Adaptive Q-Loop: Self-Tuning Continuous Dynamical Decoupling for Field-Deployable Quantum Sensors

Ethan A. Caspian

Published Dec 2, 2025• 0 citations

Abstract

Portable solid-state quantum sensors, such as nitrogen-vacancy (NV) center magnetometers operated on moving platforms, must contend with strongly nonstationary “robotic” noise: a composite of low-frequency drift, narrow spectral lines, and mechanically induced sidebands that evolves on timescales of seconds to minutes. Recently, the Q-Loop protocol—a continuous, open-loop phasemodulation scheme in which a smooth folded waveform y(t) replaces discrete π pulses—was introduced and shown to provide substantial coherence enhancement under composite robotic noise in static conditions [1]. However, in realistic field operation the underlying noise correlation time and line amplitudes drift, and a fixed choice of Q-Loop parameters (η, T) can lose most of its advantage. Here we propose and analyze adaptive Q-Loop, a self-tuning continuous dynamical decoupling protocol that compensates slow spectral drift using low-overhead calibration and a precomputed control policy. We model the dephasing noise as a parametric power spectral density with three dominant components (Ornstein–Uhlenbeck band, line spectrum, and regularized 1/f tail), and treat the corresponding parameters θ(t) as slowly varying stochastic processes. Within the standard filterfunction formalism (Gaussian, weak noise, quasi-stationary over a single interrogation), we first map the static coherence-extension factor R(η, T; θ) and show that Q-Loop exhibits broad performance plateaus in (η, T) for realistic composite spectra. We then construct an offline-optimized policy P that assigns near-optimal (η, T) to each point in a low-dimensional parameter space (τc, Alines, A1/f ) and combine it with a lightweight online estimator based on a small set of short Ramsey and bandselective probe sequences. Numerical simulations using time-dependent Hamiltonian evolution (QuTiP) with drifting correlation time and line amplitudes demonstrate that, for factor-of-five drifts in τc and order-unity variations in line strength, a static Q-Loop tuned at t = 0 can degrade from an initial ∼ 50× coherence extension down to ∼ 2×. In contrast, adaptive Q-Loop maintains ∼ 40× extension and recovers roughly 80–90% of the coherence advantage of an oracle policy that knows θ(t) exactly, using a calibration overhead fcal ≲ 10% of the total shots. This performance far exceeds that of fixed CPMG/XY8 sequences with realistic timing and amplitude errors under the same drifting noise. Adaptive Q-Loop therefore realizes a practical, low-complexity “firmware layer” for plug-andforget coherence protection in field-deployable quantum sensors.

DOI: 10.36227/techrxiv.176470631.19639964/v1

Related papers

Q-Loop-Enabled Deterministic Coherence Management for Portable Quantum Sensors

Ethan Arthur Caspian

Portable NV-center magnetometers suffer severe coherence loss from composite "robotic" noise (low-frequency 1/f components, line harmonics, and mechanical vibrations). Standard high-order dynamical decoupling (CPMG, XY8, UDD) delivers spectacular performance in the lab but degrades under realistic timing jitter and flip-angle errors encountered in mobile platforms. We introduce Q-Loop, a continuous, open-loop phase-modulation protocol that replaces discrete π-pulses with a smooth, folded waveform y(t) implemented via IQ modulation. Using the filter-function formalism, we derive analytic and numerical expressions for the resulting |F (ω, τ)| 2 and optimize parameters (η ≃ 3.6, T ≃ 0.97τ c) from a coarse estimate of the noise correlation time τ c. Stochastic simulations incorporating composite robotic noise and realistic control errors (0-20% timing jitter and 0-20% flip-angle error) show that optimized Q-Loop consistently delivers ∼ 5.6× coherence extension-outperforming realistically implementable CPMG and UDD, and remaining within ∼ 10-15% of XY8-8 even under severe conditions-while requiring no pulse calibration or high-bandwidth switching. To our knowledge, Q-Loop is the first continuousmodulation protocol systematically optimized and benchmarked against pulsed sequences under field-relevant composite noise and control imperfections. Q-Loop thus emerges as a robust, low-overhead "firmware layer" for field-deployed quantum sensors.

Nov 27, 20250 citations

Q-Loop-Enabled Deterministic Coherence Management for Portable Quantum Sensors

Caspian, Ethan A.

This repository contains the simulation and analysis code accompanying the manuscript “Q-Loop-Enabled Deterministic Coherence Management for Portable Quantum Sensors”. It provides a fully reproducible Python/NumPy/QuTiP workflow to evaluate Q-Loop, a continuous phase-modulation protocol for coherence protection in NV-centre magnetometers and related qubit sensors operating under “robotic” composite noise.The code computes filter functions (|F(\omega,\tau)|^2) and decoherence functionals (\chi(\tau)) for free evolution, Q-Loop, and standard pulsed dynamical decoupling sequences (CPMG, XY8, UDD), under both Lorentzian and composite (1/f +) line-rich noise spectra. It generates the main figures in the paper, including: (i) filter-function and noise-weighting plots, and (ii) jitter–flip-angle robustness heatmaps comparing Q-Loop to XY8/CPMG. A Colab-friendly script is included so all figures can be regenerated on a laptop without specialized hardware. The repository also provides example waveform generators (I/Q baseband) for implementing Q-Loop and pulsed sequences on standard AWGs, together with reference parameter sets ((\eta \simeq 3.6, T \simeq 0.97\tau_c)) that reproduce the reported (\sim 5)–(6\times) coherence extension. This archive is intended as a starting point for both numerical studies of coherence management under realistic platform noise and for experimental implementation on ensemble NV magnetometers or other driven qubit sensors.

Nov 21, 20250 citations

Q-Loop Control: Spectral Noise Filtering and Reduced Control Power for GHZ Stabilization under 1/f Dephasing

Ethan Arthur Caspian

Dynamical decoupling and shaped control are standard tools for mitigating dephasing noise in solid-state qubits. Most analyses either assume commuting pure dephasing with longitudinal control, where filter-function theory applies exactly, or move to fully numerical non-commuting dynamics with limited interpretability. Here I introduce Q-Loop, a scalar modulation protocol designed to suppress low-frequency 1/f γ dephasing more efficiently than standard Carr–Purcell (CP) sequences, and I connect its performance across both regimes. First, in the commuting pure-dephasing limit I compute the filter functions F(ω, τ ) and decoherence functionals χ(τ ) for free evolution, CP, and Q-Loop under 1/f γ noise with γ ∈ {0.5, 0.9, 1.2}. For γ = 0.9 I find that Q-Loop reduces the integrated noise susceptibility to χQ-Loop/χCP ≈ 0.36, with even stronger suppression for steeper spectra (χQ-Loop/χCP ≈ 0.14 at γ = 1.2). Monte Carlo simulations of classical noise trajectories verify that the resulting coherence W(τ ) = ⟨e −iϕ⟩ matches the analytic prediction e −χ(τ) . Second, I move beyond the commuting approximation and study an n-qubit GHZ state evolving under a time-dependent Hamiltonian with transverse control and longitudinal dephasing noise, H(t) = β 2 y(t) Xn i=1 Xi + 1 2 Xn i=1 Bi(t)Zi, where Bi(t) are mixed local+collective classical 1/f 0.9 fields and y(t) is either Q-Loop or CP. Using QuTiP to simulate unitary dynamics conditioned on classical noise, I determine the minimum control amplitude β ∗ (n) required to maintain fidelity F ≥ 0.9 with the GHZ state at a fixed interrogation time. For n = 2, 3, 4 qubits I obtain β ∗ Q-Loop ≃ (2.03, 2.28, 2.53) and β ∗ CP ≃ (15.53, 15.53, 16.00), corresponding to an average efficiency gain ⟨β ∗ CP/β∗ Q-Loop⟩ ≈ 6.9. Thus the spectral advantage of Q-Loop in reducing χ(τ ) is reflected in a substantial reduction of transverse control power required to stabilize entanglement under realistic non-commuting dynamics. I also discuss the limitations of the present analysis—classical noise, finite system size, and the restricted range n = 2–4—and outline how the framework can be extended to larger systems and more microscopic noise models.

Dec 9, 20250 citations

Noise-Adaptive Predictive Dynamical Decoupling

Ali Abu-Nada, Subhashish Banerjee, Lian-Ao Wu

Protecting quantum coherence against realistic environmental noise remains one of the fundamental obstacles to scalable quantum technologies. We develop a noise-adaptive dynamical decoupling framework that combines analytical open-quantum-system modeling with machine-learning-based forecasting for a qubit interacting with random telegraph noise. Unlike conventional dynamical decoupling protocols based on fixed pulse schedules, the proposed approach continuously forecasts short-time coherence evolution and adaptively applies control pulses according to the instantaneous noise dynamics. We investigate stationary and non-stationary environments spanning both Markovian and non-Markovian regimes. Numerical simulations demonstrate that the machine-learning-assisted adaptive control strategy substantially outperforms conventional periodic dynamical decoupling while using a comparable number of control pulses. The improvement becomes particularly pronounced in non-Markovian and non-stationary regimes, where memory effects, coherence revivals, and temporally evolving noise strongly limit the effectiveness of static pulse protocols. These results establish predictive machine-learning-assisted dynamical decoupling as a promising and scalable framework for adaptive quantum control in realistic noisy quantum devices.

quant-pharXiv:2606.15769Jun 16, 20260 citations

Q-Loop 2: Filter-Function Engineering for Deterministic Coherence Enhancement on NISQ Hardware

Ethan Arthur Caspian

We study a deterministic modulation protocol, Q-Loop, that replaces multi-axis, measurementdependent control with a single scalar waveform and is implementable as an open-loop schedule on current NISQ hardware. Building on earlier work where Q-Loop was shown to reduce both the dephasing functional χ(τ ) and the minimum control amplitude required to stabilize few-qubit GHZ states under classical 1/f noise, we revisit the protocol with a more conservative analytic treatment and a quantitative comparison to alternative empirical scaling forms for modest system sizes (n = 2–8 qubits). In a pure-dephasing (or effectively commuting) regime we derive the filter-function structure of an idealized Q-Loop waveform and show that it produces a discrete comb of sidebands with tunable low-frequency suppression relative to free evolution and standard Carr–Purcell (CP) sequences. For more general noise channels the scalar filter-function picture is used as a qualitative guide, and quantitative statements are obtained from stochastic-Hamiltonian simulations with classical 1/f γ noise. Using QuTiP-based simulations of Schrödinger dynamics with time-dependent Hamiltonians and classical 1/f γ local and correlated noise, we compute the minimum control amplitude β ∗ (n) needed to maintain a fixed target fidelity as a function of system size n for n = 2–8 qubits. Over this range the ensemble-averaged β ∗ (n) data are empirically well described by β ∗ (n) ≈ Aeαn with α ≃ 0.23–0.25, while quadratic and power-law forms yield systematically higher residuals and poorer goodness-of-fit metrics (including R2 , RMSE, and Bayesian information criterion), with BIC values for the exponential model typically 10–15 units lower than for polynomial alternatives. We emphasize that this exponential behaviour is an empirical description over a finite range, not an asymptotic law. We further map the Q-Loop-induced reduction in decoherence onto a toy physical error rate p and propagate this into a repetition-code capacity model. For a fixed phenomenological mapping, Q-Loop preprocessing yields an apparent 5–6× elevation of toy thresholds relative to an amplitudematched CP baseline, although these numbers are not intended as accurate thresholds for any specific fault-tolerant architecture. Finally, we discuss deployment on superconducting, trapped-ion, and photonic platforms using a Python-based waveform generator with PyVISA AWG integration, and outline how platform-specific noise spectra could be used to tune Q-Loop parameters in situ

Dec 9, 20250 citations

Measurement-Induced Fixed Points in Dynamical-Decoupling Noise Spectroscopy

Hiroyuki Shioiri

Dynamical-decoupling spectroscopy using solid-state spin qubits provides a powerful probe of environmental noise across wide frequency ranges. Here we identify a previously unrecognized observable fixed-point behavior in decoherence dynamics arising from additive decoherence channels sampled under narrow-band spectral filtering. We show that the ratio of decoherence times between two experimentally distinguishable configurations converges to a parameter-independent plateau value R ≈ 4/3 over an intermediate range of pulse numbers. This fixed point is independent of microscopic noise models and admits a natural geometric interpretation in terms of decoherence channel structure in the influence-functional description of open quantum systems. A central consequence is a strong experimental prediction: the existence of the plateau requires a band-limited spectral distortion, such that reconstructed noise spectra satisfy S_XY(ω)/S_CPMG(ω) ≈ 4/3 within a finite frequency window while approaching unity outside it. Numerical simulations demonstrate robustness of the effect across realistic noise environments without parameter fine tuning. We further propose an experimentally accessible protocol based solely on standard dynamical-decoupling measurements on a single qubit, enabling direct falsification of the predicted behavior. More broadly, these results indicate that measurement filtering can generate emergent invariant structures in open quantum-system dynamics, suggesting a new organizing principle for interpreting decoherence spectroscopy across quantum sensing platforms. Version 2 (v2) update: This version clarifies the logical structure of the paper by separating the existence of the measurement-induced plateau from the specific benchmark value R=4/3R=4/3R=4/3. The plateau mechanism is presented as a consequence of narrow-band spectral cancellation, while the value R=4/3R=4/3R=4/3 is treated more conservatively as a minimal effective-weight assignment consistent with pulse-sequence symmetry and the numerical results. The revised manuscript also makes the falsifiability structure more explicit and narrows the scope of the numerical claims: the simulations support the existence and robustness of the plateau, but are not presented as a unique microscopic derivation of the effective channel weights. Note (updated): This record corresponds to an earlier exploratory version of the idea. Parts of the theoretical interpretation presented here were not fully established and have been refined in subsequent work. In particular, the universal interpretation of ratio plateaus should be treated with caution. Readers are encouraged to consult the later work (see related records) for a more robust and conservative formulation of the effective description.

Feb 28, 20260 citations