Steady States of a Single Trapped-Ion Spin Coupled to an Engineered Non-Markovian Bath

Source: arXiv:2607.19286 · Published 2026-07-21 · By Anthony Vogliano, Lewis Hahn, Fabien Lefebvre, Jingwen Zhu, Sakshee Patil, Mahmood Sabooni et al.

TL;DR

This paper presents an experimental quantum simulation of a single spin-1/2 system (implemented with a trapped 171Yb+ ion) coupled to either a Markovian or a non-Markovian engineered dissipative bath. Whereas most open quantum systems assume Markovian environments where the bath instantaneously forgets information, here the authors realize non-Markovian dissipation by simulating correlated, temporally structured quantum jump events that possess memory effects. The experiment demonstrates that the steady-state of a driven-dissipative single qubit changes qualitatively under non-Markovian dissipation, yielding steady-state magnetization regimes inaccessible in the purely Markovian case. In particular, the system can reach positive magnetization along the reset axis despite the dissipation always resetting to the negative eigenstate, a phenomenon forbidden by Markovian dynamics. Such results validate theoretical predictions that non-Markovian baths enrich the steady-state phases accessible to open quantum systems.

Methodologically, the study relies on simulating many quantum trajectories where probabilistic, temporally correlated quantum jumps (resets) interrupt coherent Rabi oscillations driven by a microwave field. By sampling reset intervals from a chosen probability distribution function f(t), the authors engineer arbitrary bath memory kernels. They focus on a Pareto distribution for f(t), capturing non-Markovian power-law temporal correlations, and contrast this with exponential distributions representing Markovian baths. Collecting ensembles of these trajectories approximates the steady-state density matrix. The setup uses Floquet-controlled optical pumping pulses to implement deterministic resets applied probabilistically in time according to the sampled f(t). Extensive calibration ensures suppression of leakage to auxiliary ion states outside the qubit manifold. The measured steady-state magnetizations along orthogonal axes quantitatively match theoretical predictions from generalized master equations with memory kernels derived from f(t).

This work provides a versatile quantum simulation framework for studying non-Markovian open quantum systems, demonstrating experimentally accessible signatures of non-Markovianity in steady-states that single-qubit Markovian models cannot produce. The approach naturally extends to many-body trapped-ion systems, where complexity escalates and classical simulation becomes infeasible, opening new directions in reservoir engineering within structured, memory-retaining environments.

Key findings

• Non-Markovian dissipative channel with a Pareto distributed reset time produces steady-state magnetization opposite to the reset direction, unattainable by any Markovian dissipation strength ratio (Fig. 2 and Fig. 7).
• Maximum steady-state z-magnetization for non-Markovian dissipation reaches ⟨Mz⟩ = 0.13 ± 0.03 at effective dissipation to coherent drive ratio Γeff/B ≈ 0.53.
• Markovian model steady-state magnetization is always negative, with maximum coherence buildup near Γ/B = 2 (Fig. 1 and Fig. 6).
• Experimental ensemble averages over ~10,000 quantum trajectories per data point accurately reproduce modeled survival functions f(t) for both Markovian (exponential) and non-Markovian (Pareto) baths (Figs. 4, 5, 9, 10).
• Shot noise and SPAM errors were controlled below 1% across data collection, enabling precise discrimination of steady-state regimes.
• Non-Markovian steady-state condition requires longer experiment time to reach ergodicity due to reduced variance in inter-jump times (Fig. 5: ¯N ≈ 26 ≫ ¯t²/σ² = 8 compared to Markovian ¯N ≈ 10 ≫ 1).
• Implementation of dissipation via Floquet optical pumping with pause of coherent drive suppresses leakage to auxiliary bright states |Z±⟩, maintaining effective 2-level qubit dynamics.
• The steady-state density matrix can be expressed as a renewal integral over the survival function g(t) weighted by coherent evolution operators U(t), enabling analytic connection between f(t) and observables (Appendix A).

Methodology — deep read

1. Threat Model & Assumptions: The study assumes an open quantum system comprising a single spin-1/2 system representing the qubit degrees of freedom of a trapped 171Yb+ ion interacting with an engineered environment (bath) that causes dissipative resets. Adversaries or noise sources are not explicitly considered—the focus is on simulating and understanding non-Markovian bath memory effects. The bath is controlled via temporal correlation functions f(t) that determine the timing distribution of reset quantum jumps. The system-bath coupling is assumed tunable to mimic both Markovian (memoryless) and non-Markovian (finite memory) dissipation channels.

2. Data and Experimental Setup: The qubit is encoded in the hyperfine ground states |F=0,mF=0⟩ = |↓⟩ and |F=1,mF=0⟩ = |↑⟩ of 171Yb+ ion trapped in an RF-Paul trap. Coherent Rabi oscillations between |↑⟩ and |↓⟩ at microwave frequency 12.642 GHz provide the Hamiltonian drive. The dissipative channel is engineered by implementing optical pumping resets at stochastically sampled times from distribution f(t). The experimental data comprises approximately 10,000 reset profiles per dissipation rate Γeff/B value, repeated over various values to sweep from Zeno-polarized to depolarized regimes. Calibration includes controlling for state preparation and measurement (SPAM) errors (<1%), rescaling durations to satisfy steady-state conditions, and ensuring suppression of population leakage into auxiliary bright states |Z±⟩ that do not evolve coherently.

3. Architecture and Algorithm: The underlying driven-dissipative dynamics are described by a Lindblad master equation for the Markovian bath case, and by a generalized Nakajima-Zwanzig master equation with time-dependent memory kernel K(t) in the non-Markovian case. The memory kernel K(t) derives from the stochastic reset time distribution f(t) via Laplace transforms, connecting the temporal correlations in the dissipation to the system's evolution.

Experimentally, the protocol simulates many quantum trajectories where a coherent drive Hamiltonian H = Bσ_y acts continuously except at discrete times sampled from f(t), where the system is reset deterministically to |↓⟩ via optical pumping. Each trajectory consists of segments of coherent evolution punctuated by resets with correlated times. Ensemble averaging many such trajectories approximates the solution to the generalized master equation and yields the steady-state density matrix ρ_ss.

4. Training Regime / Experimental Protocol: Though "training" is not applicable, the experiment carefully controls hyperparameters equivalent to dissipation rate Γeff, drive B, reset distributions (Pareto α=4 for non-Markovian), sequence length T ensuring many resets occur to eliminate initial state memory (steady state), and number of repetitions (~10,000 resets profiles). The procedure includes (1) sampling reset times from f(t) until total time exceeds T, (2) driving coherent evolution segments of length t_i between resets, (3) stopping evolution to apply optical pumping reset, (4) optionally rotating spin for measurement along x or z, (5) repeated measurements and ensemble average to calculate magnetization observables.

5. Evaluation Protocol: Primary metrics are steady-state magnetization components ⟨M_x⟩ and ⟨M_z⟩ measured via state-dependent fluorescence, compared against theoretical predictions for both Markovian and non-Markovian models. Statistical errors are shot-noise limited over 10,000 runs; SPAM errors calibrated and below 1%. Experimental data is shown alongside theory curves demonstrating qualitative differences between Markovian and non-Markovian steady-states. Distributions of reset counts, survival functions, and reset-time histograms verify the underlying f(t) distributions used. The steady-state approximation is tested by ensuring the expected number of resets ¯N satisfies the inequality ¯N ≫ ¯t^2/σ^2.

6. Reproducibility: The authors provide detailed descriptions of the ion-level structure, optical pumping protocols, and the stochastic reset sampling method but do not explicitly release code or datasets. The non-Markovian model is formulated analytically in Appendix A, and the experimental procedures are detailed enough for replication given equivalent trapped-ion hardware. The base quantum trajectory simulation method is a standard approach in open system quantum optics, adapted here for arbitrary reset-time distributions. Some technical hardware specifics (e.g., microwave horn parameters, pumping laser wavelengths) are documented.

End-to-end example: For fixed drive strength B, one chooses a non-Markovian reset distribution f(t) (Pareto with α=4), samples reset times {t_i} until total time ∑t_i > T, applies coherent drive segments for durations t_i interleaved with optical pumping resets. After completing the full experimental sequence per trajectory, one measures spin projection along chosen axes and repeats 10,000 times. Averaging yields steady-state magnetization data showing a positive ⟨M_z⟩ in opposition to the reset target state, confirming the distinctive steady-state allowed only by non-Markovian dissipation.

Technical innovations

• Implementation of non-Markovian dissipation via stochastic quantum jump events sampled from arbitrary temporal distributions f(t), enabling simulation of correlated reset dynamics beyond simple Poisson processes.
• Use of Floquet-controlled optical pumping pauses to deterministically reset the qubit state while suppressing leakage to auxiliary bright ion states outside the computational manifold.
• Analytic characterization of steady-state density matrices via renewal theory linking the survival function g(t) of f(t) to the expectation values of observables under coherent evolution.
• Experimental demonstration that non-Markovian dissipation can produce steady-states with magnetization opposite to the reset direction, an effect not realizable under any Markovian Lindblad dissipation strength.

Datasets

• Trapped 171Yb+ ion quantum simulator data — approx. 10,000 reset profiles per setting — in-house experimental

Baselines vs proposed

• Markovian dissipation (exponential f(t)): steady-state magnetization ⟨M_z⟩ reaches a maximum negative value near Γ/B = 2, never positive (Fig. 6).
• Non-Markovian dissipation (Pareto f(t), α=4): steady-state ⟨M_z⟩ reaches positive values up to 0.13 ± 0.03 at Γeff/B ≈ 0.53, opposite the reset direction (Fig. 7).

Figures from the paper

Figures are reproduced from the source paper for academic discussion. Original copyright: the paper authors. See arXiv:2607.19286.

Fig 1: Expected steady-state magnetization as reset

Fig 2: Shift in steady-state magnetization induced by

Fig 3: Relevant energy levels for in 171Yb+ ions.

Fig 4: Probabilistic operation of a deterministic optical

Fig 5: Distribution of total resets within a given profile.

Fig 6: Steady-state magnetization under Markovian

Fig 7: Steady-state magnetization expected vs

Fig 8: Procedure to implement steady-state experiment for arbitrarily structured dissipation invoking the framework

Limitations

• Experiment currently limited to single qubit systems; extension to many-body trapped-ion chains is suggested but not yet demonstrated.
• Non-Markovian bath simulated through correlated reset times rather than direct engineering of bath modes, an indirect approach that may omit some bath-system correlations present in natural environments.
• Steady-state results rely on ergodicity assumptions requiring sufficiently long experiment times (large ¯N); finite experimental runtime may limit full convergence in some regimes.
• The Pareto distribution parameter α=4 used represents one specific class of non-Markovian noise; other forms of bath spectral densities and associated dynamics remain to be studied experimentally.
• No adversarial or error model evaluation beyond intrinsic experimental noise—robustness to device imperfections or external perturbations not explicitly analyzed.
• Code and raw data are not publicly released, potentially hindering exact reproducibility.

Open questions / follow-ons

• How do non-Markovian dissipative effects manifest in many-body driven-dissipative quantum systems with spatially and temporally correlated baths?
• Can engineered non-Markovian reservoirs improve quantum error correction or enable novel dissipative state preparation protocols beyond Markovian reservoir engineering?
• What are the transient dynamics and relaxation pathways under different memory kernels f(t), beyond steady-state properties, especially following quantum quenches?
• How do spatial correlations and crosstalk between multi-qubit reservoirs in trapped-ion or other platforms affect non-Markovian dissipation and emergent phases?

Why it matters for bot defense

While the paper focuses on open quantum system simulation rather than classical bot defense or CAPTCHA challenges, the work demonstrates precise engineering and control of non-Markovian dissipation at the single-qubit level with trapped ions. For bot-defense and CAPTCHA practitioners interested in quantum-enhanced or quantum-secured authentication schemes, the methodological advances in controlled dissipative environments may inspire analogies in designing systems with temporal memory effects that modulate adversarial detection signals. Moreover, the concept of engineered baths with long temporal correlations could enrich future quantum platform noise models or security analyses where temporal correlation of system resets impacts vulnerability. More broadly, the demonstrated capability to simulate complex non-Markovian open dynamics experimentally offers insights for any application area considering memory-bearing environments or feedback mechanisms, including sophisticated bot-detection heuristics relying on temporally correlated adversary behaviors. However, direct application to CAPTCHA or bot defense requires further translational work beyond the quantum physics domain primarily investigated here.

Cite

@article{arxiv2607_19286,
  title={ Steady States of a Single Trapped-Ion Spin Coupled to an Engineered Non-Markovian Bath },
  author={ Anthony Vogliano and Lewis Hahn and Fabien Lefebvre and Jingwen Zhu and Sakshee Patil and Mahmood Sabooni and Zhexuan Gong and Rajibul Islam },
  journal={arXiv preprint arXiv:2607.19286},
  year={ 2026 },
  url={https://arxiv.org/abs/2607.19286}
}

Read the full paper

• arXiv:2607.19286 (abstract)
• arXiv:2607.19286 (PDF)