions, per-input 256-shot tests, and explicit rotation-schedule ablations expose both transfer and limitations. A family-matched damped-alignment response outperforms an equal-parameter-count polynomial comparator under identical grouped folds but transfers weakly to some widths, encodings, and schedules. Explicit local-folding zero-noise extrapolation and tensor-product measurement-error mitigation improve information and held-out likelihood accuracy in selected simulated budgets; small-cluster sensitivity and failure cases are retained. The accompanying reproducibility archive contains the complete source, raw counts, parameter registries, 2,000-replicate bootstrap vectors, mitigation replicates, automated tests, and resumable one-click workflow. I make no hardware, trained-VQNN, clinical, quantum-advantage, or universal-scaling claim.
Quantum machine learning (QML) is increasingly proposed for predictive analytics in agriculture, yet two operational questions remain underexplored: platform reproducibility across open-source software-development kits (SDKs), and robustness under climate-distribution shift. This paper contributes a methodological reproducibility benchmark, not an agricultural result: a synthetic agronomic-suitability classification task on which a Quantum Support Vector Machine (QSVM) is trained identically across three SDKs (Qiskit, Cirq, PennyLane) with a fixed feature map. Across 40 independently generated train/test instances at the primary n = 60 operating point, all three SDKs produce identical predictions at every evaluated seed and regime; Qiskit and PennyLane agree at the double-precision floor and Cirq’s default kernel differs at approximately 1.8 × 10 − 7 , a gap that collapses under forced double precision and that an SDK-free, exact analytic evaluation of the same circuit reproduces to the same floor, directly showing that the present four-qubit circuit carries no computational quantum advantage. Under repeated climate-shift evaluation, the QSVM shows no statistically significant accuracy change under drought or extreme heat after Holm correction; under wet conditions, the QSVM and several non-linear classical baselines show a significant accuracy gain , with balanced accuracy and AUROC rising alongside raw accuracy, that moves opposite to a significant majority-class-baseline drop. Test-set class prevalence shifts sharply under wet, but prevalence change alone does not account for the gain, which is shared across several non-linear classical models and is therefore not specific to the QSVM. Both the cross-SDK parity finding and the general shape of the shift result replicate under two structurally different synthetic generators (correlated features; an alternative noisy decision rule), though the wet-regime effect weakens under correlated sampling. Three real-data anchors reproduce the same numerical pattern, with one isolated prediction-level exception. These results are limited to a synthetic benchmark and noise-free statevector simulators; they establish cross-SDK numerical reproducibility and controlled shift behaviour, not quantum advantage or field-level crop-yield validity.
We theoretically study the dephasing of an Andreev spin qubit (ASQ) due to electric and magnetic noise. Using a tight-binding model, we calculate the Andreev states formed in a Josephson junction where the link is a semiconductor with strong spin-orbit interaction. As a result of both the spin-orbit interaction and induced superconductivity, the local charge and spin of these states vary as a function of externally controllable parameters: the phase difference between the superconducting leads, an applied magnetic field, and filling of the underlying semiconductor. Concomitantly, coupling to fluctuations of the electric or magnetic environment will vary, which informs the rate of dephasing. We qualitatively predict the dependence of dephasing on the nature of the environment, magnetic field, phase difference between the junction, and filling of the semiconductor. Comparing the simulated electric- and magnetic-noise-induced dephasing rate to experiment suggests that the dominant source of noise is magnetic. Moreover, by appropriately tuning these external parameters, we find sweet spots at which we predict an enhancement in ASQ coherence times.
Note: description expanded on 2026-10-14 for readability. It adds no new predictions or criteria. The frozen pre-registration is the attached PDF, unchanged since publication. Status: the experiment was run on 2026-10-14, after this record was published. Results, including every pre-registered test, will be reported in a separate record linked here. In plain language Theory says a quantum system never completely forgets where it started, even when its dynamics are chaotic. This leftover memory is called a quantum birthmark (Graf et al., Phys. Rev. X, 2026). Theory also says the birthmark should be larger when the system's rules run the same forwards and backwards in time (time-reversal symmetry) than when that symmetry is broken. This record pre-registers a test of that prediction on IBM's ibm_kingston quantum processor. We built two nearly identical sets of 4-qubit circuits. One set has time-reversal symmetry and the other does not. Both use exactly the same gates, so they experience the same hardware noise. We measure how often each circuit returns to its starting state, compared with the 1-in-16 chance level, and predict that the symmetric set shows about 1.67× more memory than the broken set. All predictions, pass/fail criteria, circuits and analysis code were deposited here before any quantum hardware was used. What is tested, and what is not Tested: whether the return probability stays above the ergodic baseline 1/16 at depths 8–23, and whether that excess is larger for the time-reversal-symmetric class by the ratio exact numerics predict for these fixed instances. Not tested: the withdrawn temporal-memory claims of 10.5281/zenodo.20470129. The birthmark is closed-system coherent memory, which measurement and reset erase, so no outcome here bears on those claims. Not tested: the universal random-matrix value 1.889. At this size (N = 16) the circuit family itself gives about 1.66, so a finite-size difference is expected. Not a quantum-advantage claim: a 16-dimensional system is exactly simulable. This is a hardware-validation experiment. Design Hardware: ibm_kingston, physical qubits 108–109–118–129 as an open chain, selected by a no-QPU calibration audit (included as v6_backend_audit.zip). Circuits: a kicked-Ising Floquet period Kx(a/2) → D(h1) → Kx(b) → D(h2) → Kx(a/2), with D(h) = exp(−iΣhiZi)·exp(−iπ/4 ΣZiZj), at exactly 6 CZ per period. COE (symmetric): h1 = h2. TRSB (broken): h1 = h + d, h2 = h − d. The classes differ only in virtual-Z angles. Instances: 16 seeded COE/TRSB pairs (seed 20261013), run at every depth from 8 to 23 (48–138 CZ). Observable: P(t) = |⟨0000|Ft|0000⟩|². The primary statistic is the readout-mitigated excess ratio R = ECOE / ETRSB, where E is the mean of P(t) − 1/16. A noise fidelity common to both classes cancels in R. Controls: 32 deep circuits (depths 60–63) as a decoherence-floor check. 8 state-preparation monitors. 16 readout-calibration circuits. Run: one SamplerV2 job of 568 circuits at 256 shots each, with dynamical decoupling and twirling off and the execution order shuffled. Predictions Quantity COE TRSB R Excess over 1/16, depths 8–23 (noiseless, primary) 0.0896 0.0537 1.667 Same instances, infinite-time limit 0.0952 0.0634 1.502 Circuit family, 10,000 random draws — — 1.661 Random-matrix reference, N = 16 — — 1.889 Under four simulated noise models (nominal, pessimistic, and each with coherent errors), the mitigated R stays between 1.64 and 1.77. The shot-noise standard deviation of R at 256 shots is 0.070–0.125. Pre-registered decision criteria Test Pass condition H1 (primary): symmetry dependence 99% CI lower bound of R > 1.0 H2 (primary): quantitative agreement 99% CI of R intersects 1.54–1.87 and is at most 1.0 wide S1: birthmark above decohered floor Window excess minus depth 60–63 excess, 99% CI lower bound > 0, for both classes S2: matched fidelity between classes 0.85 ≤ FCOE/FTRSB ≤ 1.15 S3: instance-level agreement Pearson r ≥ 0.80 against exact predictions S4 (flag only): state prep and drift Monitors ≥ 0.95, drift ≤ 0.045 Overall verdict: CONFIRMED if H1, H2 and S2 all pass; PARTIAL if H1 passes but H2 or S2 fails; NOT CONFIRMED if H1 fails. Every test is reported whatever the outcome. Uncertainty comes from a 10,000-sample parametric multinomial bootstrap over every circuit's counts, with instances held fixed. Amendment 1 (v1.1.0, made before any data) Amendment 1 changed two things relative to v1.0.0 (10.5281/zenodo.23355804): Shots: reduced from 1024 to 256 per circuit for QPU budget, cutting estimated execution from about 152 s to about 38 s. H2: changed from "point estimate in 1.54–1.87" to the CI rule above, because at 256 shots the point rule would falsely fail 20–25% of the time under the pessimistic models. A CI-width cap of 1.0 keeps H2 informative. Circuits, instances and predictions are unchanged. The trade-off is that H2 is now a weaker test. Provenance Every reported number is derived by script from raw counts: Source and transpiled circuits are saved as QPY. Before submission, every transpiled circuit is mapped back to 4 qubits and its unitary checked against the exact model to < 10−8. The analysis reads its decision bands only from predictions.json and its data only from the raw results file. Submission is gated on the DOI of this record. Files: The PDF is the authoritative pre-registration. predictions.json holds every predicted number and decision band. birthmark_prereg_v1.1.0.zip contains the circuit builder, verification, submission and frozen analysis scripts, plus simulated-data self-tests. v6_backend_audit.zip is the pre-submission calibration audit. Known limitations Small system size (N = 16), so the random-matrix value is not expected. The prediction is specific to depths 8–23 and to these 16 fixed instances. The noise models exclude leakage, TLS fluctuations and crosstalk from outside the chain. T1 relaxation biases P(0000) upward, which pulls R toward 1 (conservative). One backend, one path, one job: no replication. Data: CC BY 4.0. Code: MIT. Principal investigator: Amit Brahmbhatt, Quantum-Clarity LLC.
Abstract We benchmark the decoherence of superconducting transmon qubits to examine the temporal stability of energy relaxation, dephasing, and qubit transition frequency. By collecting statistics during measurements spanning multiple days, we find the mean parameters T1 T 1 ¯ = 49 μs and T2∗ T 2 * ¯ = 95 μs; however, both of these quantities fluctuate, explaining the need for frequent re-calibration in qubit setups. Our main finding is that fluctuations in qubit relaxation are local to the qubit and are caused by instabilities of near-resonant two-level-systems (TLS). Through statistical analysis, we determine sub-millihertz switching rates of these TLS and observe the coherent coupling between an individual TLS and a transmon qubit. Finally, we find evidence that the qubit’s frequency stability produces a 0.8 ms limit on the pure dephasing which we also observe. These findings raise the need for performing qubit metrology to examine the reproducibility of qubit parameters, where these fluctuations could affect qubit gate fidelity.