Pre-Calibrated Phase-Coherence Compensation for Superconducting Qubits: Emulator Study and Hardware Feasibility of the Aurora Method
Futoshi Hamanoue · 2025
Read Pre-Calibrated Phase-Coherence Compensation for Superconducting Qubits: Emulator Study and Hardware Feasibility of the Aurora Method by Futoshi Hamanoue (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-20741.txt.
Introduction
Futoshi Hamanoue reports in 2025 a feasibility study of Aurora-DD, a method that keeps superconducting qubits from losing their phase. Qubits drift out of phase as they idle, and that drift corrupts quantum computations. Aurora pairs a fixed XY8 pulse sequence, which echoes away slow noise, with one global phase offset. The offset is tuned in advance on a calibrated emulator, not live on the machine.
Across 30 randomised trials at each of four phase settings, the emulator runs cut mean-squared error by 68 to 97 percent against an unprotected baseline. A three-run campaign on IBM's real ibm_fez hardware pointed to roughly 99 percent lower absolute error. Hamanoue stresses those hardware numbers are proof-of-concept only. A side branch adding zero-noise extrapolation proved unstable, sometimes amplifying errors badly, and now sits in an appendix as a warning.
Catalog record
| CREATOR | Futoshi Hamanoue |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 44 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-20741
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Pre-Calibrated Phase-Coherence Compensation for Superconducting Qubits: Emulator Study and Hardware Feasibility of the Aurora Method Futoshi Hamanoue Abstract We present an emulator-based and hardware feasibility study of Aurora-DD, a phase-coherence compensation method that integrates a sign-based feedback update of a global phase offset Δ ϕ \Delta\phi with a fixed-depth XY8 dynamical decoupling (DD) scaffold. From a control perspective Aurora is defined as a closed-loop controller, but in this work the feedback optimization is performed offline on a calibrated emulator and the resulting Δ ϕ ∗ \Delta\phi^{\ast} is deployed as a pre-calibrated (open-loop) phase compensation on hardware. Thus, our contribution should be interpreted as an “offline closed-loop, online open-loop” feasibility demonstration rather than a full on-device adaptive controller. Using an Aer-based emulator calibrated with ibm_fez device parameters, Aurora-DD achieves substantial reductions in the mean-squared error (MSE) of the measured expectation value ⟨ Z ⟩ \langle Z\rangle , yielding 68 68 – 97 % 97\% improvement across phase settings ϕ ∈ { 0.05 , 0.10 , 0.15 , 0.20 } \phi\in\{0.05,0.10,0.15,0.20\} over n = 30 n=30 randomized trials. These large- n n emulator results provide statistically stable evidence that the combined effect of XY8 and Δ ϕ ∗ \Delta\phi^{\ast} suppresses both dephasing and systematic phase bias. On real superconducting hardware ( ibm_fez ), we perform a small-sample ( n = 3 n=3 ) multi-phase validation campaign. In this limited- n n regime, Aurora-DD yields point estimates corresponding to approximately 99.2 99.2 – 99.6 % 99.6\% reduction in absolute error relative to a no-DD baseline across all tested phase points, while DD-only and Δ ϕ \Delta\phi -only controls provide smaller but uniformly positive improvements. These hardware numbers are non-inferential and are reported transparently as feasibility evidence under tight queue, drift, and credit constraints; they are not intended as definitive population-level statistics. In contrast, the auxiliary Aurora+ZNE branch exhibits instability: shallow two-point ZNE occasionally amplifies calibration inconsistencies and produces large error outliers, especially when combined with deep DD blocks. We therefore relegate all ZNE analysis to the Appendix as a cautionary study, and position Aurora-DD (without ZNE) as the primary and practically reproducible contribution of this work. Overall, the combined emulator and hardware results support pre-calibrated Aurora-DD as a practical, stable, and hardware-compatible phase-coherence compensator for NISQ devices in single-qubit settings. The hardware campaign presented here is explicitly framed as a proof-of-concept (feasibility) study. The full experimental protocol is described in sufficient detail to enable replication on calibrated emulators and hardware. Keywords: Closed-loop quantum control, Phase-coherence compensation, Superconducting qubits, Dynamical decoupling, NISQ error mitigation, Zero-noise extrapolation 1 Introduction The reliability of current Noisy Intermediate-Scale Quantum (NISQ) devices is strongly limited by dephasing, control imperfections, and readout errors. Even in shallow single-qubit circuits, idle-time dephasing alone can distort simple observables such as the expectation value \langle Z\rangle of a phase-rotated state R_{X}(\phi)|0\rangle , producing deviations far larger than predicted by the ideal model \langle Z\rangle_{\mathrm{ideal}}=\cos\phi . These fluctuations pose a practical challenge for variational algorithms, calibration routines, and sensing protocols that rely on accurate single-qubit statistics. A wide range of error-mitigation techniques have been proposed to address these issues without invoking full quantum error correction. These include dynamical decoupling (DD) sequences such as XY8, which suppress low-frequency dephasing through echo-based refocusing; zero-noise extrapolation (ZNE), which attempts to reconstruct a zero-noise limit by artificially scaling the noise; and classical post-processing such as M3 readout mitigation. While each method can improve performance, they share an important limitation: they operate in an open-loop fashion. Their parameters—pulse timing, scaling factors, or readout matrices—are fixed in advance and do not adapt to run-to-run variations in hardware noise. When the instantaneous noise profile deviates from calibration (e.g., T_{2} drift, residual couplings, pulse distortion), open-loop mitigation can become inconsistent or even counterproductive. To address this gap, the Aurora framework was introduced as a closed-loop phase-coherence compensation method. Instead of relying solely on static calibrations of the decoherence channel, Aurora measures an operational phase-error proxy \delta Z(\phi)=\langle Z\rangle_{\mathrm{ideal}}-\langle Z\rangle_{\mathrm{meas}}, and applies an additional corrective phase rotation \Delta\phi chosen to minimize the phase-error mean-squared deviation. This feedback-based formulation is grounded in the Bloch-equation description of superconducting qubits and results in a lightweight sign-based update rule that is robust to measurement noise and compatible with existing NISQ hardware. The present paper advances this line of research by conducting an integrated emulator and hardware feasibility study of the Aurora method on IBM’s superconducting backend ibm_fez . Among the Aurora variants, we focus on the practically deployable configuration in which the closed-loop phase offset \Delta\phi is combined with a fixed-depth XY8 decoupling sequence. We refer to this hybrid as Aurora-DD . Importantly, in the current work the sign-based update of \Delta\phi is executed on a calibrated emulator, yielding an offline-optimized offset \Delta\phi^{\ast} that is then applied as a fixed, pre-calibrated phase compensation on hardware. Thus, the on-device implementation is operationally open-loop, even though it inherits its value from a closed-loop optimization. Crucially, we exclude Zero-Noise Extrapolation (ZNE) from the primary definition of Aurora-DD. As we show later, even shallow ZNE can introduce instability when combined with XY8, occasionally amplifying calibration-inconsistent runs. The ZNE-enhanced branch is therefore treated as an auxiliary option and is analyzed separately in Appendix A . Our contributions are summarized as follows: • We implement a hardware-ready Aurora-DD protocol that is fully compatible with IBM’s primitives-based runtime API and requires no circuit-depth increase beyond XY8 itself. The corrective phase \Delta\phi is optimized in a calibrated emulator via a closed-loop sign-based update and then frozen as a pre-calibrated offset \Delta\phi^{\ast} for hardware execution. • Through n=30 randomized emulator trials across four phase points \phi\in\{0.05,0.10,0.15,0.20\} , we show that Aurora-DD reduces the mean-squared error (MSE) of \langle Z\rangle by 68 – 97\% relative to an unmitigated baseline. These large- n results provide statistically stable evidence for the effectiveness of the controller under calibrated NISQ noise. • We perform a small-sample ( n=3 ) hardware validation on ibm_fez under the same phase settings. Within this limited- n regime, Aurora-DD yields point estimates corresponding to approximately 99.2 – 99.6\% reduction in absolute error, clearly outperforming both DD-only and \Delta\phi -only controls. We explicitly frame these hardware results as a proof-of-concept feasibility demonstration rather than as a definitive statistical study. • We show that the auxiliary Aurora+ZNE branch exhibits instability: shallow noise-scaling occasionally yields extrapolation artifacts and large outliers. This reinforces the need to treat ZNE as optional rather than integral to Aurora-DD in the current hardware regime. Taken together, the emulator and hardware findings support pre-calibrated Aurora-DD (without ZNE) as a practical and stable method for suppressing phase-coherence errors in superconducting qubits under realistic NISQ constraints. The remainder of this paper is organized as follows. Section 2 formalizes the Aurora update law and its integration with XY8. Section 3 describes the emulator and hardware experiments. Section 4 presents the multi-trial error analysis. Section 5 discusses limitations and future extensions. Appendix A provides the auxiliary analysis of ZNE. 2 Theoretical Framework of Aurora Aurora is formalized as a closed-loop phase–coherence compensation method that operates on top of standard superconducting–qubit noise described by the Bloch equations. In this section, we provide a complete theoretical specification of the controller, consisting of (i) a calibrated decoherence model, (ii)