Pith. sign in

REVIEW 4 major objections 5 minor 67 references

Convergence monitoring of quantum Gibbs samplers

T0 review · 4 major / 5 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read This paper proposes a Hamiltonian-agnostic stopping criterion for quantum Gibbs samplers based on the equilibrium shifted symmetry of quasi-frequencies recorded during weak measurements.

desk verdict A clean, honestly-scoped convergence diagnostic with a real gap between the analytical support and the moment-based rule; worth reviewing. read the letter →

arxiv 2608.02038 v1 pith:ASXFMO2O submitted 2026-08-03 quant-ph

Nikolaos Louloudis , Ruben Ibarrondo , Mikel Sanz , Mario Berta This is my paper
classification quant-ph
keywords quantumGibbssamplersconvergencemonitoringweakmeasurementquasi-frequencydistributiondetailedbalancebatch-meansstoppingrulethermalstatepreparationenergyrecoverability
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 convergence of a quantum Gibbs sampler can be monitored for free: record the quasi-frequencies the sampler already outputs during its weak measurements, and watch for a specific equilibrium symmetry. At thermal equilibrium, the distribution of these quasi-frequencies is symmetric about a computable center C = −βσ_E²/2, independently of the Hamiltonian. The authors turn this into a stopping rule that halts when a batch-means confidence ellipsoid for the mean and third centered moment fits inside a tolerance ellipsoid around (C, 0). They show numerically on small transverse-field Ising models that the monitored moments co-relax with energy and that the online rule reproduces temperature-dependent slowing. They explicitly present it as a practical diagnostic rather than a certificate of global mixing, and under simplifying assumptions they prove that quasi-frequency statistics determine the full energy distribution.

What carries the argument

The shifted reflection identity of Proposition 1 is the load-bearing object: it converts detailed balance at equilibrium into a testable property of the weak-measurement record. On top of it sits a batch-means multivariate stopping rule: from accepted quasi-frequencies ω_i, form Y_i = (ω_i, (ω_i−C)³), estimate the long-run covariance with non-overlapping batch means, and stop when the confidence ellipsoid is contained in a tolerance ellipsoid centered at θ* = (C, 0). A second analytic tool, Theorem 1, shows under Gaussian and uniform-prior assumptions that the quasi-frequency POVM spans the same operator space as the energy projectors, so the frequency record carries full energy-distribution

What would settle it

Find a concrete instance—a Hamiltonian, initial state, and sampler—where the quasi-frequency moments are within tolerance (C_α(M) ⊆ T_Λ) while a relevant observable such as energy or magnetization is still far from its Gibbs value; or analytically exhibit a family of states with identical quasi-frequency statistics but different energy distributions outside Theorem 1's assumptions. A numerical search across larger system sizes and alternative coupling families would suffice as evidence.

Watch

Extended reading notes

Core claim

At equilibrium the net energy flow between system and bath vanishes, and for the algorithmic Lindbladians used in modern quantum Gibbs samplers this balance survives in the weak-measurement record as a shifted reflection symmetry. Concretely (Proposition 1): with a Gaussian frequency window and a transition rate obeying γ(C−u) = e^{βu}γ(C+u), the equilibrium quasi-frequency density satisfies π(C−u) = π(C+u), with center C = −βσ_E²/2; hence the equilibrium mean quasi-frequency is known a priori, independent of the Hamiltonian. The paper converts this into an online stopping rule: from accepted quasi-frequencies ω_i it forms Y_i = (ω_i, (ω_i−C)³), builds a batch-means confidence ellipsoid for

Load-bearing premise

The monitored quasi-frequency moments co-relax with physically relevant observables such as energy; this is demonstrated numerically on small transverse-field Ising models, not proven, so a state whose slow modes are invisible to the frequency record could stop the rule prematurely.

Editorial extensions

If this is right

  • Practitioners can run quantum Gibbs samplers with a built-in, no-extra-cost stopping signal, avoiding repeated destructive state preparation at many monitoring times.
  • The stopping rule is Hamiltonian-agnostic: the target (C, 0) depends only on β and σ_E, so it can be deployed without spectral knowledge of the system.
  • For qubit-efficient samplers with a single reusable ancilla, the monitored signed-frequency record obeys the same symmetry, preserving their architectural advantages.
  • Under Theorem 1's assumptions, if two states yield the same quasi-frequency statistics they have the same energy distribution, so frequency-based stopping certifies energy convergence in that restricted setting.
  • The rules are asymptotically consistent output-analysis procedures: if the monitored moments converge inside the tolerances, the rule eventually passes almost surely, though this is not a finite-time global-mixing certificate.

Reading between the lines

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

  • Editorial extension: the same symmetry target could be checked as a full distributional test, such as histogram symmetry or a Kolmogorov–Smirnov-type comparison, rather than only first and third moments; this might catch higher-order non-equilibrium deviations earlier.
  • Editorial extension: Theorem 1's energy-to-frequency kernel suggests Hamiltonian-informed diagnostics: with approximate spectral data, one could invert G to estimate energy error or choose monitored features tailored to observables of interest.
  • Editorial extension: a natural stress test would be to search on larger or frustrated systems for slow modes that are invisible to the quasi-frequency record; if such modes outlive the stopping time, a second, observable-specific monitor would be needed.
  • Editorial extension: because the rule's eventual-passage property is asymptotic, it could be paired with independent single-trajectory estimators or interleaved measurements as a safeguard against premature stopping in practice.
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

4 major / 5 minor

Summary. The paper proposes a low-cost convergence-monitoring criterion for quantum Gibbs samplers (QGSs) based on the weak-measurement quasi-frequency record that the sampler already produces. The central analytical result, Proposition 1 (Sec. III.a), shows that for the Gaussian-filtered algorithmic Lindbladian, any transition rate satisfying the shifted detailed-balance identity γ(C−u)=e^{βu}γ(C+u), with C=−βσ_E^2/2, yields an equilibrium quasi-frequency density symmetric about C; hence the equilibrium mean is C and the third centered moment is zero. This target is Hamiltonian-agnostic. The paper then constructs a batch-means multivariate stopping rule (Sec. III.b, Algorithm 1) that stops when a confidence ellipsoid for the monitored moments (ω, (ω−C)^3) is contained in a tolerance ellipsoid around the known target. Numerical benchmarks on small transverse-field Ising models (n=3,5,7) show co-decay of a deterministic target distance with the energy error (Fig. 3) and reproduce the inverse-temperature ordering of an offline energy-relaxation time (Fig. 4). A further analytical result, Theorem 1 (Sec. IV), states that under Gaussian filter/rate and a normalized unitary 1-design condition on the couplings, the quasi-frequency operator density spans the same operator subspace as the energy projectors, so the full quasi-frequency distribution determines the energy distribution. The paper explicitly frames the criterion as a practical diagnostic rather than a global mixing certificate.

Significance. If the proposed criterion is reliable, it is a useful methodological contribution: it requires no additional quantum operations or ancillas, reuses data already generated by the sampler, and is Hamiltonian-agnostic at the level of the equilibrium target. Proposition 1 is cleanly derived, parameter-free, and extends to the linear-combination-of-Gaussians rate families; Theorem 1 is a nontrivial span result that gives a clear information-theoretic sense in which quasi-frequencies can carry full energy-distribution information under restrictive assumptions. The paper is also unusually honest about its limitations, explicitly acknowledging in App. F.h that the rule is not time-uniform and in App. F.g that moment-based calibration does not prove error bounds for every observable. The main weakness is that the practical stopping rule monitors only two low-order moments, and no quantitative bound connects those moments to the physical convergence of the energy; the numerical evidence is limited to three small Ising instances and per-family calibration of tolerances. These gaps are load-bearing for the paper's central claim.

major comments (4)
  1. [Sec. III.b, Eq. (16)-(21), Algorithm 1; App. F.g] The stopping rule controls only the mean and third centered moment of the accepted-event quasi-frequency record. These are two linear functionals of the state-dependent transition density, and the set of states with exactly the equilibrium values of these two moments is generically large. No inequality of the form |Tr[H(ρ−ρβ)]| ≤ L ||θ(ρ)−θ*|| is stated or proved, and Fig. 3 provides only numerical co-decay for n=3,5,7 Ising models. The paper's own App. F.g concedes that calibration 'does not by itself prove that passing a moment-based rule bounds the error of every observable.' This is load-bearing because the practical claim is that the rule reliably monitors convergence of physically relevant observables such as energy. The authors should either provide a quantitative Lipschitz-type bound under stated assumptions, exhibit and analyze the failure modes explicitly, or substantially soft
  2. [Sec. IV.d-e and Sec. III.c-d] Theorem 1 is proved under Gaussian filter and Gaussian rate (Eqs. (3) and (8)) and under the uniform-prior/1-design condition (Eq. (39)). The numerical benchmark in Sec. III.c-d uses the Metropolis rate (Eq. (7)) and couplings {X_i/√(2n), Z_i/√(2n)}, which are neither Gaussian nor a 1-design (for example, summing A_a^† Y_j A_a over these couplings does not give a multiple of the identity). Thus Theorem 1 does not analytically support the numerical setup it is juxtaposed with. The paper should either run the benchmark in the Gaussian/1-design setting, prove an analogue of Theorem 1 for the Metropolis rate and the actual coupling family, or explicitly state that the numerics and Theorem 1 concern different regimes and that the theorem is only motivational.
  3. [Sec. III.c-d, Table I, Eq. (34)] The tolerance P_Λ is calibrated per model family using the deterministic energy-error curves (Table I), and the benchmark scale a_m in Eq. (34) is also chosen per model family from the same stopping-time and energy-relaxation data. This weakens the 'Hamiltonian-agnostic' claim: the user must know the model family and run an offline energy-relaxation study before applying the rule. More importantly, the claim that the rule 'reproduces the temperature ordering' is made after this calibration; while a positive multiplicative scale cannot change the ordering, P_Λ itself can affect which cases stop early or late. The authors should report sensitivity of Fig. 4 to reasonable variations of P_Λ and a_m, and ideally provide a principled, parameter-free choice for P_Λ or an explicit calibration protocol that does not use the same energy signal being predicted.
  4. [Sec. III.b and App. F.h] Algorithm 1 inspects at random checkpoints and stops at the first passage. The asymptotic confidence statements in App. F apply to each fixed checkpoint, not to the random stopping time. The paper correctly acknowledges that the rule is not a time-uniform confidence sequence, but this is a central limitation for a stopping procedure: the probability of a false stop is not controlled. The authors should state this caveat in the main text near Algorithm 1 and, if possible, report empirical false-positive rates (e.g., the fraction of trajectories that stop after the energy has actually converged) on the benchmark models.
minor comments (5)
  1. [Sec. III.b, after Eq. (21)] The phrase 'In the simulations below, we prefer the latter' is informal and the typography is inconsistent. Please rephrase and clarify whether the online estimate of Λ is the default choice for the reported trajectory results.
  2. [Fig. 3] The caption says color gives the physical Lindblad time, but no color scale or colorbar is visible in the figure. Please add a colorbar or use a clear grayscale/linestyle mapping so the reader can identify t values.
  3. [Sec. III.d, Eq. (33)] The confidence interval notation uses t_{0.975,N_runs−1}; please define the quantile notation explicitly or use the standard Student-t notation consistently.
  4. [App. F, Eq. (F26)] The one-dimensional maximization over φ is stated without details on how the maximum is found. Please add a sentence describing the numerical procedure (e.g., dense grid followed by local refinement) and its tolerance.
  5. [General] The paper does not provide the simulation code or data. Given the empirical nature of the central validation, a code repository or a statement about availability would improve reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the equilibrium target is derived parameter-free, the stopping rule is a disclosed practical calibration, and the analytical span result is an independent theorem.

full rationale

The paper's central target — the equilibrium quasi-frequency center θ*=(C,0) — is obtained from Proposition 1, which is a genuine mathematical consequence of the KMS/detailed-balance condition and the Gaussian filter identity, with no fitted data entering the derivation. The stopping rule then tests the sample mean and third centered moment against this known target; no fitted target is used inside the criterion itself. The numerical benchmark does calibrate the per-family tolerance P_Λ and a multiplicative scale a_m against the energy relaxation being studied, but the paper explicitly limits the benchmark claim to temperature ordering and notes that a single constant per model family cannot impose the within-family β-ordering; the relevant passages (Sec. III.d, Eq. (34)) disclose rather than hide this calibration. The analytical content of Theorem 1 is a span-equality statement proven under explicitly stated Gaussian and 1-design assumptions; it does not assume the moment rule or the energy convergence it is meant to monitor. The paper repeatedly and explicitly disclaims that the rule is a global mixing certificate (Sec. III.e, App. F.h) and concedes that calibration alone does not prove a moment-based rule bounds every observable (App. F.g). These are honest limitations, not circular reasoning. Citations to prior sampler constructions are external and not load-bearing in a way that reduces the derivation to self-citation. No equation or fitted parameter is demonstrably equivalent to its own output by construction, so the circularity score is 0.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The core derivation rests on prior QGS constructions and standard detailed-balance conditions. The free parameters are mostly sampler design choices and benchmark calibration tolerances; no new physical entities are introduced. The strongest extra assumption for Theorem 1 is the uniform prior (unitary 1-design), which is stated explicitly.

free parameters (3)
  • frequency resolution sigma_E = 1 in numerics
    Chosen by hand as a sampler design parameter; the equilibrium center C = -beta sigma_E^2/2 depends on it, but it is not fitted to the target.
  • tolerance radius P_Lambda = 2.0e-2 (Ising), 1.3e-2 (fully connected)
    Calibrated per model family against the Gibbs energy error to achieve a desired empirical operating point (Table I, Sec. III.d). It directly controls when the online rule stops.
  • benchmark calibration scale a_m = median_c S_c/O_c per model family
    Introduced in Eq. (34) to convert sampler steps to Lindblad time when comparing S_c with O_c. It preserves ordering but is a fitted scale used only for the benchmark comparison.
assumptions (6)
  • domain assumption The algorithmic Lindbladians of [16-18,24] satisfy KMS detailed balance and have the Gibbs state as a stationary state.
    Invoked in Sec. II.a, Eq. (6), as the foundation for the quasi-frequency record and its equilibrium symmetry.
  • domain assumption Ergodicity of the Lindbladian dynamics, so that the evolution converges to the Gibbs state.
    Explicitly stated in Sec. II.a as the usual assumption under which convergence holds.
  • domain assumption The Gaussian window and the reflection identity gamma(C-u)=e^{beta u} gamma(C+u) hold for the transition rate.
    Needed for Proposition 1 (Sec. III.a, Eq. (13)); satisfied by the Gaussian and Metropolis families and the linear-combination-of-Gaussians family.
  • domain assumption The accepted-event process satisfies a Markov-chain central limit theorem and the batch-means covariance estimator is consistent.
    Assumed in Appendix F, Eq. (F3), as the statistical basis for the confidence ellipsoids and stopping rules.
  • domain assumption The coupling operators satisfy the uniform prior condition sum_a A_a^dagger X A_a = Tr(X)/d I (normalized unitary 1-design).
    Imposed in Sec. IV.d, Eq. (39), for Theorem 1; the proof in Appendix H notes it is stronger than necessary.
  • domain assumption For the qubit-efficient variant, the native one-ancilla channel is well approximated by the Gaussian KMS surrogate in the large-T, large-sigma regime.
    Used in Appendix E to extend the symmetry results to the qubit-efficient sampler; the approximation error is stated but not bounded in the main experiments.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Convergence monitoring of quantum Gibbs samplers." pith.science (2026). https://pith.science/paper/ASXFMO2O

@misc{pith2026260802038,
  author       = {Pith},
  title        = {Pith review of: Convergence monitoring of quantum Gibbs samplers},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ASXFMO2O}},
  note         = {Machine review of arXiv:2608.02038}
}
read the original abstract

Recent progress in fully quantum Markov chain Monte Carlo methods enables efficient Gibbs-state sampling on quantum computers [Chen et al., Nature 646, 561 (2025)]. Although rigorous worst-case bounds on mixing times remain largely inaccessible for classically intractable systems, experience from classical Monte Carlo suggests that convergence of relevant observables may nevertheless be rapid. This raises the practical question of how to diagnose convergence efficiently, i.e., with at most polynomial overhead. We propose a low-cost criterion for convergence monitoring that exploits the weak measurements inherent in quantum Gibbs samplers and their qubit-efficient variants [Ding et al., arXiv:2508.05703 (2025)]. Our approach is based on the observation that, at thermal equilibrium, the net energy flow between system and environment vanishes and energy-exchange statistics satisfy a balance condition. This condition appears in the distribution of (quasi-)frequencies extracted from the weak-measurement record and we use it to construct a Hamiltonian-agnostic stopping criterion based solely on data already generated by the sampler. We provide a statistical analysis, along with numerical and analytical studies to understand its performance, assumptions, and limitations.

Figures

Figures reproduced from arXiv: 2608.02038 by the authors.

Figure 1
Figure 1. Davies limit intuition for the convergence monitor [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Implementation of Lindbladian simulation via weak [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Relation between the deterministic target distance [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Inverse-temperature dependence of the deterministic energy-relaxation time [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Naive convergence monitoring by repeated state preparation and destructive measurement. The black curve represents [PITH_FULL_IMAGE:figures/full_fig_p015_5.png]
Figure 6
Figure 6. Figure 6: Ideal Bohr-frequency resolution via phase estimation. Starting from a block encoding [PITH_FULL_IMAGE:figures/full_fig_p018_6.png]
Figure 7
Figure 7. Figure 7: Construction of the block-encoding unitary [PITH_FULL_IMAGE:figures/full_fig_p020_7.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

67 extracted references · 1 canonical work pages

  1. [1]

    A learning algorithm for Boltzmann machines

    D. H. Ackley, G. E. Hinton, and T. J. Sejnowski.“A learning algorithm for Boltzmann machines”. Cognitive science9(1):147–169 (1985)

    1985

  2. [2]

    Quantum boltzmann machine

    M. H. Amin, E. Andriyash, J. Rolfe, B. Kulchytskyy, and R. Melko.“Quantum boltzmann machine”. Physical Review X8(2):021050 (2018)

    2018

  3. [3]

    A tutorial on adaptive MCMC

    C. Andrieu and J. Thoms.“A tutorial on adaptive MCMC”. Statistics and Computing18:343–373 (2008)

    2008

  4. [4]

    The multiplicative weights update method: a meta-algorithm and applica- tions

    S. Arora, E. Hazan, and S. Kale.“The multiplicative weights update method: a meta-algorithm and applica- tions”. Theory of computing8(1):121–164 (2012)

    2012

  5. [5]

    Cen- tral limit theorems for open quantum random walks and quantum measurement records

    S. Attal, N. Guillotin-Plantard, and C. Sabot.“Cen- tral limit theorems for open quantum random walks and quantum measurement records”. Annales Henri Poincaré 16(1):15–43 (2015)

    2015

  6. [6]

    Rapid thermalization of spin chain commuting Hamiltonians

    I. Bardet, Á. Capel, L. Gao, A. Lucia, D. Pérez- García, and C. Rouzé.“Rapid thermalization of spin chain commuting Hamiltonians”. Physical Review Let- ters130(6):060401 (2023)

    2023

  7. [7]

    Quantum Gibbs Sampling in Infinite Dimensions: Generation, Mixing Times and Circuit Implementation

    S. Becker, C. Rouzé, and R. Salzmann.“Quantum Gibbs Sampling in Infinite Dimensions: Generation, Mixing Times and Circuit Implementation”, (2026). Available online:https://arxiv.org/abs/2604.01192

    arXiv 2026

  8. [8]

    Simulating Ther- mal Properties of Bose-Hubbard Models on a Quantum Computer

    S. Becker, C. Rouzé, and R. Salzmann.“Simulating Ther- mal Properties of Bose-Hubbard Models on a Quantum Computer”, (2026). Available online:https://arxiv. org/abs/2604.06077

    arXiv 2026

Show all 67 references
  1. [9]

    Fast Mixing of Quan- tum Spin Chains at All Temperatures

    T. Bergamaschi and C.-F. Chen.“Fast Mixing of Quan- tum Spin Chains at All Temperatures”, (2025). Available online:https://arxiv.org/abs/2510.08533

    2025

  2. [10]

    Binder, editor.Applications of the Monte Carlo Method in Statistical Physics

    K. Binder, editor.Applications of the Monte Carlo Method in Statistical Physics. volume 36 ofTopics in Current Physics, Springer Berlin, Heidelberg (1987)

    1987

  3. [11]

    Quantum speed-ups for solving semidefinite programs

    F. G. Brandao and K. M. Svore.“Quantum speed-ups for solving semidefinite programs”. In2017 IEEE 58th Annual Symposium on Foundations of Computer Science (FOCS), pages 415–426, (2017)

    2017

  4. [12]

    Breuer and F

    H.-P. Breuer and F. Petruccione.The theory of open quantum systems. OUP Oxford (2002)

    2002

  5. [13]

    Central limit theorem and large deviation principle for continuous time open quantum walks

    H. Bringuier.“Central limit theorem and large deviation principle for continuous time open quantum walks”. In Annales Henri Poincaré, volume 18, pages 3167–3192, (2017)

    2017

  6. [14]

    Brooks, A

    S. Brooks, A. Gelman, G. Jones, and X.-L. Meng.Hand- book of Markov Chain Monte Carlo. CRC press (2011)

    2011

  7. [15]

    Efficient Shadow Tomography of Thermal States

    C.-F.ChenandA.Gilyén.“Efficient Shadow Tomography of Thermal States”, (2026). Available online:https:// arxiv.org/abs/2603.16845

    2026

  8. [16]

    Efficient quantum thermal simulation

    C.-F. Chen, M. Kastoryano, F. G. Brandão, and A. Gi- lyén.“Efficient quantum thermal simulation”. Nature 646(8085):561–566 (2025)

    2025

  9. [17]

    Quantum Thermal State Preparation

    C.-F. Chen, M. J. Kastoryano, F. G. S. L. Brandão, and A. Gilyén.“Quantum Thermal State Preparation”, (2023). Available online:https://arxiv.org/abs/2303. 18224

    2023

  10. [18]

    An ef- ficient and exact noncommutative quantum Gibbs sam- pler

    C.-F. Chen, M. J. Kastoryano, and A. Gilyén.“An ef- ficient and exact noncommutative quantum Gibbs sam- pler”, (2023). Availableonline:https://arxiv.org/abs/ 2311.09207

    2023

  11. [19]

    Thermal expectation estimation via single-trajectory Gibbs sam- pling with non-destructive measurements

    H. Chen, J. Jiang, B. Li, and L. Ying.“Thermal expectation estimation via single-trajectory Gibbs sam- pling with non-destructive measurements”. arXivpreprint arXiv:2603.21595 , (2026)

    2026 arXiv

  12. [20]

    Quantum Replica Exchange

    Z. Chen, J. Basso, Z. Ding, and L. Lin.“Quantum Replica Exchange”, (2025). Available online:https://arxiv. org/abs/2510.07291

    2025

  13. [21]

    Markovian master equations

    E. B. Davies.“Markovian master equations”. Communi- cations in mathematical Physics39(2):91–110 (1974)

    1974

  14. [22]

    Markovian master equations. II

    E. B. Davies.“Markovian master equations. II”. Mathe- matische Annalen219(2):147–158 (1976)

    1976

  15. [23]

    Polynomial-time preparation of low-temperature Gibbs states for 2d toric code

    Z. Ding, Z. Landau, B. Li, L. Lin, and R. Zhang. “Polynomial-time preparation of low-temperature Gibbs states for 2d toric code”. arXiv preprint arXiv:2410.01206 , (2024)

    2024 arXiv

  16. [24]

    Efficient Quantum Gibbs Samplers with Kubo–Martin–Schwinger Detailed Balance Condition

    Z. Ding, B. Li, and L. Lin.“Efficient Quantum Gibbs Samplers with Kubo–Martin–Schwinger Detailed Balance Condition”. Communications in Mathematical Physics 406:67 (2025)

    2025

  17. [25]

    End-to-End Efficient Quantum Thermal and Ground State Prepara- tion Made Simple

    Z. Ding, Y. Zhan, J. Preskill, and L. Lin.“End-to-End Efficient Quantum Thermal and Ground State Prepara- tion Made Simple”, (2025). Available online:https: //arxiv.org/abs/2508.05703

    2025

  18. [26]

    Simulating physics with computers

    R. P. Feynman.“Simulating physics with computers”. International Journal of Theoretical Physics21(6):467– 488 (1982)

    1982

  19. [27]

    Markov chain Monte Carlo: Can we trust the third significant figure?

    J. M. Flegal, M. Haran, and G. L. Jones.“Markov chain Monte Carlo: Can we trust the third significant figure?”. Statistical Science pages 250–260 (2008)

    2008

  20. [28]

    Batch means and spectral variance estimators in Markov chain Monte Carlo

    J. M. Flegal and G. L. Jones.“Batch means and spectral variance estimators in Markov chain Monte Carlo”. The Annals of Statistics38(2):1034–1070 (2010)

    2010

  21. [29]

    Quantum Monte Carlo simulations of solids

    W. M. Foulkes, L. Mitas, R. Needs, and G. Rajagopal. “Quantum Monte Carlo simulations of solids”. Reviews of Modern Physics73(1):33 (2001)

    2001

  22. [30]

    Perfect Sampling for Quantum Gibbs States

    D. S. França.“Perfect Sampling for Quantum Gibbs States”. Quantum Information & Computation 18(5&6):361–388 (2018)

    2018

  23. [31]

    Gibbs sampling

    A.E.Gelfand.“Gibbs sampling”. JournaloftheAmerican statistical Association95(452):1300–1304 (2000)

    2000

  24. [32]

    Inference from Iterative Simulation Using Multiple Sequences

    A. Gelman and D. B. Rubin.“Inference from Iterative Simulation Using Multiple Sequences”. Statistical Science 7(4):457 – 472 (1992)

    1992

  25. [33]

    Practical markov chain monte carlo

    C. J. Geyer.“Practical markov chain monte carlo”. Sta- tistical science pages 473–483 (1992). 12

    1992

  26. [34]

    An adaptive Metropolis algorithm

    H. Haario, E. Saksman, and J. Tamminen.“An adaptive Metropolis algorithm”. Bernoulli7(2):223–242 (2001)

    2001

  27. [35]

    Sign problem in Monte Carlo simulations of frustrated quantum spin systems

    P. Henelius and A. W. Sandvik.“Sign problem in Monte Carlo simulations of frustrated quantum spin systems”. Physical Review B62(2):1102–1113 (2000)

    2000

  28. [36]

    Exchange Monte Carlo Method and Application to Spin Glass Simulations

    K. Hukushima and K. Nemoto.“Exchange Monte Carlo Method and Application to Spin Glass Simulations”. Journal of the Physical Society of Japan65:1604–1608 (1996)

    1996

  29. [37]

    Predicting properties of quantum thermal states from a single trajectory

    J. Jiang, J. Leng, and L. Lin.“Predicting properties of quantum thermal states from a single trajectory”, (2026). Available online:https://arxiv.org/abs/2602.12539

    2026

  30. [38]

    Fixed-width output analysis for Markov chain Monte Carlo

    G. L. Jones, M. Haran, B. S. Caffo, and R. Neath. “Fixed-width output analysis for Markov chain Monte Carlo”. Journal of the American Statistical Association 101(476):1537–1547 (2006)

    2006

  31. [39]

    Quantum Gibbs Samplers: The Commuting Case

    M. J. Kastoryano and F. G. S. L. Brandão.“Quantum Gibbs Samplers: The Commuting Case”. Communica- tions in Mathematical Physics344(3):915–957 (2016)

    2016

  32. [40]

    Quantum logarith- mic Sobolev inequalities and rapid mixing

    M. J. Kastoryano and K. Temme.“Quantum logarith- mic Sobolev inequalities and rapid mixing”. Journal of Mathematical Physics54(5) (2013)

    2013

  33. [41]

    Rapid thermalization of dissipative many-body dynamics of commuting Hamiltonians

    J. Kochanowski, A. M. Alhambra, A. Capel, and C. Rouzé.“Rapid thermalization of dissipative many-body dynamics of commuting Hamiltonians”. Communications in Mathematical Physics406(8):176 (2025)

    2025

  34. [42]

    Landau and K

    D. Landau and K. Binder.A guide to Monte Carlo simu- lations in statistical physics. Cambridge university press (2021)

    2021

  35. [43]

    D. A. Levin and Y. Peres.Markov chains and mixing times. volume 107, American Mathematical Soc. (2017)

    2017

  36. [44]

    Lecture notes on the theory of open quantum systems

    D.A.Lidar.“Lecture notes on the theory of open quantum systems”. arXiv preprint arXiv:1902.00967 , (2019)

    1902 arXiv

  37. [45]

    Eternal black holes in anti-de Sitter

    J. Maldacena.“Eternal black holes in anti-de Sitter”. Journal of High Energy Physics2003(04):021 (2003)

    2003

  38. [46]

    Simulated Tempering: A New Monte Carlo Scheme

    E. Marinari and G. Parisi.“Simulated Tempering: A New Monte Carlo Scheme”. Europhysics Letters (1992)

    1992

  39. [47]

    Equation of state calculations by fast computing machines

    N. Metropolis, A. W. Rosenbluth, M. N. Rosenbluth, A. H. Teller, and E. Teller.“Equation of state calculations by fast computing machines”. The journal of chemical physics21(6):1087–1092 (1953)

    1953

  40. [48]

    Sign problem in quantum Monte Carlo simulation

    G. Pan and Z. Y. Meng.“Sign problem in quantum Monte Carlo simulation”. arXiv preprint arXiv:2204.08777 , (2022)

    2022 arXiv

  41. [49]

    C. P. Robert and G. Casella.Monte Carlo Statistical Methods. Springer-Verlag New York (1999)

    1999

  42. [50]

    Effi- cient thermalization and universal quantum computing with quantum Gibbs samplers

    C. Rouzé, D. S. França, and Á. M. Alhambra.“Effi- cient thermalization and universal quantum computing with quantum Gibbs samplers”. InProceedings of the 57th Annual ACM Symposium on Theory of Computing, pages 1488–1495, (2025)

    2025

  43. [51]

    Op- timal Quantum Algorithm for Gibbs State Preparation

    C. Rouzé, D. Stilck França, and Á. M. Alhambra.“Op- timal Quantum Algorithm for Gibbs State Preparation”. Physical Review Letters136(6):060601 (2026)

    2026

  44. [52]

    Convergence diagnostics for Markov chain Monte Carlo

    V. Roy.“Convergence diagnostics for Markov chain Monte Carlo”, (2019). DOI: 10.48550/arXiv.1909.11827

    2019 doi arXiv

  45. [53]

    Polynomial-time quantum Gibbs sampling for the weak and strong coupling regime of the Fermi-Hubbard model at any temperature

    Š. Šmíd, R. Meister, M. Berta, and R. Bondesan. “Polynomial-time quantum Gibbs sampling for the weak and strong coupling regime of the Fermi-Hubbard model at any temperature”. Nature Communications16(1):10736 (2025)

    2025

  46. [54]

    Rapid Mixing of Quantum Gibbs Samplers for Weakly-Interacting Quantum Systems

    Š. Šmíd, R. Meister, M. Berta, and R. Bonde- san.“Rapid Mixing of Quantum Gibbs Samplers for Weakly-Interacting Quantum Systems”. arXiv preprint arXiv:2510.04954 , (2025)

    2025 arXiv

  47. [55]

    How to beat critical slowing-down: 1990 update

    A. D. Sokal.“How to beat critical slowing-down: 1990 update”. Nuclear Physics B-Proceedings Supplements 20:55–67 (1991)

    1990

  48. [56]

    Replica Monte Carlo simulation of spin glasses

    R. H. Swendsen and J.-S. Wang.“Replica Monte Carlo simulation of spin glasses”. Physical review letters 57(21):2607–2609 (1986)

    1986

  49. [57]

    Nonuniversal critical dynamics in Monte Carlo simulations

    R. H. Swendsen and J.-S. Wang.“Nonuniversal critical dynamics in Monte Carlo simulations”. Physical Review Letters58:86–88 (1987)

    1987

  50. [58]

    Thermo field dy- namics

    Y. Takahashi and H. Umezawa.“Thermo field dy- namics”. International journal of modern Physics B 10(13n14):1755–1805 (1996)

    1996

  51. [59]

    Lower bounds to the spectral gap of Davies generators

    K. Temme.“Lower bounds to the spectral gap of Davies generators”. Journal of Mathematical Physics54(12) (2013)

    2013

  52. [60]

    Quantum metropolis sampling

    K. Temme, T. J. Osborne, K. G. Vollbrecht, D. Poulin, and F. Verstraete.“Quantum metropolis sampling”. Na- ture471(7336):87–90 (2011)

    2011

  53. [61]

    Fast mixing of weakly interacting fermionic systems at any temperature

    Y. Tong and Y. Zhan.“Fast mixing of weakly interacting fermionic systems at any temperature”. PRX Quantum 6(3):030301 (2025)

    2025

  54. [62]

    Quantum SDP-solvers: Better upper and lower bounds

    J.VanApeldoorn, A.Gilyén, S.Gribling, andR.deWolf. “Quantum SDP-solvers: Better upper and lower bounds”. In2017 IEEE 58th Annual Symposium on Foundations of Computer Science (FOCS), pages 403–414, (2017)

    2017

  55. [63]

    Multivariate out- put analysis for Markov chain Monte Carlo

    D. Vats, J. M. Flegal, and G. L. Jones.“Multivariate out- put analysis for Markov chain Monte Carlo”. Biometrika 106(2):321–337 (2019)

    2019

  56. [64]

    Critical slowing down

    U. Wolff.“Critical slowing down”. Nucl. Phys. B-Proc. Sup17:93–102 (1989)

    1989

  57. [65]

    Collective Monte Carlo Updating for Spin Sys- tems

    U. Wolff.“Collective Monte Carlo Updating for Spin Sys- tems”. Physical Review Letters62:361–364 (1990)

    1990

  58. [66]

    Generation of thermofield double states and critical ground states with a quantum com- puter

    D. Zhu, S. Johri, N. M. Linke, K. Landsman, C. Huerta Alderete, N. H. N. H, A. Matsuura, T. Hsieh, and C. Monroe.“Generation of thermofield double states and critical ground states with a quantum com- puter”. Proceedings of the National Academy of Sciences 117(41):25402–25406 (2020)

    2020

  59. [67]

    The modified logarithmic Sobolev inequality for quantum spin systems: classical and commuting nearest neighbour interactions

    Ángela Capel, C. Rouzé, and D. S. França.“The modified logarithmic Sobolev inequality for quantum spin systems: classical and commuting nearest neighbour interactions”, (2021). Available online:https://arxiv.org/abs/2009. 11817. 13 CONTENTS I. Introduction 1 II. Quantum Gibbs ...

    2021

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.