Pith. sign in

REVIEW 4 major objections 6 minor 62 references

By training a reinforcement-learning agent in two curriculum phases, the paper establishes that a near-optimal bosonic code under both single- and double-photon loss is the Fock-state pair |4> and |7>, with a cascading recovery operator, an

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-03 22:01 UTC pith:D5BN6V6O

load-bearing objection The useful result is the simple |4>,|7> AQEC code with a cascade recovery operator, which looks genuinely robust to double-photon loss; the performance claims, though, are not yet fully backed because the analytic solver is never fidelity-checked against a full master-equation simulation. the 4 major comments →

arxiv 2511.12482 v2 pith:D5BN6V6O submitted 2025-11-16 quant-ph cs.LG

Discovering autonomous quantum error correction via deep reinforcement learning

Yue Yin , Tailong Xiao , Xiaoyang Deng , Ming He , Jianping Fan , Guihua Zeng This is my paper
classification quant-ph cs.LG PACS 03.67.Pp
keywords autonomous quantum error correctionbosonic codesreinforcement learningcurriculum learningdouble-photon lossKnill-Laflamme conditionsFock statesmaster equation solver
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper claims that a deep reinforcement-learning agent, trained in two curriculum phases, can discover a practical autonomous quantum error-correction code for a bosonic cavity exposed to both single- and double-photon loss. The discovered code uses the Fock states |0L>=|4> and |1L>=|7> with a cascading recovery operator, and is reported to keep the mean logical fidelity at 91% after gamma_a t = 0.6 when the double-photon loss rate is 1.2% of the single-photon loss rate, above the breakeven threshold. The search is made tractable by an analytical solution of the effective master equation that splits the density matrix into decoupled diagonals, cutting simulation cost by a factor of order N. If correct, this matters because it suggests measurement-free error correction against higher-order photon loss can be discovered automatically and implemented with only nearest-neighbor Fock-state couplings.

Core claim

Under the approximate AQEC dynamics dρa/dt = (γa/2)D[a] + (γaλ/2)D[Leng], the agent converges to codewords |0L>=|4> and |1L>=|7> and to the recovery operator Leng ∝ |4><3| + |7><6| + |3><2| + |6><5|, a Hamiltonian-distance-1, cascading operator that maps second-order error states back toward the code space. The key structural fact is that <4|a^2|7>=0, so double-photon loss cannot flip the logical qubit directly; the residual violation of the standard exact-correction conditions appears only as dephasing. Using a mod-3 parity syndrome instead of photon-number parity enlarges the correctable error space. The paper reports that this code surpasses the breakeven threshold at γat=0.6 with mean fi

What carries the argument

Two mechanisms carry the argument. (1) An analytical solver for the effective master equation: because each density-matrix element couples only to elements offset by the same index, the equation decomposes into at most 2N−1 decoupled linear systems along the diagonals, reducing simulation complexity by about a factor of N and making RL training fast. (2) The discovered cascading recovery operator Leng ∝ |4><3|+|7><6|+|3><2|+|6><5|, a Hamiltonian-distance-1 operator (it connects only neighboring Fock levels) that implements a mod-3 parity error syndrome: it shuttles population from the second-order error space {|2>,|3>,|5>,|6>} back toward the code space. The code states themselves, Fock stat

Load-bearing premise

The entire result is computed from the reduced master equation that assumes the helper qubit stays in its ground state and the cavity-qubit coupling is much weaker than the helper qubit's decay; the paper inherits this reduction from earlier work and does not independently validate it.

What would settle it

Integrate the full three-part master equation (storage cavity, transmon, readout) without the adiabatic approximation ρ(t)=ρa(t)⊗|0><0| for the discovered code at the paper's test parameters (g/γa=600, γb/γa=1800, γa2=0.012γa). If the mean logical fidelity at γat=0.6 falls to or below the breakeven value of 0.84, the central claim fails. A hardware alternative: implement the pump-and-dump recovery sequence on a cavity-transmon module and measure logical-state fidelity under engineered double-photon loss.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • The discovered GRL code keeps mean fidelity above breakeven over longer evolution times under double-photon loss, while the T4C, binomial, and earlier RL codes fall below it.
  • Because the recovery operator has Hamiltonian distance d=1, the code can be implemented without nonlinear interactions; single-qubit logical gates require only third-order nonlinearity, versus fourth or sixth order for other codes.
  • The analytical solver reduces simulation time by a factor of roughly N, so reinforcement-learning searches remain feasible as the Fock-space truncation grows.
  • The mod-3 parity syndrome enlarges the correctable error space, giving a concrete mechanism for resisting second-order photon loss.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the reduced master equation is trusted, the same two-phase reward schedule should transfer to other error sets; one test is to add a^3 loss and see whether the agent converges to Fock-state pairs separated by four photons, giving a mod-4 syndrome.
  • The solver's speedup opens the door to searches at larger truncations (N=16 or 32); the paper's restriction to N=8 may have biased the discovery toward high-mean-photon states like |4> and |7>.
  • The mod-3 cascading structure suggests a family of codes of the form |m> and |m+k> with cascading recovery operators, which could be screened analytically before any training.
  • The initial fidelity dip implies practical implementations should pair the code with fast logical-state preparation or trajectory-resolved fidelity metrics, since the protection is not instantaneous.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper trains a deep reinforcement-learning agent (PPO with curriculum learning) to search for an approximate autonomous quantum error correction code for a bosonic mode subject to both single-photon and double-photon loss. A two-phase curriculum first maximizes the fidelity relative to the breakeven threshold over a short time horizon, then refines the policy for longer evolution. The reported result is a code with logical states |0L>=|4> and |1L>=|7>, with a nearest-neighbor engineered Lindblad operator of Hamiltonian distance d=1, which the authors claim outperforms the T4C, binomial, and prior RL codes and remains above breakeven at γa t=0.6 and beyond. The paper also presents a semi-analytical solver for the reduced master equation, claims a speed-up relative to QuTiP, analyzes robustness to phase and amplitude damping, and sketches an experimental implementation.

Significance. If the central claim holds, the paper would be a useful contribution to the growing area of RL-designed bosonic AQEC codes: it identifies a simple, low-nonlinearity code that is robust against double-photon loss, and it demonstrates a curriculum-training strategy that improves long-time fidelity. The machine-checkable and reproducible parts are genuine strengths: the GitHub repository is referenced, the comparison against established codes is concrete, and the Wigner-function visualizations support the reported fidelities within the chosen model. However, the significance is currently bounded by two unresolved technical points: the analytical solver is not validated against an independent master-equation integration, and the explicit analytic formula in Eq. (38) contains an arithmetic inconsistency. Until these are fixed, the headline claims are conditional on approximations inherited from Ref. [43].

major comments (4)
  1. [§II.B and Appendix C] The analytical solver is never validated against a direct numerical integration of the full master equation (4). Table I reports only wall-clock times; no fidelity trajectories or density-matrix comparisons are shown. Because the RL rewards r1 and r2 (Eqs. 19 and 21) are computed from this solver, any systematic error in the adiabatic elimination leading to Eq. (5), or in the eigendecomposition, propagates directly into every reported advantage. The implementation simulation in Fig. 9 uses γb/γa=10 and g0/γb≈60, which violate the assumptions g,γa≪γb under which Eq. (5) was derived, so it cannot serve as the missing validation. Please add a QuTiP master-equation comparison of fidelities for the test parameters and for at least one trained trajectory.
  2. [Eq. (38)] The stated equality u = 11/56 − √2/14 + (27/112)η ≈ (7.44 + 241.07η)×10^-3 is arithmetically inconsistent. For η=0, 11/56 − √2/14 ≈ 0.0954, which differs from 7.44×10^-3 by a factor of about 12.8. The bound u<26.7×10^-3 and the mean-fidelity expression Eq. (39) rely on this numerical approximation. Since Eq. (39) is presented as the analytic explanation of the code's protection, this error is load-bearing and needs to be corrected or the derivation clarified.
  3. [§II.B, Eq. (5)] All training and performance claims are based on the reduced master equation dρa/dt = (γa/2)D[a] + (γaλ/2)D[Leng], which is inherited from Ref. [43] under the assumptions g,γa≪γb and γa≪g. The paper does not independently test this reduction against the full master equation (4) for the discovered code |4>,|7>, nor does it show from first principles how the multi-photon terms in Eq. (10) are included consistently. Given that the RL environment itself is this reduced equation, the 'state-of-the-art' claim is conditional on the validity of that approximation; please provide either a derivation or a numerical check in the relevant parameter regime.
  4. [§III and Appendix B] The claim that the RL agent 'discovers the optimal set of codewords' should be qualified. The reward functions r1=f1ε and r2=f1ε+f2α directly maximize the fidelity relative to the breakeven threshold, so finding a high-fidelity code is the optimization target rather than an independent prediction. Moreover, the search is restricted by the disjoint-Fock-support ansatz in Eq. (13) and by the nearest-neighbor form of Lo in Eq. (14). The comparison with T4C, binomial, and prior RL codes is still valuable, but the word 'optimal' is not supported beyond this restricted class; please soften the claim or provide a broader search.
minor comments (6)
  1. [Appendix B vs. §II.D] There is an inconsistency in the reward scale: the main text states f1=250 and f2=2, while Appendix B gives r=50ε in phase 1 and r=250ε+2α in phase 2. Please clarify which values were used.
  2. [Table I] The text says the analytical solver achieves 'nearly twofold acceleration' at small scales, but Table I shows 0.55 s vs 0.30 s, a factor of 1.8. Also, the statement that the solver accelerates training by 40% is not fully explained; please specify the total training-time breakdown.
  3. [§III] The phrase 'state-of-art performance' in the abstract and conclusion is stronger than what is demonstrated, since only four comparison codes are considered. Please temper the claim (e.g., 'outperforms the compared codes in this model').
  4. [Fig. 5 caption] The caption says the fidelity distribution is solved with step π/10 and π/20 respectively; it would be clearer to state which step is used for θ and which for φ.
  5. [References] Reference [49] appears to be unrelated to the statement about double-photon loss rates of 1%–10% of γa; please check and replace with the appropriate source.
  6. [Eq. (1)] The sentence following Eq. (1), 'surpassing the break-even threshold' appears incomplete in the manuscript text; please rephrase for clarity.

Circularity Check

0 steps flagged

No significant circularity: the code is the output of the stated optimization, not a prediction derived from its own inputs.

full rationale

The paper's derivation chain is not circular. The effective master equation, Eq. (5), is taken from Ref. [43] under explicitly stated conditions (g,γa ≪ γb and γa ≪ g); Ref. [43] is an external prior work, not a self-citation by the present authors, so the inheritance is an external modeling assumption rather than a self-referential loop. The diagonal decomposition leading to Eq. (10) is a genuine algebraic reduction of Eq. (5), and the eigen-expansion ρ(m)(t)=Σ_l c_l e^{w_l t}v_l is a standard linear-ODE solution method; it does not assume the discovered code. The RL reward functions r1 = f1ε and r2 = f1ε + f2α (Eqs. 19, 21) directly maximize fidelity relative to the breakeven threshold, so the later statement that the GRL code 'surpasses the breakeven' is the reported value of the optimization objective. That is the output of a search, not a fitted parameter renamed as a prediction, and no equation is reduced to its own input by construction. The analytical formula for u, Eq. (38), and the resulting mean fidelity F̄ = 2/3 + (1/3)e^{-uγat}, Eq. (39), are post-hoc explanations of the discovered |4⟩,|7⟩ code; they are not used to force the code. The main weaknesses are validation gaps: Appendix C compares only wall-clock times and not fidelity against QuTiP, and the implementation Hamiltonian in Eq. (36) uses parameters outside the strict validity regime of Eq. (5). These are correctness risks, not circularity. There is no load-bearing self-citation, no imported uniqueness theorem, no ansatz smuggled via the authors' own prior work, and no renaming of a known result. The central claim is an optimization result within an externally inherited model.

Axiom & Free-Parameter Ledger

5 free parameters · 6 axioms · 0 invented entities

The central claim rests on the effective single-mode master equation inherited from Ref [43], on a heavily restricted encoding search (disjoint Fock partitions and nearest-neighbor Lo), and on hand-chosen RL hyperparameters. No new physical entities are introduced. The paper's own Appendices D and E document where the discovered operator coefficients were manually simplified and where an alternative reward failed.

free parameters (5)
  • Reward normalization factors f1, f2 = f1=250, f2=2 in main text; Appendix B states phase-1 reward r=50ε and phase-2 r=250ε+2α
    Hand-chosen hyperparameters determine which codes are favored; main text and Appendix B disagree on the phase-1 scale.
  • Fock truncation N = 7 (Sec. II.C), 8 (Appendix C)
    Truncation bounds the search space; 'optimal code' is only optimal up to N=7/8.
  • Lo equal-weight simplification ξ = ξ=1
    RL outputs time-dependent projection coefficients (Appendix D); authors fix Lo ∝ |3><2|+|4><3|+|6><5|+|7><6|, choosing equal weights. Robustness is tested for ξ∈[0.5,1.3].
  • Action-consistency threshold = 0.97
    Hand-chosen tolerance for freezing actions; affects phase-2 training dynamics.
  • Phase-2 below-breakeven penalty = -20
    Hand-chosen penalty for crossing below breakeven; shapes which policies survive.
axioms (6)
  • domain assumption Adiabatic elimination and ancilla ground-state ansatz ρ(t)=ρa(t)⊗|0><0| leading to Eq. (5)
    Central to all simulations; conditions g,γa≪γb, γa≪g stated in Sec. II.B but not validated for the reported parameters.
  • domain assumption Born-Markov/Lindblad master equation (4) with γa/2 D[a] + γb/2 D[σ-]
    Underlying open-system model taken from Ref [43]; no microscopic derivation in this paper.
  • ad hoc to paper Fock-state partition restriction: logical states have disjoint Fock support
    Sec. II.C assumes all Fock states are assigned exclusively to |0L> or |1L>; this excludes codes with shared basis states and is justified only by training complexity.
  • ad hoc to paper Lo restricted to nearest-neighbor ladder operators (Eq. 14)
    The 'cascading error-correction' restriction forces Hamiltonian distance d=1; the claimed optimality is within this class.
  • domain assumption Mean fidelity over the Bloch sphere and breakeven defined by an unencoded Fock pair
    Used for both reward and evaluation; depends on the chosen input-state distribution and breakeven definition.
  • domain assumption Phase-damping and amplitude-damping noise models (Eqs. 25-32) from Ref [56]
    Robustness claims rely on these specific spectral densities and parameter ranges.

pith-pipeline@v1.3.0-alltime-deepseek · 19330 in / 21858 out tokens · 175577 ms · 2026-08-03T22:01:34.398281+00:00 · methodology

0 comments
read the original abstract

Quantum error correction is essential for fault-tolerant quantum computing. However, standard methods relying on active measurements may introduce additional errors. Autonomous quantum error correction (AQEC) circumvents this by utilizing engineered dissipation and drives in bosonic systems, but identifying practical encoding remains challenging due to stringent Knill-Laflamme conditions. In this work, we utilize curriculum learning enabled deep reinforcement learning to discover Bosonic codes under approximate AQEC framework to resist both single-photon and double-photon losses. We present an analytical solution of solving the master equation under approximation conditions, which can significantly accelerate the training process of reinforcement learning. The agent first identifies an encoded subspace surpassing the breakeven point through rapid exploration within a constrained evolutionary time-frame, then strategically fine-tunes its policy to sustain this performance advantage over extended temporal horizons. We find that the two-phase trained agent can discover the optimal set of codewords, i.e., the Fock states $\ket{4}$ and $\ket{7}$ considering the effect of both single-photon and double-photon loss. We identify that the discovered code surpasses the breakeven threshold over a longer evolution time and achieve the state-of-art performance. We also analyze the robustness of the code against the phase damping and amplitude damping noise. Our work highlights the potential of curriculum learning enabled deep reinforcement learning in discovering the optimal quantum error correct code especially in early fault-tolerant quantum systems.

Figures

Figures reproduced from arXiv: 2511.12482 by Guihua Zeng, Jianping Fan, Ming He, Tailong Xiao, Xiaoyang Deng, Yue Yin.

Figure 1
Figure 1. Figure 1: Schematic diagram of a typical AQEC system, in￾cluding storage cavity A, transmon ancilla q and readout R. For the state |A, q, R⟩, the AQEC process can be sum￾marized as follows, as depicted in Fig.2: 1. An error happens (arrow 1): |code, g, 0⟩ → |error, g, 0⟩ 2. The AQEC pump drives the transition (arrow 2 and 3): |error, g, 0⟩ → |error, e, 0⟩ → |code, g, 1⟩ 3. The readout R decays, returning the state t… view at source ↗
Figure 2
Figure 2. Figure 2: Hardware-module energy level diagram of approx￾imate AQEC process. This diagram illustrates the physi￾cal ”pump-and-dump” mechanism for a single recovery cycle. The labels 1-4 depict the error and recovery process of an AQEC code [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Definition of ρ (m) in the case of N = 3 The equation for each diagonal element can be written as: d d(γαt) ρ (m) = [∑ N n=1 (ηnA (m,n) ) + λL (m) ]ρ (m) (10) [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: The diagram of training the PPO agent with the quantum environment. a The training process of GRL code in each episode. The training of the agent lasts for 300k episodes. In each episode, the quantum environment first creates the six basic quantum states s1 according to the codeword generated by its action a1 in the first step, and takes the st−1 as the initial states in the following steps. These states e… view at source ↗
Figure 5
Figure 5. Figure 5: a The fidelity distribution F(θ, ϕ, t) of other codes (T4C [18], Binomial [9] and RL [43] code) and GRL code solved as the function of Bloch angle θ and ϕ analytically with step π/10 and π/20 respectively. The result is shown with γat = 0.6 and λ = 104 . b Plot of average reward(solid blue line) as function of train episode. The evolution time for each step is γat = 0.06 c Comparison of AQEC performance of… view at source ↗
Figure 6
Figure 6. Figure 6: a-d Wigner function of six cardinal-point states of GRL code and breakeven before and after γat = 4.2 evolution. a GRL code before evolution. b GRL code after evolution with AQEC. c Breakeven before evolution. d Breakeven after free evolution. The performance of the GRL code can also be better understood through the KL condition. The KL condi￾tion, which imposes two fundamental rules: ⟨0L|L † nat,iLnat,j |… view at source ↗
Figure 7
Figure 7. Figure 7: a-c The fidelity comparison between different ωc with s = 2 (a), s = 3.7 (b) and s = 5 (c). The phase damping noise is determined by 2 factors: ωc and s. ωc is the reservoir cutoff frequency and s is the ohmicity parameter. breakeven). As shown in these figures, the GRL code maintains a clearer and more distinct structure than the breakeven code, higher fidelity observed for the GRL code. Specifically, the… view at source ↗
Figure 8
Figure 8. Figure 8: Fidelity evolution with amplitude damping. Re￾sults with different coupling coefficients are compared with breakeven, GRL code without noise and GRL code without both noise and error correction. The detuning frequency and width of Lorentzian distribution are 100kHz and 140kHz in (a) and 1kHz and 40Hz in (b) respectively. Another significant noise source is amplitude damping, [PITH_FULL_IMAGE:figures/full_… view at source ↗
Figure 9
Figure 9. Figure 9: The fidelity of GRL code considering the coupling engineering. The parameters here are γa = 0.2×2πkHz, γa2 = 2 × 2πHz, γb = 2 × 2πkHz, γc = 0.24 × 2πMHz, g0 = 0.12 × 2πMHz, g1 = 0.16 × 2πMHz, ωa = 3.5 × 2πGHz, ωb = ωc = 5 × 2πGHz From the simulation with the Heff in Equ.36, the Fig.9 shows the GRL code surpasses the breakeven, with the gain G = (1 − F¯ be)/(1 − F¯) ≈ 2.64 at t = 3ms. The analytical solutio… view at source ↗
Figure 10
Figure 10. Figure 10: The mean fidelity of RL, T4C and Bino￾mial codes under different second-order conditions (γa2/γa = 0, 0.04, 0.08). The performance of each code is evaluated un￾der the low coupling rate (g/γa = 600) and high coupling rate (g/γa = 900). The high coupling rate shows little levitation in the mean fidelity. To validate the nontrivial impact of second-order pho￾ton loss on fidelity, we evaluated how fidelity v… view at source ↗
Figure 11
Figure 11. Figure 11: Mean fidelity as the function of γat with ξ = 0.5, 1.0, 1.3. The red shaded region represents the ±1% error band surrounding the ξ = 1.0 curve. The numerical results of the GRL code in Tab.II is time-dependent with different coefficient for some pro￾jection operators. The coefficients are evolving through the time, which is challenging for real implementation. However, tests in Fig.11 shows that consideri… view at source ↗
Figure 12
Figure 12. Figure 12: The mean fidelity as the function of time step. Each step is γat = 0.06 long. An alternative reward structure based on the incre￾mental fidelity improvement can be ∆Fk = Fk − Fk−1, which rewards the RL agent for fidelity gains while pe￾nalizing declines. Although intuitively appealing, this approach was found ineffective in practice. Specifically, when implemented, the RL agent quickly learned to arti￾fic… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

62 extracted references · 10 linked inside Pith

  1. [1]

    An error happens (arrow 1): |code,g, 0⟩ → | error,g, 0⟩

  2. [2]

    The AQEC pump drives the transition (arrow 2 and 3): |error,g, 0⟩ → | error,e, 0⟩ → | code,g, 1⟩

  3. [3]

    Hardware-module energy level diagram of approx- imate AQEC process

    The readout R decays, returning the state to code space (arrow 4): |code,g, 1⟩ → | code,g, 0⟩ 1 2 3 4 Figure 2. Hardware-module energy level diagram of approx- imate AQEC process. This diagram illustrates the physi- cal ”pump-and-dump” mechanism for a single recovery cycle. The labels 1-4 depict the error and recovery process of an AQEC code. 3 AQEC has s...

  4. [4]

    M. A. Nielsen and I. L. Chuang, Quantum Computation and Quantum Information: 10th Anniversary Edition (Cambridge University Press, 2010)

    2010

  5. [5]

    Asfaw, A

    A. Asfaw, A. Megrant, C. Jones, C. Gidney, D. Bacon, D. Debroy, D. Kafri, E. Lucero, H. Neven, J. Hilton, et al. , Suppressing quantum errors by scaling a surface code logical qubit, Nature 614, 676 (2023)

    2023

  6. [6]

    B. M. Terhal, Quantum error correction for quantum memories, Reviews of Modern Physics 87, 307 (2015) , publisher: American Physical Society

    2015

  7. [7]

    Schindler, J

    P. Schindler, J. T. Barreiro, T. Monz, V. Nebendahl, D. Nigg, M. Chwalla, M. Hennrich, and R. Blatt, Exper- imental Repetitive Quantum Error Correction, Science 332, 1059 (2011) , publisher: American Association for the Advancement of Science

    2011

  8. [8]

    Chiaverini, D

    J. Chiaverini, D. Leibfried, T. Schaetz, M. D. Barrett, R. B. Blakestad, J. Britton, W. M. Itano, J. D. Jost, E. Knill, C. Langer, R. Ozeri, and D. J. Wineland, Re- alization of quantum error correction, Nature 432, 602 (2004), publisher: Nature Publishing Group

    2004

  9. [9]

    L. Hu, Y. Ma, W. Cai, X. Mu, Y. Xu, W. Wang, Y. Wu, H. Wang, Y. P. Song, C.-L. Zou, S. M. Girvin, L.-M. Duan, and L. Sun, Quantum error correction and uni- versal gate set operation on a binomial bosonic logical qubit, Nature Physics 15, 503 (2019) , number: 5 Pub- lisher: Nature Publishing Group

    2019

  10. [10]

    N. Ofek, A. Petrenko, R. Heeres, P. Reinhold, Z. Leghtas, B. Vlastakis, Y. Liu, L. Frunzio, S. M. Girvin, L. Jiang, M. Mirrahimi, M. H. Devoret, and R. J. Schoelkopf, Ex- tending the lifetime of a quantum bit with error correc- tion in superconducting circuits, Nature 536, 441 (2016) , publisher: Nature Publishing Group

    2016

  11. [11]

    Acharya, L

    R. Acharya, L. Aghababaie-Beni, I. Aleiner, T. I. Ander- sen, M. Ansmann, F. Arute, K. Arya, A. Asfaw, N. As- trakhantsev, J. Atalaya, et al. , Quantum error correc- tion below the surface code threshold, arXiv preprint arXiv:2408.13687 (2024)

    Pith/arXiv arXiv 2024

  12. [12]

    D. Gao, D. Fan, C. Zha, J. Bei, G. Cai, J. Cai, S. Cao, X. Zeng, F. Chen, J. Chen, et al. , Establishing a new benchmark in quantum computational advantage with 105-qubit zuchongzhi 3.0 processor, arXiv preprint arXiv:2412.11924 (2024)

    Pith/arXiv arXiv 2024

  13. [13]

    Campagne-Ibarcq, A

    P. Campagne-Ibarcq, A. Eickbusch, S. Touzard, E. Zalys- Geller, N. E. Frattini, V. V. Sivak, P. Reinhold, S. Puri, S. Shankar, R. J. Schoelkopf, L. Frunzio, M. Mirrahimi, and M. H. Devoret, Quantum error correction of a qubit encoded in grid states of an oscillator, Nature 584, 368 (2020), publisher: Nature Publishing Group

    2020

  14. [14]

    Y.-H. Chen, W. Qin, R. Stassi, X. Wang, and F. Nori, Fast binomial-code holonomic quantum computation with ultrastrong light-matter coupling, Physical Review Research 3, 033275 (2021) , publisher: American Physical Society

    2021

  15. [15]

    Y. Zhao, Y. Ye, H.-L. Huang, Y. Zhang, D. Wu, H. Guan, Q. Zhu, Z. Wei, T. He, S. Cao, F. Chen, T.-H. Chung, H. Deng, D. Fan, M. Gong, C. Guo, S. Guo, L. Han, N. Li, S. Li, Y. Li, F. Liang, J. Lin, H. Qian, H. Rong, H. Su, L. Sun, S. Wang, Y. Wu, Y. Xu, C. Ying, J. Yu, C. Zha, K. Zhang, Y.-H. Huo, C.-Y. Lu, C.-Z. Peng, X. Zhu, and J.-W. Pan, Realization of...

    2022

  16. [16]

    Y. Ma, Y. Xu, X. Mu, W. Cai, L. Hu, W. Wang, X. Pan, H. Wang, Y. P. Song, C.-L. Zou, and L. Sun, Error- transparent operations on a logical qubit protected by quantum error correction, Nature Physics 16, 827 (2020) , publisher: Nature Publishing Group. 12

    2020

  17. [17]

    K. C. Young, M. Sarovar, and R. Blume-Kohout, Error Suppression and Error Correction in Adiabatic Quantum Computation: Techniques and Challenges, Physical Re- view X 3, 041013 (2013) , publisher: American Physical Society

    2013

  18. [18]

    Atalaya, S

    J. Atalaya, S. Zhang, M. Y. Niu, A. Babakhani, H. C. H. Chan, J. M. Epstein, and K. B. Whaley, Continu- ous quantum error correction for evolution under time- dependent Hamiltonians, Physical Review A 103, 042406 (2021), publisher: American Physical Society

    2021

  19. [19]

    J. P. Barnes and W. S. Warren, Automatic quantum error correction, Phys. Rev. Lett. 85, 856 (2000)

    2000

  20. [20]

    D. R. Pérez and E. Kapit, Improved autonomous er- ror correction using variable dissipation in small logical qubit architectures, Quantum Science and Technology 6, 015006 (2020) , publisher: IOP Publishing

    2020

  21. [21]

    L. Egan, D. M. Debroy, C. Noel, A. Risinger, D. Zhu, D. Biswas, M. Newman, M. Li, K. R. Brown, M. Cetina, and C. Monroe, Fault-tolerant control of an error- corrected qubit, Nature 598, 281 (2021) , publisher: Na- ture Publishing Group

    2021

  22. [22]

    A. G. Fowler, M. Mariantoni, J. M. Martinis, and A. N. Cleland, Surface codes: Towards practical large-scale quantum computation, Physical Review A 86, 032324 (2012), publisher: American Physical Society

    2012

  23. [23]

    J. M. Gertler, B. Baker, J. Li, S. Shirol, J. Koch, and C. Wang, Protecting a bosonic qubit with autonomous quantum error correction, Nature 590, 243 (2021) , num- ber: 7845 Publisher: Nature Publishing Group abstract- Translation:

    2021

  24. [24]

    Hillmann and F

    T. Hillmann and F. Quijandría, Quantum error cor- rection with dissipatively stabilized squeezed-cat qubits, Physical Review A 107, 032423 (2023) , publisher: Amer- ican Physical Society

    2023

  25. [25]

    W. Cai, Y. Ma, W. Wang, C.-L. Zou, and L. Sun, Bosonic quantum error correction codes in superconducting quan- tum circuits, Fundamental Research 1, 50 (2021)

    2021

  26. [26]

    A. M. Steane, Error Correcting Codes in Quantum The- ory, Physical Review Letters 77, 793 (1996) , publisher: American Physical Society

    1996

  27. [27]

    J.-M. Lihm, K. Noh, and U. R. Fischer, Implementation- independent sufficient condition of the Knill-Laflamme type for the autonomous protection of logical qudits by strong engineered dissipation, Physical Review A 98, 012317 (2018) , publisher: American Physical Society

    2018

  28. [28]

    Zanardi, J

    P. Zanardi, J. Marshall, and L. Campos Venuti, Dissipa- tive universal Lindbladian simulation, Physical Review A 93, 022312 (2016)

    2016

  29. [29]

    E. Kapit, Hardware-Efficient and Fully Autonomous Quantum Error Correction in Superconducting Circuits, Physical Review Letters 116, 150501 (2016) , publisher: American Physical Society

    2016

  30. [30]

    Faist, S

    P. Faist, S. Nezami, V. V. Albert, G. Salton, F. Pastawski, P. Hayden, and J. Preskill, Continuous Symmetries and Approximate Quantum Error Correc- tion, Physical Review X 10, 041018 (2020) , publisher: American Physical Society

    2020

  31. [31]

    Krastanov, M

    S. Krastanov, M. Heuck, J. H. Shapiro, P. Narang, D. R. Englund, and K. Jacobs, Room-temperature pho- tonic logical qubits via second-order nonlinearities, Na- ture Communications 12, 191 (2021) , publisher: Nature Publishing Group

    2021

  32. [32]

    Reiter, A

    F. Reiter, A. S. Sørensen, P. Zoller, and C. A. Muschik, Dissipative quantum error correction and application to quantum sensing with trapped ions, Nature Communi- cations 8, 1822 (2017) , publisher: Nature Publishing Group

    2017

  33. [33]

    Kerckhoff, H

    J. Kerckhoff, H. I. Nurdin, D. S. Pavlichin, and H. Mabuchi, Designing Quantum Memories with Embed- ded Control: Photonic Circuits for Autonomous Quan- tum Error Correction, Physical Review Letters 105, 040502 (2010) , publisher: American Physical Society

    2010

  34. [34]

    D. W. Leung, M. A. Nielsen, I. L. Chuang, and Y. Ya- mamoto, Approximate quantum error correction can lead to better codes, Physical Review A 56, 2567 (1997) , pub- lisher: American Physical Society

    1997

  35. [35]

    Fösel, P

    T. Fösel, P. Tighineanu, T. Weiss, and F. Marquardt, Reinforcement Learning with Neural Networks for Quan- tum Feedback, Physical Review X 8, 031084 (2018) , pub- lisher: American Physical Society

    2018

  36. [36]

    Kong and Z.-W

    L. Kong and Z.-W. Liu, Near-Optimal Covariant Quan- tum Error-Correcting Codes from Random Unitaries with Symmetries, PRX Quantum 3, 020314 (2022) , pub- lisher: American Physical Society

    2022

  37. [37]

    Bény, Perturbative Quantum Error Correction, Physi- cal Review Letters 107, 080501 (2011) , publisher: Amer- ican Physical Society

    C. Bény, Perturbative Quantum Error Correction, Physi- cal Review Letters 107, 080501 (2011) , publisher: Amer- ican Physical Society

    2011

  38. [38]

    Bény and O

    C. Bény and O. Oreshkov, General Conditions for Ap- proximate Quantum Error Correction and Near-Optimal Recovery Channels, Physical Review Letters 104, 120501 (2010), publisher: American Physical Society

    2010

  39. [39]

    T. Xiao, J. Huang, H. Li, J. Fan, and G. Zeng, Intel- ligent certification for quantum simulators via machine learning, npj Quantum Information 8, 138 (2022)

    2022

  40. [40]

    T. Xiao, J. Huang, H. Li, J. Fan, and G. Zeng, Quantum generative adversarial imitation learning, New Journal of Physics 25, 033034 (2023)

    2023

  41. [41]

    Y. Baum, M. Amico, S. Howell, M. Hush, M. Liuzzi, P. Mundada, T. Merkh, A. R. Carvalho, and M. J. Biercuk, Experimental Deep Reinforcement Learning for Error-Robust Gate-Set Design on a Superconducting Quantum Computer, PRX Quantum 2, 040324 (2021) , publisher: American Physical Society

    2021

  42. [42]

    Bolens and M

    A. Bolens and M. Heyl, Reinforcement Learning for Dig- ital Quantum Simulation, Physical Review Letters 127, 110502 (2021) , publisher: American Physical Society

    2021

  43. [43]

    that γa,g ≪ γb, γa ≪ g, and γat ≪ 1,λ ≫ 24, the approximate solution of the first eight Fock states of the storage cavity is ρa(t) ≈   0 · · · 0 0 0 0 ... . . . ... ... ... ... 0 · · · ρ44(0) 0 0 ρ47(0)e−uγat 0 · · · 0 0 0 0 0 · · · 0 0 0 0 0 · · · ρ74(0)e−uγat 0 0 ρ77(0)   (37) The ρij(0) represents the initial states. The protecting 11 f...

  44. [44]

    T. Xiao, J. Fan, and G. Zeng, Parameter estimation in quantum sensing based on deep reinforcement learning, npj Quantum Information 8, 2 (2022)

    2022

  45. [45]

    J. Olle, R. Zen, M. Puviani, and F. Marquardt, Si- multaneous discovery of quantum error correction codes and encoders with a noise-aware reinforcement learning agent, npj Quantum Information 10, 1 (2024) , publisher: Nature Publishing Group

    2024

  46. [46]

    M. A. Webster and D. E. Browne, Engineering quan- tum error correction codes using evolutionary algorithms, IEEE Transactions on Quantum Engineering 6, 1 (2025)

    2025

  47. [47]

    V. P. Su, C. Cao, H.-Y. Hu, Y. Yanay, C. Tahan, and B. Swingle, Discovery of optimal quantum codes via rein- 13 forcement learning, Phys. Rev. Appl. 23, 034048 (2025)

    2025

  48. [48]

    Zeng, Z.-Y

    Y. Zeng, Z.-Y. Zhou, E. Rinaldi, C. Gneiting, and F. Nori, Approximate Autonomous Quantum Error Cor- rection with Reinforcement Learning, Physical Review Letters 131, 050601 (2023) , arXiv:2212.11651 [cond-mat, physics:quant-ph]

    Pith/arXiv arXiv 2023

  49. [49]

    Mauron, T

    C. Mauron, T. Farrelly, and T. M. Stace, Optimization of tensor network codes with reinforcement learning, New Journal of Physics 26, 023024 (2024)

    2024

  50. [50]

    V. V. Sivak, A. Eickbusch, B. Royer, S. Singh, I. Tsiout- sios, S. Ganjam, A. Miano, B. L. Brock, A. Z. Ding, L. Frunzio, S. M. Girvin, R. J. Schoelkopf, and M. H. De- voret, Real-time quantum error correction beyond break- even, Nature 616, 50 (2023) , number: 7955 Publisher: Nature Publishing Group

    2023

  51. [51]

    B. C. A. Freire, N. Delfosse, and A. Leverrier, Opti- mizing hypergraph product codes with random walks, simulated annealing and reinforcement learning (2025), arXiv:2501.09622 [quant-ph]

    Pith/arXiv arXiv 2025

  52. [52]

    V. P. Su, C. Cao, H.-Y. Hu, Y. Yanay, C. Tahan, and B. Swingle, Discovery of Optimal Quantum Error Correcting Codes via Reinforcement Learning (2023), arXiv:2305.06378 [quant-ph]

    Pith/arXiv arXiv 2023

  53. [53]

    for codes). III. RESUL T The analytical master equation solver offers a signifi- cant advantage due to its lower algorithmic complexity. Solving the master equation numerically, without any simplification, would typically involve a single differen- tial equation with (2N )2 pivots, where N represents the number of Fock states considered. This makes it com...

  54. [54]

    Y. Zeng, W. Qin, Y.-H. Chen, C. Gneiting, and F. Nori, Neural-network-based design of approximate gottesman- kitaev-preskill code, Physical Review Letters 134, 60601 (2025)

    2025

  55. [55]

    Tosca, M

    J. Tosca, M. C. Strinati, C. Conti, and C. Ciuti, Emer- gent equilibrium in all-optical single quantum-trajectory ising machines (2024), arXiv:2412.12768 [quant-ph]

    Pith/arXiv arXiv 2024

  56. [56]

    Z. Wang, T. Rajabzadeh, N. Lee, and A. H. Safavi- Naeini, Automated Discovery of Autonomous Quantum Error Correction Schemes, PRX Quantum 3, 020302 (2022)

    2022

  57. [57]

    Z. Li, T. Roy, D. R. Perez, K.-H. Lee, E. Kapit, and D. I. Schuster, Autonomous error correction of a single logi- cal qubit using two transmons (2023), arXiv:2302.06707 [quant-ph]

    Pith/arXiv arXiv 2023

  58. [58]

    com/astarrf/GRLcode

    GRLcode, GitHub - astarrf/GRLcode , https://github. com/astarrf/GRLcode

  59. [59]

    Z. Ni, S. Li, X. Deng, Y. Cai, L. Zhang, W. Wang, Z.- B. Yang, H. Yu, F. Yan, S. Liu, C.-L. Zou, L. Sun, S.- B. Zheng, Y. Xu, and D. Yu, Beating the break-even point with a discrete-variable-encoded logical qubit, Na- ture 616, 56 (2023) , arXiv:2211.09319 [quant-ph]

    Pith/arXiv arXiv 2023

  60. [60]

    Sarma and H

    G. Sarma and H. Mabuchi, Gauge subsystems, separa- bility and robustness in autonomous quantum memories, New Journal of Physics 15, 035014 (2013)

    2013

  61. [61]

    Dakir, A

    Y. Dakir, A. Slaoui, L. B. Drissi, and R. A. Laamara, Quantifying non-Markovianity via local quantum Fisher information (2024), arXiv:2409.10163

    Pith/arXiv arXiv 2024

  62. [62]

    Schulman, F

    J. Schulman, F. Wolski, P. Dhariwal, A. Radford, and O. Klimov, Proximal policy optimization algorithms (2017), arXiv:1707.06347 [cs.LG] . Appendix A: The effect of double-photon loss on the code fidelity Figure 10. The mean fidelity of RL, T4C and Bino- mial codes under different second-order conditions ( γa2/γa = 0, 0.04, 0.08). The performance of each ...

    Pith/arXiv arXiv 2017