Pith. sign in

REVIEW 6 major objections 4 minor 47 references

Harnessing Environmental Memory with Reinforcement Learning in Open Quantum Systems

T0 review · 6 major / 4 minor · reviewed 2026-08-03 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Reinforcement learning can control a driven open quantum system to amplify environmental memory, beating optimal control by spreading information backflow across multiple revival windows.

desk verdict Plausible but under-supported: RL may beat OCT at maximizing non-Markovianity, but the comparison is not information-neutral and the paper is missing key reproducibility details. read the letter →

arxiv 2601.01252 v3 pith:VY2ZGOZB submitted 2026-01-03 quant-ph

Safae Gaidi , Abdallah Slaoui , Mohammed EL Falaki , Amine Jaouadi This is my paper · ORCID
classification quant-ph
keywords non-MarkovianityreinforcementlearninginformationbackflowBLPmeasurequantumcontrolopensystemstructuredreservoir
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to establish that reinforcement learning (RL) can amplify non-Markovian memory effects in a driven two-level system coupled to a structured reservoir more effectively than standard optimal control theory (OCT). The central claim is that RL agents rewarded whenever the trace distance between two candidate states grows learn to spread information backflow over several memory windows, producing a larger integrated Breuer-Laine-Piilo non-Markovianity than OCT's strategy of sharpening a single dominant revival. In the model studied, PPO reaches a total non-Markovianity of about 0.37 and SAC about 0.29, both far above the OCT baselines, even though OCT creates higher local peaks. This works because the BLP measure is a time integral of the positive part of the trace-distance slope, so a moderate backflow sustained across many intervals beats a tall narrow burst. The result matters because non-Markovianity is increasingly treated as a usable quantum resource.

What carries the argument

The central object is the Breuer-Laine-Piilo (BLP) non-Markovianity measure, N_Tot = integral over [0,T] of max(0, dD/dt) dt, where D(t) is the trace distance between two optimally chosen initial states. The RL reward function is a discrete version of the same expression, r_k = max(0, (D_{k+1}-D_k)/dt), so the agent's objective matches the final figure of merit exactly. The second piece of machinery is the observation vector s_k = [t/T, D_k, dD/dt_{k-1}, gamma_k, Omega_k], which gives the policy direct access to the instantaneous decay kernel gamma(t) and the previous trace-distance slope as an indicator of imminent backflow, letting the agent anticipate backflow windows. The OCT baselines m

What would settle it

Run PPO and SAC with the observation vector reduced to [t/T, Omega_k], omitting gamma_k and the D / dD/dt history; if total non-Markovianity drops to OCT levels, the RL advantage is an information-access effect. A second check: set the reward penalties alpha(Delta Omega)^2 and beta Omega^2 in Algorithm 3 to zero and verify whether PPO's smooth-pulse and highest-N_Tot conclusions still hold, and test whether an OCT solver restarted from a pulse pre-aligned to all three negative-gamma windows reaches N_Tot approximately 0.37.

Watch

Extended reading notes

Core claim

The paper's central claim is that reward signals shaped by the BLP integral lead RL agents to a distributed-backflow strategy: instead of concentrating the driving field into one strong revival, PPO and SAC synchronize the system dynamics with multiple intervals where the decay rate gamma(t) is negative, sustaining positive trace-distance growth over a longer fraction of the evolution. Because the total non-Markovianity N_Tot is the time integral of max(0, dD/dt), these moderate but prolonged revivals sum to more than the taller but narrower peak produced by gradient-based OCT. The authors report N_Tot approximately 0.37 for PPO, approximately 0.29 for SAC, and substantially lower values for

Load-bearing premise

The load-bearing premise is that the RL agents' access to the analytic memory kernel gamma(t) and full distinguishability history in their observations is not what produces the reported advantage over OCT, which only sees scalar objective values; the paper reports no ablation of the observation vector, so if gamma_k or the trajectory history is essential, the central comparison collapses.

Editorial extensions

If this is right

  • RL can serve as a model-free alternative to OCT for non-Markovian quantum control, requiring no analytic gradients and no microscopic model beyond sampled trajectories.
  • Maximizing integral measures like BLP favors sustained, multi-window control; future optimizers should reward duration as well as intensity of backflow.
  • PPO's clipped updates yield smoother, phase-aligned pulses than SAC while achieving the highest total non-Markovianity, suggesting hardware-friendly control and strong memory exploitation are compatible.
  • The learned policies synchronize the driving field with the reservoir's negative-gamma(t) windows, indicating that the timing of memory windows is a key physical resource.
  • Taken together, the results support the view of non-Markovianity as an engineered resource, with RL as a practical tool for memory-based quantum technologies.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Inference: The paper leaves untested whether the RL advantage survives if the analytic memory kernel gamma_k is removed from the observation vector; if PPO/SAC collapse to OCT levels without it, the headline result would reflect privileged information rather than model-free discovery.
  • Inference: The same reward-shaping principle likely transfers to other integral non-Markovianity measures and to multi-qubit networks, where coordinating backflow across several subsystems could multiply total memory, an extension the paper does not run.
  • Inference: Because the BLP measure ignores intervals with dD/dt < 0, the learned policies may waste time; penalizing interruptions of backflow in the reward could push N_Tot even higher.
  • Inference: The OCT baseline uses finite-difference gradients; an OCT solver with exact adjoint gradients over the full time-nonlocal equation might close the gap, meaning the reported hierarchy could be partly an artifact of gradient noise.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

6 major / 4 minor

Summary. The paper studies enhancement of non-Markovianity in a driven two-level system coupled to a Lorentzian reservoir. It defines the BLP measure N_Tot = ∫ max(0, Ẋ)dt, and optimizes piecewise-constant driving fields with two OCT methods (Powell, L-BFGS-B) and two RL algorithms (PPO, SAC) whose reward is the positive part of the trace-distance derivative. The central claim is that RL, especially PPO, achieves a larger integrated N_Tot (≈0.37) than SAC (≈0.29) and than OCT, because RL produces moderate but sustained backflow across multiple memory windows, whereas OCT concentrates effort into a single tall revival.

Significance. If the comparison is fair and the results are reproducible, the paper would offer a concrete, minimal model demonstrating that optimizing a time-integrated non-Markovianity measure favors temporally distributed control strategies, and that RL is a practical tool for such non-convex, memory-dependent landscapes. The master equation and BLP measure are standard and correctly assembled, and the reward design is natural. However, as submitted, the evidence for the headline hierarchy is incomplete, and several load-bearing details are missing or contradictory.

major comments (6)
  1. [Abstract vs. Sec. VI B] The Abstract states that 'SAC achieving the largest overall enhancement and PPO delivering slightly lower but still strongly improved performance,' while Sec. VI B and the Conclusion report the opposite: PPO reaches N_Tot≈0.37 and SAC≈0.29, with PPO 'largest total non-Markovianity.' This contradiction concerns the paper's central quantitative claim and must be corrected.
  2. [Sec. V A, Eq. (8)] The RL observation vector s_k includes the instantaneous decay rate γ_k, i.e., the analytic memory kernel of the reservoir, while the OCT methods (Sec. IV) receive only scalar evaluations of the objective. The paper's conclusion that RL 'uncovers' memory-exploiting strategies in a model-free way (Sec. VII) is therefore not supported unless the observation vector is ablated. Please train PPO/SAC without γ_k (and ideally without D and Ddot) and show that the RL-vs-OCT hierarchy persists; otherwise the headline result may be an information-access effect.
  3. [Algorithm 3 vs. Eq. (10)] Algorithm 3 defines the reward as r_k = max(0, Ddot_{k+1}) − α(ΔΩ_k)^2 − βΩ_{k+1}^2, but Eq. (10) defines it as r_k = max(0, Ddot_k). The coefficients α and β are never stated. If nonzero, the RL agents maximize a different objective than the BLP integral, and the claim that they 'directly encode the BLP information backflow' is inaccurate. Specify α, β, or remove the penalty terms and repeat the training.
  4. [Table I / Sec. VI B] Only a single seed (Seed=42) is reported for each RL algorithm, and no multiple trials or error bars are given for any method. The differences between PPO (0.37), SAC (0.29), and OCT could be within stochastic variation, especially for SAC, which is explicitly stochastic. Please report mean and standard deviation over several independent runs (or over several initializations for OCT) and assess whether the hierarchy is statistically robust.
  5. [Sec. IV / Appendix A] The OCT comparison lacks a 'level playing field' specification. No iteration budget, convergence tolerance, number of random starts, or finite-difference step size ϵ is given beyond generic pseudocode. L-BFGS-B with finite-difference gradients may be inaccurate or slow for N_c control parameters, and Powell's method has its own convergence caveats. Without explicit computational budgets and a fair termination criterion, the claim that RL outperforms OCT is not robust. Please specify the exact settings used to produce the curves in Figs. 2–4 and, ideally, include random-restart results.
  6. [Sec. II / Appendix A] The manuscript does not list the numerical values of the physical parameters (Δ, Γ, λ, T, N_c, Δt, Ω_min, Ω_max) nor the OCT iteration/tolerance settings. The observed hierarchy may depend on these values, and the results are not reproducible without them. Please provide a complete parameter table and, if possible, a sensitivity analysis in the strong-coupling regime.
minor comments (4)
  1. [Sec. VI B] Figure reference 'Fig.??' is unresolved; the sentence 'As shown in Fig.??, RL broadens...' needs a proper citation.
  2. [Sec. VI A] Minor grammar: 'Figure 3 summarize the behavior' should be 'summarizes'; also the text refers to 'Figure 3' when discussing RL curves that are shown in Fig. 6.
  3. [Throughout] Notation is inconsistent: N_Tot, N T ot, N_{Tot}, and N_{Tol} are used interchangeably. Please use a single symbol consistently.
  4. [Sec. V B] The text says 'PPO achieves a higher total non-Markovianity N_Tot than SAC' and later explains why, but the physical interpretation is presented before the numerical results. Consider moving this discussion to Sec. VI where the data are shown.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the RL reward is the BLP objective by design, the same objective OCT optimizes, and the self-citations are not load-bearing.

full rationale

Eq. (10) defines r_k = max(0, Ddot_k), which is exactly the integrand of N_Tot in Eq. (7). This is intentional reward design rather than a hidden prediction: the paper does not claim to derive N_Tot from independent data; it reports the value of an objective that both RL and the OCT baselines explicitly maximize (Algorithms 1-2 state the same N(Omega)=sum max(0,Ddot)dt objective). Reporting the objective value after optimization is not circular, and the RL-vs-OCT hierarchy is a numerical optimization outcome, not a quantity forced by definition. The observation vector in Eq. (8) includes gamma_k and Ddot_{k-1}, and Algorithm 3 contains penalty terms absent from Eq. (10); these are information-asymmetry and reproducibility limitations that undermine the model-free/discovery framing, but they are not circular reductions: the policy is not defined as the reported N_Tot, and no fitted parameter is renamed as a prediction. Self-citations (Refs. 13, 32, 34) support background claims and the standard Lorentzian decay-rate formula, which is co-cited with external Ref. 5 and is not used as a uniqueness theorem or ansatz to force the conclusions. Therefore no load-bearing step reduces to its own input.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on standard physics (Lorentzian-reservoir amplitude damping, BLP measure) plus a handful of numerical settings. The most important ledger entries are the undisclosed physical parameters, the undisclosed PPO reward penalties, and the assumption that OCT received a fair computational budget. No new physical entities, forces, or dimensions are postulated.

free parameters (3)
  • Physical parameters (Delta, Gamma, lambda, T, N_c, Delta_t, Omega_min, Omega_max) = not stated
    The decay rate gamma(t) (Eq. 3), the number and width of negative-gamma windows, and the total time T are set by these values, yet no numerical values appear in the text or appendix. The reported N_Tot values (~0.37 PPO, ~0.29 SAC) cannot be reproduced or even sanity-checked without them.
  • PPO reward penalty coefficients alpha, beta = not stated
    Algorithm 3 defines r_k = max(0, Ddot_{k+1}) - alpha(Delta Omega_k)^2 - beta Omega^2_{k+1}, but Eq. (10) gives r_k = max(0, Ddot_k) with no penalties. alpha and beta are never quantified; if nonzero they shape the smoothness of PPO pulses and directly affect the 'PPO is smoother' conclusion.
  • RL training hyperparameters (LR, batch, buffer, steps, action range, seed) = Table I: LR 3e-4/6e-4; batch 256/64; buffer 300k; steps 5e5; range [-5,5]; seed 42
    Disclosed in Table I; standard chosen-by-hand hyperparameters. The main concern is that all reported training curves and final values come from a single seed (42), with no repeats or variance reporting.
assumptions (4)
  • domain assumption The time-local master equation with time-dependent decay rate gamma(t) is a valid reduced description of the driven system coupled to a Lorentzian reservoir (Eqs. 2-3).
    Standard for a two-level system coupled to a structured environment; invoked in Sec. II without derivation. If the Lorentzian/unitary approximation fails, the control problem itself is mis-modeled.
  • domain assumption The BLP trace-distance measure, evaluated with the fixed initial pair |1><1| and |0><0|, is the appropriate operational non-Markovianity quantifier for this control problem.
    Sec. III; optimality of this pair is asserted 'following the standard result for amplitude-damping channels', but with a time-dependent drive present the optimal pair need not remain this one. The paper never checks whether the chosen pair bounds the true measure.
  • domain assumption gamma(t) is computable and observable at every step and can be fed to the agent (Eq. 8).
    The observation vector includes the analytic decay rate gamma_k; in an experiment this would require real-time estimation of the memory kernel, which is not discussed. This privileged observation is the load-bearing premise for the 'model-free' narrative.
  • ad hoc to paper OCT runs were converged fairly, with adequate iteration budget and comparable starting conditions.
    No OCT iteration counts or runtimes are reported. A finite-difference L-BFGS-B in N_c dimensions costs N_c full master-equation propagations per iteration, so budget fairness against 5e5 RL steps is unverified; if OCT was under-budgeted, 'RL beats OCT' may be an artifact of the comparison.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Harnessing Environmental Memory with Reinforcement Learning in Open Quantum Systems." pith.science (2026). https://pith.science/paper/VY2ZGOZB

@misc{pith2026260101252,
  author       = {Pith},
  title        = {Pith review of: Harnessing Environmental Memory with Reinforcement Learning in Open Quantum Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VY2ZGOZB}},
  note         = {Machine review of arXiv:2601.01252}
}
read the original abstract

Non-Markovian quantum dynamics, characterized by information backflow from the environment to the system, has emerged as a potential resource for quantum technologies. A key challenge is therefore to control and enhance such memory effects. In this work, we investigate the use of reinforcement learning (RL) to maximize non-Markovianity in a driven two-level system coupled to a structured reservoir. We compare RL-based control strategies with standard optimal control theory (OCT). We show that OCT produces localized but relatively weak revivalsin the instantaneous non-Markovianity rate, whereas RL policies generate significantly stronger and better-timed information backflow by synchronizing the system dynamics with favorable memory intervals of the environment. This enhanced exploitation of memory effects leads to a higher total integrated non-Markovianity for RL than for OCT, with SAC achieving the largest overall enhancement and PPO delivering slightly lower but still strongly improved performance with smoother, experimentally attractive pulses. Our results contribute to the emerging view of non-Markovianity as an operational resource and illustrate how RL can serve as a flexible, model-free tool for non-Markovian quantum control.

Figures

Figures reproduced from arXiv: 2601.01252 by the authors.

Figure 1
Figure 1. FIG. 1. Time-dependent decay rate [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Convergence history of the total non-Markovianity [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Instantaneous non-Markovianity [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: FIG. 4. Optimized control pulses Ω( [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5. Reinforcement-learning control loop for maximizing [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6. Instantaneous non-Markovianity [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 8
Figure 8. Figure 8: FIG. 8. Optimized control pulses obtained with SAC (top) [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 9
Figure 9. Figure 9: FIG. 9. Pseudocode for Powell’s derivative-free optimal-control algorithm applied to the maximization of total non [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 10
Figure 10. Figure 10: FIG. 10. Pseudocode for L-BFGS-B optimal control applied to the maximization of total non-Markovianity with bound [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]
Figure 11
Figure 11. Figure 11: FIG. 11. Pseudocode for PPO applied to maximizing the BLP non-Markovianity measure. [PITH_FULL_IMAGE:figures/full_fig_p013_11.png]
Figure 12
Figure 12. Figure 12: FIG. 12. Pseudocode for SAC applied to maximizing the BLP non-Markovianity measure in the driven open quantum system. [PITH_FULL_IMAGE:figures/full_fig_p014_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

47 extracted references · 15 canonical work pages

  1. [32]

    Reinforcement learning in different phases of quantum control,

    M. Bukov, A. G. Day, D. Sels, P. Weinberg, A. Polkovnikov, and P. Mehta, “Reinforcement learning in different phases of quantum control,”Phys. Rev. X8, 031086 (2018), doi:10.1103/PhysRevX.8.031086

    doi:10.1103/physrevx.8.031086 2018

  2. [1]

    Initializeρ 1(0),ρ 2(0), and a random initial ampli- tude Ω0

  3. [2]

    Apply the actiona k and propagate the master equation for one step ∆t

  4. [3]

    ComputeD k, ˙Dk, and the rewardr k

  5. [4]

    Store the transition (s k, ak, rk, sk+1): SAC uses a replay buffer, while PPO uses on-policy rollouts

  6. [5]

    Update the actor and critics using gradient descent according to the chosen RL algorithm. FIG. 5. Reinforcement-learning control loop for maximizing non-Markovianity. The agent observes (D, ˙D, γ,Ω), selects actions, and receives rewards based on information backflow. VI. RESUL TS AND DISCUSSION A. OCT versus RL: peaks versus area Figure 3 summarize the b...

  7. [6]

    PPO attains the largest total non-Markovianity NTot

  8. [7]

    SAC reaches a slightly smaller but still strongly en- hanced value

Show all 47 references
  1. [8]

    mountain

    OCT improves upon the uncontrolled dynamics but remains far below the RL policies in totalN Tot. At first sight this appears paradoxical: as shown in Fig. 3, OCT often produces slightly higher peaks in the instantaneous non-MarkovianityN loc(t) than the RL policies. The resolu...

  2. [9]

    Breuer and F

    H.-P. Breuer and F. Petruccione,The Theory of Open Quantum Systems(Oxford University Press, 2002), doi:10.1093/acprof:oso/9780199213900.001.0001

    2002

  3. [10]

    Rivas and S

    ´A. Rivas and S. F. Huelga,Open Quantum Systems: An Introduction(Springer, 2012), doi:10.1007/978-3-642- 23354-8

    2012 doi

  4. [11]

    Measure for the degree of non-Markovian behavior of quantum processes in open systems,

    H.-P. Breuer, E.-M. Laine, and J. Piilo, “Measure for the degree of non-Markovian behavior of quantum processes in open systems,”Phys. Rev. Lett.103, 210401 (2009), doi:10.1103/PhysRevLett.103.210401

    2009 doi

  5. [12]

    Quan- tum non-Markovianity: Characterization, quantification and detection,

    ´A. Rivas, S. F. Huelga, and M. B. Plenio, “Quan- tum non-Markovianity: Characterization, quantification and detection,”Rep. Prog. Phys.77, 094001 (2014), doi:10.1088/0034-4885/77/9/094001

    2014 doi

  6. [13]

    Quantum speed limit of Jaynes-Cummings model with detuning for arbitrary initial states,

    Y. Shahri, M. A. Jafarizadeh, S. Salimi, and M. S. Asl, “Quantum speed limit of Jaynes-Cummings model with detuning for arbitrary initial states,”Phys. Lett. A470, 128783 (2023), doi:10.1016/j.physleta.2023.128783

    2023

  7. [14]

    De- gree of non-Markovianity of quantum evolu- 9 tion,

    D. Chru´ sci´ nski and S. Maniscalco, “De- gree of non-Markovianity of quantum evolu- 9 tion,”Phys. Rev. Lett.112, 120404 (2014), doi:10.1103/PhysRevLett.112.120404

    2014 doi

  8. [15]

    Quantifying non-Markovianity via local quantum Fisher information,

    Y. Dakir, A. Slaoui, L. B. Drissi, and R. A. Laamara, “Quantifying non-Markovianity via local quantum Fisher information,”Annals of Physics480, 170142 (2025), doi:10.1016/j.aop.2025.170142

    2025

  9. [16]

    Non-Markovianity and reservoir memory of quan- tum channels,

    B. Bylicka, D. Chru´ sci´ nski, and S. Maniscalco, “Non-Markovianity and reservoir memory of quan- tum channels,”Scientific Reports4, 5720 (2014), doi:10.1038/srep05720

    2014 doi

  10. [17]

    Dynamics of non-Markovian open quantum systems,

    I. de Vega and D. Alonso, “Dynamics of non-Markovian open quantum systems,”Rev. Mod. Phys.89, 015001 (2017), doi:10.1103/RevModPhys.89.015001

    2017 doi

  11. [18]

    Non-Markovian dynamics, dense coding capacity, and non-locality in coupled two-qubit systems interacting with bosonic thermal environments,

    Y. Dakir, L. Bouhouch, A. Slaoui, and R. Ahl Laa- mara, “Non-Markovian dynamics, dense coding capacity, and non-locality in coupled two-qubit systems interacting with bosonic thermal environments,”Physica A: Statis- tical Mechanics and its Applications676, 130865 (2025). doi.o...

    2025

  12. [19]

    Ex- tracting quantum dynamical resources: consumption of non-Markovianity for noise reduction,

    G. D. Berk, S. Milz, F. A. Pollock, and K. Modi, “Ex- tracting quantum dynamical resources: consumption of non-Markovianity for noise reduction,”npj Quantum In- formation9, 104 (2023), doi:10.1038/s41534-023-00774- w

    2023 doi

  13. [20]

    Control-enhanced non- Markovian quantum metrology,

    X. Yang, X. Long, R. Liu, K. Tang, Y. Zhai, X. Nie, T. Xin, J. Li, and D. Lu, “Control-enhanced non- Markovian quantum metrology,”Commun. Phys.7, 282 (2024), doi:10.1038/s42005-024-01758-8

    2024 doi

  14. [21]

    Impact of non-Markovianity on quantum precision and synchronization in open quantum systems,

    S. Gaidi, A. Slaoui, M. E. Falaki, and R. A. Laamara, “Impact of non-Markovianity on quantum precision and synchronization in open quantum systems,”Physica A: Statistical Mechanics and its Applications681, 131129 (2026), doi:10.1016/j.physa.2025.131129

    2026

  15. [22]

    Trajectory Tracking of Stochas- tic Open Quantum Systems Based on Online Estimated State Feedback Control,

    S. Cong and Y. Zhou, “Trajectory Tracking of Stochas- tic Open Quantum Systems Based on Online Estimated State Feedback Control,”Adv. Quantum Technol.7(9), 2300217 (2024), doi:10.1002/qute.202300217

    2024 doi

  16. [23]

    Opti- mal control theory for a unitary operation under dis- sipative evolution,

    M. H. Goerz, D. M. Reich, and C. P. Koch, “Opti- mal control theory for a unitary operation under dis- sipative evolution,”New J. Phys.16, 055012 (2014), doi:10.1088/1367-2630/16/5/055012

    2014 doi

  17. [24]

    Training Schr¨ odinger’s cat: Quantum op- timal control: Strategic report on current status, visions and goals for research in Europe,

    S. J. Glaser, U. Boscain, T. Calarco, C. P. Koch, W. K¨ ockenberger, R. Kosloff, I. Kuprov, B. Luy, S. Schirmer, T. Schulte-Herbr¨ uggen, D. Sugny, and F. Wilhelm, “Training Schr¨ odinger’s cat: Quantum op- timal control: Strategic report on current status, visions and goals f...

    2015 doi

  18. [25]

    Quan- tum optimal control in quantum technologies: Strate- gic report on current status, visions and goals for re- search in Europe,

    C. P. Koch, M. H. Goerz, and S. J. Glaser, “Quan- tum optimal control in quantum technologies: Strate- gic report on current status, visions and goals for re- search in Europe,”EPJ Quantum Technol.9, 19 (2022), doi:10.1140/epjqt/s40507-022-00138-x

    2022 doi

  19. [26]

    Non-Markovian reinforcement learn- ing using fractional dynamics,

    G. Guptaet al., “Non-Markovian reinforcement learn- ing using fractional dynamics,” inProc. 60th IEEE Conf. on Decision and Control (CDC), IEEE, 2021, doi:10.1109/CDC45484.2021.9683076

    2021

  20. [27]

    Machine Learning Non-Markovian Quantum Dynamics,

    I. A. Luchnikov, S. V. Vintskevich, D. A. Grigoriev, and S. N. Filippov, “Machine Learning Non-Markovian Quantum Dynamics,”Phys. Rev. Lett.124, 140502 (2020), doi:10.1103/PhysRevLett.124.140502

    2020 doi

  21. [28]

    Memory ef- fect and non-Markovian dynamics in an open quan- tum system,

    F. Liu, X. Zhou, and Z.-W. Zhou, “Memory ef- fect and non-Markovian dynamics in an open quan- tum system,”Phys. Rev. A99, 052119 (2019), doi:10.1103/PhysRevA.99.052119

    2019 doi

  22. [29]

    Quan- tification of memory effects in topological two-band open quantum systems,

    H. Trivi˜ no, F. Mesa, and V. A. Ballesteros, “Quan- tification of memory effects in topological two-band open quantum systems,”Heliyon10, e40552 (2024), doi:10.1016/j.heliyon.2024.e40552

    2024 doi

  23. [30]

    R. S. Sutton and A. G. Barto,Reinforcement Learning: An Introduction, Vol. 1, No. 1, pp. 229–235 (MIT Press, Cambridge, 1998), doi:10.1017/S0263574799271172

    1998 doi

  24. [31]

    An introduction to deep rein- forcement learning,

    V. Fran¸ cois-Lavet, P. Henderson, R. Islam, M. G. Belle- mare, J. Pineau,et al., “An introduction to deep rein- forcement learning,”Found. Trends Mach. Learn.11(3– 4), 219–354 (2018), doi:10.1561/2200000071

    2018 doi

  25. [33]

    Mul- tipartite Entangled States in Dipolar Quantum Sim- ulators,

    T. Comparin, F. Mezzacapo, and T. Roscilde, “Mul- tipartite Entangled States in Dipolar Quantum Sim- ulators,”Phys. Rev. Lett.129, 150503 (2022), doi:10.1103/PhysRevLett.129.150503

    2022 doi

  26. [34]

    Sample-efficient model-based reinforcement learning for quantum control,

    I. Khalid, C. A. Weidner, E. A. Jonckheere, S. G. Schirmer, and F. C. Langbein, “Sample-efficient model-based reinforcement learning for quantum control,”Phys. Rev. Research5, 043002 (2023), doi:10.1103/PhysRevResearch.5.043002

    2023 doi

  27. [35]

    Model-free quantum control with reinforcement learning,

    V. V. Sivak, A. Eickbusch, H. Liu, B. Royer, I. Tsioutsios, and M. H. Devoret, “Model-free quantum control with reinforcement learning,”Phys. Rev. X12, 011059 (2022), doi:10.1103/PhysRevX.12.011059

    2022 doi

  28. [36]

    Deep reinforcement learning for quantum gate control,

    Z. An and D. L. Zhou, “Deep reinforcement learning for quantum gate control,”Europhys. Lett.126, 60002 (2019), doi:10.1209/0295-5075/126/60002

    2019 doi

  29. [37]

    Physics-informed reinforcement learning for optimal control of nonlinear systems,

    Y. Wang and Z. Wu, “Physics-informed reinforcement learning for optimal control of nonlinear systems,” AIChE J.70, e18542 (2024), doi:10.1002/aic.18542

    2024 doi

  30. [38]

    Comparing quantum hy- brid reinforcement learning to classical methods,

    M. Moll and L. Kunczik, “Comparing quantum hy- brid reinforcement learning to classical methods,”Hum.- Intell. Syst. Integr.3, 15–23 (2021), doi:10.1007/s42454- 021-00025-3

    2021 doi

  31. [39]

    Non- Markovian quantum control via model maximum likelihood estimation and reinforcement learn- ing,

    T. Neema, S. Jha, and T. Sahai, “Non- Markovian quantum control via model maximum likelihood estimation and reinforcement learn- ing,”arXiv preprintarXiv:2402.05084 (2024), doi:10.48550/arXiv.2402.05084

    2024 doi arXiv

  32. [40]

    Re-exploring control strategies in a non-Markovian open quantum system by reinforce- ment learning,

    A. Jaouadi, E. Mangaud, and M. Desouter- Lecomte, “Re-exploring control strategies in a non-Markovian open quantum system by reinforce- ment learning,”Phys. Rev. A109, 013104 (2024), doi:10.1103/PhysRevA.109.013104

    2024 doi

  33. [41]

    From non-Markovian dissipation to spatiotemporal control of quantum nanodevices,

    T. Lacroix, B. W. Lovett, and A. W. Chin, “From non-Markovian dissipation to spatiotemporal control of quantum nanodevices,”Quantum8, 1305 (2024), doi:10.22331/q-2024-04-03-1305

    2024 doi

  34. [42]

    A non-Markovianity measure based on quantum speed limit,

    S. Gaidi, A. Slaoui, M. E. Falaki, and R. A. Laa- mara, “A non-Markovianity measure based on quantum speed limit,”Physica A130733 (2025), doi:10.1016/j.physa.2025.130733

    2025

  35. [43]

    Soft actor-critic: Off-policy maximum entropy deep reinforce- ment learning with a stochastic actor,

    T. Haarnoja, A. Zhou, P. Abbeel, and S. Levine, “Soft actor-critic: Off-policy maximum entropy deep reinforce- ment learning with a stochastic actor,” inProc. Int. Conf. on Machine Learning (ICML), PMLR, 1861–1870 (2018)

    2018

  36. [44]

    Reinforcement learning with neural networks 10 for quantum feedback,

    T. F¨ osel, P. Tighineanu, T. Weiss, and F. Mar- quardt, “Reinforcement learning with neural networks 10 for quantum feedback,”Phys. Rev. X8, 031084 (2018), doi:10.1103/PhysRevX.8.031084

    2018 doi

  37. [45]

    Universal quantum control through deep reinforce- ment learning,

    M. Y. Niu, S. Boixo, V. N. Smelyanskiy, and H. Neven, “Universal quantum control through deep reinforce- ment learning,”npj Quantum Inf.5, 33 (2019), doi:10.1038/s41534-019-0141-3

    2019 doi

  38. [46]

    Robust quantum control using reinforcement learning from demonstration,

    S. Li, Y. Fan, X. Li, X. Ruan, Q. Zhao, Z. Peng, R.- B. Wu, J. Zhang, and P. Song, “Robust quantum control using reinforcement learning from demonstration,”npj Quantum Inf.11, 124 (2025), doi:10.1038/s41534-025- 01065-2

    2025 doi

  39. [47]

    Quantum optimal control via semi-automatic differenti- ation,

    M. H. Goerz, S. C. Carrasco, and V. S. Malinovsky, “Quantum optimal control via semi-automatic differenti- ation,”Quantum6, 871 (2022), doi:10.22331/q-2022-12- 07-871. 11 Algorithm 1: Powell OCT Input:initial control vectorΩ (0) ∈[Ω min,Ω max]Nc , initial set of search directi...

    2022 doi

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.