Full-Curve Validation of the Frauchiger–Renner Joint-Outcome Probability on a 156-Qubit Superconducting Quantum Processor

Full-Range Validation of the Frauchiger–Renner Joint-Outcome Probability on a 156-Qubit Superconducting Quantum Processor Amit Brahmbhatt Quantum-Clarity LLC Email: [e-mail] Date: April 2026 Platform: IBM Kingston (156-qubit Heron r3 heavy-hex superconducting processor) Companion methodology record: Zenodo DOI 10.5281/zenodo.19673112 Related infrastructure: Y⊗Z stabilizer basis migration (Zenodo DOI: 10.5281/zenodo.18498540; 10.5281/zenodo.19478241) Abstract We report a full-range experimental measurement of the Frauchiger–Renner joint-outcome probability P(ok, ok; θ) = cos²(θ/2)/4 on a 156-qubit superconducting quantum processor. The Frauchiger–Renner construction has previously been validated at canonical angle values on photonic platforms; the present work extends this validation across the full coin-preparation angle range θ ∈ [0, π] on IBM Kingston (Heron r3) using 10 angle values, each replicated across 10 parallel non-overlapping 4-qubit modules at 3000 shots per module, for an aggregated 30,000 shots per angle and 300,000 shots total in 81.91 seconds of billed quantum-processor time. The observed P(ok, ok; θ) tracks the analytic prediction with Pearson correlation 0.99955 and root-mean-square residual 0.0094 absolute. The full four-outcome joint distribution including the (fail, fail), (fail, ok), and (ok, fail) buckets follows the predicted angle dependence across the same range, with residual structure largely absorbed by the readout-error model described below. A two-parameter symmetric per-pair readout-error model with ε_A = 2.8% and ε_B = 3.4% absorbs the systematic structure of the residuals, reducing the root-mean-square residual on P(ok, ok) by a factor of 2.6 and bringing the post-correction residuals to within hardware-level fidelity expectations across the full curve. Per-module inter-chain variance tracks Poisson expectation across the angle range with mild excess at intermediate angles, indicating that the time-interleaved sampling protocol established in the companion methodology record adequately suppressed the patch-correlated drift artifacts identified in prior work. To our knowledge, this is the first reported full-range experimental measurement of the Frauchiger–Renner paradoxical-outcome probability on a superconducting platform at the 100+ qubit scale, parameterized in the coin-preparation angle, and characterizes the readout-fidelity systematics of the Wigner-basis joint readout as a function of the underlying state amplitudes. Plain-Language Summary The Frauchiger–Renner thought experiment is a famous result in the foundations of quantum mechanics. It describes a circuit that, when analyzed using ordinary quantum theory, produces a specific paradoxical outcome with a known small probability — at the standard angle setting, that probability is exactly 1 in 12. This work asks a more general question: across the full range of possible angle settings for the experiment, does a real quantum computer reproduce the entire predicted curve, not just the single canonical point? Using IBM's 156-qubit Kingston processor, we ran the Frauchiger–Renner circuit at 10 different angle settings, each repeated 10 times in parallel across different physical regions of the chip, for a total of 300,000 individual measurements completed in about 82 seconds of computing time. The observed paradoxical-outcome probability matches the theoretical curve to within about one percent over most of the angle range, with larger endpoint deviations attributable to readout-floor effects. The correlation coefficient between observation and prediction is 0.9996. This is the first time the full angle dependence has been measured on a superconducting quantum computer at this scale. The small remaining gap between observation and prediction is largely explained by an ordinary effect: the measurement step at the end of the circuit makes occasional bit-flip errors at a rate of about three percent per pair, consistent with the well-known readout fidelity of this class of hardware. Once that effect is taken into account, the residual disagreement drops by a factor of about 2.6 and falls well within hardware-level expectations, although a small residual structure remains that a more elaborate readout model could in principle further reduce. The technical machinery used in this experiment was developed and described in a separate methodology record from the same author. That earlier work showed that experiments of this kind can produce misleading apparent signals if the order in which measurements are submitted is not randomized. The present experiment uses the same drift-resistant submission protocol and confirms that under proper controls the quantum-mechanical prediction is recovered with high precision across a non-trivial functional form. 1. Introduction The Frauchiger–Renner construction (Frauchiger and Renner 2018) is a no-go theorem demonstrating that the universal validity of quantum mechanics, the assumption of single measurement outcomes, and the assumption of consistency between agents reasoning about each other's measurements cannot all jointly hold. The construction is implemented as a 4-system quantum circuit in which a coin-like initial state is propagated through two nested observer registers, with the paradoxical outcome appearing as a joint Wigner-basis measurement event whose probability is fixed by the quantum-mechanical formalism. The Frauchiger–Renner protocol has been experimentally implemented on photonic platforms (Bong et al. 2020; Proietti et al. 2019) at the canonical coin-preparation angle θ_c = 2·arccos(1/√3), with the observed paradoxical-event probability matching the predicted value of 1/12 at the level of a few percent. A natural extension of the canonical-point implementation is a parameterized scan across the coin-preparation angle θ. The joint-outcome probability is a non-trivial function of θ; reproducing the full predicted angle dependence provides a stronger test than reproducing a single point, because the curve shape constrains both the state-preparation accuracy and the joint-readout fidelity simultaneously across a range of state amplitudes. To our knowledge, no full-range measurement of the Frauchiger–Renner joint-outcome probability has been reported on a superconducting quantum processor. This paper reports such a measurement on the 156-qubit IBM Kingston (Heron r3) processor. The experiment uses 10 angle values spanning θ ∈ [0, π] and 10 parallel non-overlapping 4-qubit modules per angle, for an aggregated 30,000 shots per angle and 300,000 shots total. The hardware infrastructure (hardware-adaptive layout discovery, T1-floor filtering, geometric-mean fidelity scoring) and the experimental controls (randomized time-interleaving of circuit submission order) are reused directly from prior work by the author (Brahmbhatt 2026a, 2026b, 2026c). The methodological scaffolding is therefore not a contribution of the present work; the contribution is the full-range angle-dependence result and the readout-error characterization that emerges from the residual analysis. Section 2 derives the analytic prediction for P(ok, ok; θ) as implemented by the specific 4-qubit gate decomposition used in the experiment. Section 3 describes the experimental design. Section 4 presents the results. Section 5 introduces a two-parameter symmetric readout-error model that absorbs the systematic residual structure observed in the raw data. Section 6 reports per-module inter-chain variance as a drift-control diagnostic. Section 7 concludes. 2. Analytic Prediction for the Implemented Circuit The 4-qubit Frauchiger–Renner circuit implemented in this work assigns qubit roles as coin (q₀), friend F̄ (q₁), system S (q₂), and friend F (q₃). State preparation proceeds via Ry(θ) on q₀, followed by CX(q₀ → q₁), controlled-Hadamard CH(q₁ → q₂), and CX(q₂ → q₃). After state preparation but before Wigner-basis measurement, the 4-qubit state is |Ψ(θ)⟩ = c·|0000⟩ + (s/√2)·|1100⟩ + (s/√2)·|1111⟩ where c = cos(θ/2) and s = sin(θ/2). The propagation is verified by direct application of the gate sequence to the initial state |0000⟩: Ry(θ) on q₀ produces c|0⟩₀ + s|1⟩₀; CX(q₀ → q₁) entangles q₁ with q₀ giving c|00⟩ + s|11⟩ on (q₀, q₁); CH(q₁ → q₂) leaves the c|00⟩ branch unchanged (since q₁ = 0 makes CH the identity) and applies H to q₂ on the s|11⟩ branch, yielding (s/√2)|110⟩ + (s/√2)|111⟩ on (q₀, q₁, q₂); CX(q₂ → q₃) then maps |110⟩ → |1100⟩ and |111⟩ → |1111⟩ on the full 4-qubit register. The Wigner-basis measurement on each friend pair is implemented by CX followed by H on the leftmost qubit of the pair, reading the post-rotation computational outcome as ok = 1, fail = 0. This rotation maps the {ok, fail} basis on each pair to the post-rotation Z-eigenstates with the convention |ok⟩ = (|00⟩ − |11⟩)/√2 and |fail⟩ = (|00⟩ + |11⟩)/√2. Projecting |Ψ(θ)⟩ onto ⟨ok|₀₁ ⊗ ⟨ok|₂₃ gives an amplitude with three contributions: the c|0000⟩ component projects to c · (1/√2) · (1/√2) = c/2; the (s/√2)|1100⟩ component projects to (s/√2) · (−1/√2) · (1/√2) = −s/(2√2); and the (s/√2)|1111⟩ component projects to (s/√2) · (−1/√2) · (−1/√2) = +s/(2√2). The two s-dependent terms cancel exactly, leaving the total amplitude c/2 and the joint-outcome probability P(ok, ok; θ) = cos²(θ/2) / 4. Applying the same projection technique to the remaining three Wigner outcomes yields the full joint distribution: P(fail, ok; θ) = cos²(θ/2) / 4 P(ok, fail; θ) = [ cos(θ/2)/2 − sin(θ/2)/√2 ]² P(fail, fail; θ) = [ cos(θ/2)/2 + sin(θ/2)/√2 ]² These four probabilities sum to unity for all θ. At θ_c = 2·arccos(1/√3), cos²(θ/2) = 1/3, and the distribution evaluates to (3/4, 1/12, 1/12, 1/12), which matches both the canonical Frauchiger–Renner literature value and the result reported in the canonical-point baseline (Brahmbhatt 2026c) on the same processor. 3. Experimental Design The angle scan was implemented on IBM Kingston with the following design parameters. Ten coin-preparation angles were selected to span the full θ ∈ [0, π] range with denser sampling around the canonical point θ_c: θ ∈ {0, π/6, π/4, π/3, π/2, θ_c, 2π/3, 3π/4, 5π/6, π}. The canonical angle was retained explicitly to provide a direct numerical bridge to the prior baseline result. Ten non-overlapping 4-qubit modules were selected from the live Kingston coupling map by enumerating simple paths q₀–q₁–q₂–q₃ on the heavy-hex graph, scoring each path by the geometric mean of its three edge two-qubit gate fidelities, applying a T1-floor exclusion at 50% of the chip-mean T1, applying an edge blacklist derived from the outlier-module diagnostic in the prior baseline run (Brahmbhatt 2026c), and selecting the highest-scoring non-overlapping paths greedily. Selected module fidelity scores ranged from 0.9986 to 0.9990. The same 10 modules were used for all 10 angle values. Each (angle, module) circuit was sampled at 3000 shots, for a total of 100 distinct circuit configurations and 300,000 aggregated shots. The mitigation stack matched the prior baseline run: XY4 dynamical decoupling, Pauli twirling on both gates and measurements with 32 randomizations. Submission of the 100 circuits was randomized in a single permutation with fixed random seed prior to job creation, in accordance with the time-interleaving protocol established in the companion methodology record. Job execution required 81.91 seconds of authoritative billed QPU time as reported by the IBM Runtime portal export, against a wall-clock duration of 120.30 seconds inclusive of queue and result-serialization overhead. The authoritative IBM Runtime portal export was used to verify both the billed QPU time and the per-circuit raw outcome records underlying the aggregated angle-scan JSON. 4. Results Figure 1 (provided as a separate file in this deposit) displays the observed P(ok, ok; θ) values as points with statistical error bars, overlaid on the analytic Frauchiger–Renner prediction cos²(θ/2)/4 drawn as a solid curve and on the prediction with the symmetric readout-error correction of Section 5 applied (dashed curve, ε_A = 0.028, ε_B = 0.034). The observed points lie visibly between the two curves and land essentially on the readout-corrected curve at every angle. The canonical Frauchiger–Renner angle θ_c is marked with a dotted vertical line. The Pearson correlation between observed and predicted P(ok, ok) values across the 10 angle points is 0.99955, and the root-mean-square residual is 0.00936 absolute. The maximum absolute residual occurs at θ = π, where the observed probability of 0.0194 lies above the analytic prediction of zero — a deviation that is naturally interpretable as a readout-error floor and is treated quantitatively in Section 5. Per-angle statistical uncertainty is shot-noise-limited at 30,000 aggregated shots and is smaller than the plotted symbol size at most angles. Figure 2 (provided as a separate file in this deposit) displays the full four-outcome joint distribution in two panels. Panel (a) presents the observed joint probabilities as stacked bars at each of the 10 θ values, showing how the distribution shifts from approximately uniform at θ = 0 through the canonical (3/4, 1/12, 1/12, 1/12) configuration at θ_c and back to a bimodal (fail, fail)/(ok, fail) split at θ = π. Panel (b) overlays the observed per-bucket probabilities as points on the analytic curves for each of the four outcomes. The (fail, fail) outcome rises from 1/4 at θ = 0 to a peak of 3/4 near θ_c and then falls back to 1/2 at θ = π; the (ok, fail) outcome traces a deep well near θ = π/3 and rises sharply toward θ = π; the (fail, ok) and (ok, ok) outcomes both follow the cos²(θ/2)/4 shape. The observed values follow the analytic curves for all four outcomes, with the largest deviations appearing in the low-probability endpoint buckets. The per-θ summary table (Table 1) tabulates the observed P(ok, ok), the analytic prediction, the readout-corrected prediction of Section 5, and the residuals before and after correction. θ (deg) P_obs(ok,ok) P_QM(ok,ok) P_corr(ok,ok) Δ_raw Δ_corr 0.00 0.24713 0.25000 0.25000 −0.0029 −0.0029 30.00 0.22620 0.23325 0.22872 −0.0071 −0.0025 45.00 0.20930 0.21339 0.20785 −0.0041 +0.0014 60.00 0.18483 0.18750 0.18192 −0.0027 +0.0029 90.00 0.12590 0.12500 0.12215 +0.0009 +0.0037 109.47 0.08803 0.08333 0.08397 +0.0047 +0.0041 120.00 0.07083 0.06250 0.06542 +0.0083 +0.0054 135.00 0.04853 0.03661 0.04310 +0.0119 +0.0054 150.00 0.03037 0.01675 0.02693 +0.0136 +0.0034 180.00 0.01937 0.00000 0.01700 +0.0194 +0.0024 Table 1. Per-θ observed and predicted P(ok, ok), with raw residuals and residuals after the symmetric readout-error correction of Section 5. The raw residuals exhibit a systematic structure: the observed values lie below the analytic prediction at small θ, cross zero near θ = π/2, and lie progressively further above the prediction as θ approaches π. This sign pattern indicates that the observed distribution is being pulled toward the uniform distribution P = 1/4 across all four buckets, which is the diagnostic signature of symmetric bit-flip error in the Wigner-basis readout. 5. Readout-Error Characterization The systematic residual structure observed in Section 4 is well-modeled by independent symmetric bit-flip errors on the two friend pairs. Let ε_A denote the per-shot probability of bit-flip error in the readout of pair A = (q₀, q₁) and ε_B the corresponding probability for pair B = (q₂, q₃). The 4×4 confusion matrix mapping true joint outcomes to observed joint outcomes is the tensor product of the two single-pair confusion matrices, and the predicted observed probability for each bucket is the matrix-vector product of this confusion matrix with the analytic 4-vector. Fitting (ε_A, ε_B) by minimization of the total χ² across the 10 θ values and 4 buckets yields ε_A = 0.028 and ε_B = 0.034. The total χ² at this optimum is 125.5 across 38 degrees of freedom (4 buckets × 10 angles − 2 fit parameters), corresponding to a reduced χ² of 3.30. A constrained shared-ε fit yields ε = 0.032 with χ² = 157.2 across 39 degrees of freedom (reduced χ² = 4.03), demonstrating that allowing distinct readout-error rates for the two pairs improves the fit by approximately one σ-equivalent. The readout-corrected residuals are tabulated in the rightmost column of Table 1 and displayed graphically in Figure 3. Figure 3 (provided as a separate file in this deposit) displays the residuals on P(ok, ok) in two panels. Panel (a) shows the raw residual P_obs − P_QM as a function of θ. The raw residual is monotonically increasing with θ, crossing zero near θ = π/2 — the systematic sign pattern diagnostic of symmetric readout error pulling probabilities toward 1/4. Panel (b) shows the residual after applying the symmetric readout-error correction with ε_A = 0.028 and ε_B = 0.034. The post-correction residuals are reduced in magnitude by a factor of 2.6 in root-mean-square and exhibit no remaining sign-correlated trend with θ. Error bars on both panels are identical and reflect shot-noise-limited statistical uncertainty at 30,000 aggregated shots per angle. The root-mean-square residual on P(ok, ok) is reduced from 0.00936 (raw) to 0.00363 (after readout correction), an improvement factor of 2.58. The residual χ² of 3.30 indicates that the symmetric two-parameter model captures the dominant systematic but does not fully account for all hardware-level fidelity variation; remaining structure may reflect asymmetric bit-flip rates (ok→fail differing from fail→ok), per-qubit rather than per-pair variation in readout fidelity, or correlated errors between the two qubits within a single Wigner pair. A more elaborate readout-error model with these additional degrees of freedom could in principle further reduce the residual, at the cost of additional fit parameters; we do not pursue such a model here, as the simple symmetric per-pair model suffices to demonstrate that the systematic residual structure has a hardware origin rather than a deviation from the Frauchiger–Renner prediction. The fitted readout-error rates ε_A = 0.028 and ε_B = 0.034 are broadly consistent with the order of magnitude expected from single-qubit readout fidelities at the 1.5–2% level reported by the live Kingston backend target during the experiment, since a per-pair readout error is approximately twice the single-qubit error to first order. The mild asymmetry between ε_A and ε_B is itself plausible: ε_A averages over the readout of q₀ across 10 different physical chains while ε_B averages over the readout of q₂, and the 10 chains were selected by the layout discovery procedure based on entangling-gate fidelity rather than readout fidelity, so the two end-qubit positions in the chain are not expected to have identical aggregated readout performance. 6. Per-Module Variance and Drift Control Figure 4 (provided as a separate file in this deposit) displays the per-module diagnostic in two panels. Panel (a) plots the individual module P(ok, ok) values at each θ as scattered points overlaid on the analytic prediction curve, showing that the 10 parallel modules at each angle cluster tightly around the curve with a spread that varies with θ and with no module producing systematically anomalous behavior. Panel (b) compares the per-θ inter-module standard deviation to the Poisson expectation √[p(1−p)/N] at N = 3000 shots. The observed inter-module σ tracks the Poisson expectation across most of the angle range, with mild excess at θ near 60° and θ near 110° and a clean ratio of order unity at θ near π/2. The Poisson expectation vanishes at θ = π where p = 0; the corresponding ratio is undefined at this endpoint. The inter-module standard deviations σ_obs are within a factor of 2.2 of the Poisson expectation σ_Poisson across all 10 angle values. The mean of σ_obs/σ_Poisson across the 9 angles where the ratio is well-defined is 1.26, indicating that inter-module variance is consistent with shot-noise-limited statistics with a small additional contribution from per-chain calibration variation. This is a substantial improvement over the per-module variance excess factor of ~6 observed in the canonical-point baseline (Brahmbhatt 2026c) before the edge blacklist was applied, indicating that the targeted exclusion of two known-bad chains (qubits [80,81,82,83] and [29,30,31,32]) effectively resolved the patch-correlated outliers identified in the prior work. The successful tracking of Poisson expectation by the per-module variance, combined with the absence of any single module producing systematically anomalous P(ok, ok) values across the angle scan, supports the conclusion that the time-interleaved randomized submission protocol adequately suppressed the patch-correlated drift artifacts that motivated the methodology paper in the first place. The full curve match in Figure 1 is therefore not the artifact of a sampling pattern aligned with the θ axis; it is a genuine functional reproduction of the Frauchiger–Renner joint-outcome probability across the coin-preparation angle range. 7. Conclusion The Frauchiger–Renner joint-outcome probability P(ok, ok; θ) = cos²(θ/2)/4 is reproduced experimentally on the 156-qubit IBM Kingston superconducting quantum processor across the full coin-preparation angle range θ ∈ [0, π], at 10 angle values with 30,000 aggregated shots per angle. The Pearson correlation between observed and analytic P(ok, ok) is 0.99955, and the root-mean-square residual is 0.0094 absolute before readout-error correction and 0.0036 after. A two-parameter symmetric readout-error model with ε_A = 0.028 and ε_B = 0.034 captures the systematic residual structure and reduces the residual root-mean-square by a factor of 2.6. The full four-outcome joint distribution including the (fail, fail), (fail, ok), and (ok, fail) buckets follows the predicted angle dependence across the same range, with residual structure largely absorbed by the readout-error model. Per-module inter-chain variance is consistent with Poisson expectation across the angle range, supporting the conclusion that the time-interleaved randomized submission protocol established in the companion methodology record (Brahmbhatt 2026c) adequately suppressed the patch-correlated drift artifacts identified in prior work. To our knowledge, this is the first reported full-range experimental measurement of the Frauchiger–Renner paradoxical-outcome probability on a superconducting quantum processor, parameterized in the coin-preparation angle and scaled to the 100+ qubit class of platforms. The result extends prior canonical-point validations of the Frauchiger–Renner construction to a quantitative match against a non-trivial functional form, and characterizes the Wigner-basis joint readout fidelity of the platform as a function of the underlying state amplitudes. The total experimental cost was 81.91 seconds of billed QPU time on a single submitted job. We do not interpret the result as bearing on interpretational questions in quantum mechanics. The Frauchiger–Renner construction is a no-go theorem about the joint consistency of certain assumptions concerning quantum-mechanical observers; reproducing the predicted joint-outcome probability across θ confirms that the standard quantum-mechanical formalism continues to apply to nested-observer circuits at this scale on this platform, but does not by itself adjudicate among interpretations that all share that formalism. 8. Plain-Language Results and Significance Expressed in everyday language: ordinary quantum mechanics predicts a specific mathematical curve describing how the probability of one particular measurement outcome in a four-qubit circuit should change as a single experimental setting — an input angle — is varied from one end of its range to the other. The shape of this curve is not a flat line or a simple slope; it rises and falls in a non-trivial way, and the canonical Frauchiger–Renner value of 1 in 12 is just one point on the curve. The experiment reported here measured the curve at 10 settings on IBM's 156-qubit Kingston processor and found that the measurements land on the predicted curve, with a correlation coefficient of 0.9996 and a typical disagreement of about one percent — which is itself largely accounted for by the known measurement-error rate of the hardware. This result does not prove many-worlds, the existence of parallel universes, or any other specific interpretation of quantum mechanics. The Frauchiger–Renner name has sometimes been invoked in popular science writing to suggest that quantum experiments can adjudicate among interpretations of the theory. They cannot, and this one does not. Frauchiger and Renner's original 2018 paper was a logical argument about which combinations of philosophical assumptions can hold simultaneously within quantum theory, not a prediction that a quantum computer would reveal exotic new physics. What the present experiment confirms is much more practical: a modern 156-qubit superconducting quantum computer, when operated with appropriate experimental controls, reproduces ordinary quantum mechanics faithfully across a non-trivial functional prediction — even for a circuit construction tied to one of the most philosophically charged results in the field. The present paper belongs alongside the companion methodology record. The methodology paper documented an experiment on the same processor in which a naive measurement protocol produced an apparent 17-standard-deviation anomaly that vanished once proper controls were applied. The present paper, using the same processor with the same controls in place from the start, recovers the intended quantum-mechanical prediction to within hardware-level systematics. Taken together, the two records are a worked example of a deliberate research stance: apparent anomalies that do not survive reproducibility tests should be published as methodology lessons, and predictions that do survive should be published as validations. Neither the sensational reading — that something strange was found — nor the dismissive reading — that nothing interesting happened — captures the actual story. The accurate story is that the hardware behaves exactly as quantum mechanics predicts when it is asked properly, and that asking it properly is itself a nontrivial experimental skill. Let the computations tell the truth. Data Availability The raw experimental output JSON (fr_angle_scan_kingston_20260420_175009.json) and the IBM Runtime authoritative job export (job-d7jcii9s7cos73ejnl40-result.json), comprising aggregated and per-module joint-outcome counts at each of the 10 angle values, the assigned 4-qubit module layouts, the mitigation stack metadata, and the authoritative billed QPU timing, are deposited alongside this manuscript in the same Zenodo record. The four figures reproduced in the manuscript are included as separate files at the deposit. Experimental scripts are not included in the deposit; interested researchers may contact the author directly. Acknowledgments The author thanks IBM Quantum for access to the Kingston backend. The hardware-adaptive layout discovery and per-edge fidelity scoring procedures applied in Section 3 were developed in prior work on the Y⊗Z stabilizer basis migration architecture (Brahmbhatt 2026a, 2026b) and reused here without modification. The time-interleaved randomized submission protocol and the edge blacklist derived from outlier-module diagnostics were established in the companion methodology record (Brahmbhatt 2026c). The analytic derivation of P(ok, ok; θ) for the specific gate decomposition implemented in this work, presented in Section 2, benefited from independent verification by an external technical reviewer and is gratefully acknowledged. References Bong, K.-W., Utreras-Alarcón, A., Ghafari, F., Liang, Y.-C., Tischler, N., Cavalcanti, E. G., Pryde, G. J., and Wiseman, H. M. (2020). A strong no-go theorem on the Wigner's friend paradox. Nature Physics 16, 1199–1205. Brahmbhatt, A. (2026a). QuantaCore: A Y⊗Z Stabilizer Basis Migration Architecture for Fault-Tolerant Modular Quantum Computing. U.S. Provisional Patent 63/952,786. Zenodo prior art record DOI: 10.5281/zenodo.18498540. Brahmbhatt, A. (2026b). Y⊗Z Orthogonal Stabilizer Validation on the 156-qubit IBM Kingston Processor. Zenodo DOI: 10.5281/zenodo.19478241. Brahmbhatt, A. (2026c). Drift-Robust Multi-Time Probes on IBM Kingston: Frauchiger–Renner Baseline, Apparent Non-Markovian Signatures, and the Role of Randomized Sampling. Zenodo DOI: 10.5281/zenodo.19673112. Frauchiger, D., and Renner, R. (2018). Quantum theory cannot consistently describe the use of itself. Nature Communications 9, 3711. Proietti, M., Pickston, A., Graffitti, F., Barrow, P., Kundys, D., Branciard, C., Ringbauer, M., and Fedrizzi, A. (2019). Experimental test of local observer independence. Science Advances 5, eaaw9832.