Pith. sign in

REVIEW 3 major objections 3 minor 1 cited by

This paper establishes that an arbitrary sparse Lindbladian—every Hamiltonian coefficient and every Kossakowski entry—can be reconstructed to precision ε using only product Pauli state preparations, single uninterrupted forward evolutions,

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

2026-08-01 03:47 UTC pith:ZR6D7QUD

load-bearing objection The no-SPAM core is real and well-supported, but the advertised SPAM robustness is not: Section V's assumptions are postulates, and the corrected full-weight samples carry exponential-in-n overhead that the abstract does not disclose. the 3 major comments →

arxiv 2607.23044 v1 pith:ZR6D7QUD submitted 2026-07-25 quant-ph cs.ITcs.LGmath.IT

Characterizing Arbitrary Lindbladian Dynamics with a Few Pauli Measurements

Taiqi Zhou , Weiyuan Gong This is my paper
classification quant-ph cs.ITcs.LGmath.IT MSC 81P4581S22 PACS 03.67.-a
keywords Lindbladian learningopen quantum systemsPauli measurementsstructure learningKossakowski matrixderivative estimationquantum error mitigationsparse recovery
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.

Quantum devices are open systems, and calibration, error mitigation, and error correction all need a model of both the coherent Hamiltonian and the dissipative noise. This paper claims that the full generator of an arbitrary sparse Markovian open system—every Hamiltonian term and every entry of the Kossakowski matrix—can be learned from experiments a near-term device already supports: prepare a product state, evolve once forward, measure qubit by qubit. No ancillas, entangled probes, inverse evolutions, or mid-circuit control are needed, and neither the set of present couplings nor any locality structure is known in advance. Given a sparsity budget and a strength bound, every coefficient is recovered to precision ε from Õ(Γ²M₀²/ε⁴) experiments and Õ(ΓM₀²/ε²) total evolution time, a cost that saturates the control-free standard quantum limit. If true, this turns sparse Lindbladian noise models into directly accessible objects for near-term devices, extending error mitigation beyond Pauli-twirled channels.

Core claim

The central discovery is that the coefficient-estimation problem for a Lindbladian decouples cleanly under product-Pauli measurements, even though Hamiltonian and dissipative terms collide on shared Pauli labels. The protocol first finds the heavy diagonal noise rows by recording sign-flip patterns in the three global bases; the collision set of their labels is where the two contributions may overlap. Outside that set, a no-jump domination inequality shows that decoherence cannot hide a large Hamiltonian coefficient, so same-Pauli displacement statistics reveal its support. Coefficient values are then extracted from one-sided derivatives of trace responses at strictly positive times using a

What carries the argument

The argument is carried by three linked objects. First, the Pauli-Kossakowski form of the generator, where the Kossakowski matrix A ⪰ 0 ties every off-diagonal entry to a diagonal rate via |A_uv|² ≤ A_uu A_vv, which supplies the sparse-structure control. Second, the one-sided Legendre-kernel derivative estimator: since every response has all derivatives bounded by Γ, a single shot at a random positive time τU, reweighted by a signed kernel, estimates f′(0) with O(Γ²R⁶/α²) samples and O(Γ/α²) time, using only O(log(Γ/α)) distinct times. Third, the no-jump domination e^{tL}(ρ) ⪰ e^{tK₀}ρe^{tK₀†} with K₀ = −iH − J/2, whose imaginary/real orthogonality makes any |h_s| > ε create a detectable dis

Load-bearing premise

The evolution must be exactly the semigroup e^{tL} of a time-independent Markovian Lindbladian, and the learner must know a valid sparsity budget M₀ and strength bound Γ; if the noise is non-Markovian, time-dependent, or denser than the budget, the short-time derivative no longer encodes the generator and the guarantee fails.

What would settle it

On a system with suspected memory, run the protocol with the nominal probe window τ = 1/(2Γ) and again with τ/2, using the same total time budget; a Markovian generator must yield the same coefficient estimates within ε, so any systematic shift larger than ε between the two runs falsifies the time-homogeneous premise.

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

If this is right

  • Sparse Pauli-Lindblad noise models—including off-diagonal correlations between jump operators—can be learned in situ, so probabilistic error cancellation no longer needs twirling to a diagonal channel.
  • The sample and time scaling is optimal for control-free experiments: the ε⁻² total-evolution-time dependence matches the standard quantum limit and cannot be improved even with ancillas.
  • Support recovery from data with no locality or ansatz assumption means long-range, nonlocal, or otherwise unexpected generators are handled automatically.
  • Only a logarithmic number of strictly positive evolution times is needed, and the times can be snapped to a hardware clock lattice, so the protocol fits fixed-depth near-term experiments.
  • Calibrated state-preparation and measurement errors multiply responses by known factors and are divided out with constant overhead, preserving all scaling.

Where Pith is reading between the lines

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

  • The same one-sided differentiation machinery may carry over to weakly time-dependent generators if the probe window is short compared with the drift rate; the paper explicitly leaves time-dependence open.
  • The M₀² cost comes from estimating each fixed-sum block independently; a shared random-q estimator or compressed-sensing approach could plausibly remove that overhead, approaching the closed-system cost.
  • The phase-orthogonality identity is a Walsh transform on the symplectic space, so the technique may generalize to learning other tensor-product-basis linear maps beyond Lindbladians.
  • A practical device check: run the protocol at two probe windows τ and τ/2; if the estimated coefficients drift by more than ε, the time-homogeneous Markovian model is inadequate for that noise.

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

3 major / 3 minor

Summary. The paper proposes a control-free protocol for learning an unknown sparse Lindbladian in Pauli-Kossakowski form, using only product Pauli preparations, one forward evolution per shot, and product Pauli measurements. Given a sparsity budget M0 and a strength bound Γ, the protocol is claimed to learn every Hamiltonian coefficient h_s and every Kossakowski entry A_uv to precision ε, with both supports identified from data, using eO(Γ²M0²/ε⁴) shots and eO(ΓM0²/ε²) total evolution time. The formal no-SPAM guarantee is Theorem 4 in the Supplementary Materials, supported by a sequence of lemmas; the main text also advertises robustness to calibrated state-preparation-and-measurement errors. The numerical section reports simulations on small k-local and sparse random Lindbladians.

Significance. If the no-SPAM Theorem 4 is correct, this is a substantial contribution: it is the first ansatz-free, control-free Lindbladian learning protocol with polynomial sample and time complexity that recovers both coherent and dissipative terms without locality assumptions, ancillas, inverse evolutions, or entangled measurements. The supplement contains a complete lemma chain and explicit resource accounting, with no fit-to-data parameters; all thresholds are analytic functions of the inputs. The claimed complexity saturates the known control-free Heisenberg/SQL-type limits for the special cases discussed. However, the advertised SPAM robustness is not established: Section V relies on assumptions that assert the needed guarantees, and under the depolarizing model the correction factors can be exponentially small in n, contradicting the 'constant overhead' claim. The no-SPAM core is still valuable, but the paper's full advertised package is not currently supported.

major comments (3)
  1. [Section V, Assumptions 1–2] The SPAM robustness proof is conditional on two assumptions that are exactly the required guarantees. Assumption 1 asserts that calibrated SPAM preserves the heavy-row candidate guarantee with constant overhead; Assumption 2 asserts that each target displacement is preserved with probability ζ₁. No derivation is given showing that the depolarizing model (S193)–(S194) satisfies either assumption. This is not a matter of tuning constants: the proof of Lemma 9 and Theorem 2 is replaced by a postulate. Consequently the claim in the abstract and in the third remark after Theorem 1 that the protocol is 'provably robust to calibrated state-preparation and measurement errors' is not supported by the manuscript as written.
  2. [Section V, Lemma 20 and Corollary 1, Eq. (S204)] Even accepting the trace-response correction in Lemma 20, the stated overhead is not constant under the depolarizing model. The fixed-sum estimator of Stage 3 draws q uniformly from V, so with high probability wt(q), wt(q+s) = Θ(n). The correction factor α(Q)β(P) = r_P^{wt(q)} r_M^{wt(q+s)} is then r^{Θ(n)} for fixed depolarizing rates r_P,r_M<1. Hence ζ₂ defined in Eq. (S199) is exponentially small in n, and the ζ₂^{-2} terms in Corollary 1's complexity (S204) are r^{-Θ(n)}. This contradicts the third remark after Theorem 1, which promises dividing SPAM factors out with 'constant overhead,' and the abstract's advertised eO(Γ²M0²/ε⁴) scaling. The issue is structural: the Walsh-averaging inversion in Propositions 2–3 requires a uniform average over all Pauli labels, so a simple restriction to low-weight q does not trivially fix it. The SPAM-robustness claim must either be reproved with a
  3. [Abstract / Theorem 1 informal statement] The abstract and the informal Theorem 1 present SPAM robustness as part of the main result, but the formal theorem (Theorem 4 in the Supplement) is stated for SPAM-free evolution. The SPAM material in Section V is explicitly conditional on Assumptions 1–2. The difference between what is advertised and what is proved is large enough to affect the paper's central narrative. The authors should state in the main text that the unconditional guarantee is for SPAM-free access, and that the SPAM-robustness result is either a separate theorem under explicit assumptions or a claim to be proved from the depolarizing model.
minor comments (3)
  1. [Theorem 3, Eqs. (S132)–(S133)] The displayed bias bound appears to drop the factor 1/q(s) before the final inequality. With q(s)≥1/3, the bound becomes ≈0.09ε rather than ε/24. This still fits inside the ε/4 total budget if the constants are loosened, so the proof is repairable, but the displayed argument should be corrected.
  2. [Section V, Eq. (S198)] The text says the depolarizing special case gives 'worst-case variance overhead at most r^{-2k}' when preparation and measurement strings have weight at most k. But in the actual algorithm q is uniform, so k is not a bounded input parameter. This sentence should be reconciled with the uniform-q sampling used in Stage 3.
  3. [Numerical simulations] The sentence 'the error dependence on M is below the predicted M^{-3}' is unclear: if the mean-averaged error is expected to increase with M, 'below' should refer to the slope or the magnitude. Please rephrase to avoid confusion.

Circularity Check

1 steps flagged

SPAM-robustness claim reduces to unproved Assumptions 1–2; the no-SPAM learning result itself is independent.

specific steps
  1. self definitional [Section V (SPAM robustness), Assumptions 1–2 and Corollary 1; cf. abstract claim of 'provably robust to calibrated state-preparation and measurement errors']
    "Rather than deriving a full detector-specific SPAM model, we state the exact calibrated guarantee needed by the proof. Assumption 1 (Calibrated SPAM preserves the heavy-row candidate guarantee). At the threshold ξ and time τD, a calibrated implementation of the Lemma 9 outputs C satisfying SD,ξ ⊆ C, |C| ≤ C0M0 ... Assumption 2 ... There is a calibrated number ζ1 > 0 such that ... P[D1 = d | D0 = d] ≥ ζ1 ... Corollary 1 ... Assume ... calibrated SPAM satisfying Assumption 1 and Assumption 2 ... Then the coefficient guarantee of Theorem 4 holds ..."

    The advertised SPAM robustness is not derived from the depolarizing/preparation-measurement model (S193)-(S194). Assumption 1 asserts exactly the heavy-row support guarantee (SD,ξ ⊆ C, |C| ≤ C0M0) that Lemma 9 would need to provide under SPAM, and Assumption 2 asserts exactly the displacement-preservation probability needed by the projection stage. Corollary 1 then multiplies repetition counts by ζ1^{-1}, ζ2^{-2} and concludes the Theorem 4 guarantee. The structure-learning content of the robustness theorem is therefore the assumption restated, not a derived consequence: if calibrated SPAM does not satisfy Assumptions 1–2, no proof is given that the advertised guarantee holds. This makes the abstract's 'provably robust to calibrated SPAM errors' claim equivalent, by construction, to the un

full rationale

The core no-SPAM derivation (Lemmas 1–4, 9–19, Propositions 1–3, Theorem 4) is essentially self-contained: thresholds and constants are analytic functions of the inputs, the Legendre/Chebyshev derivative estimators are proved, the no-jump domination is proved, and the Pauli-Liouville inversion and Walsh-averaging steps are proved in the supplement. Citations to the authors' prior works [63,85,86] appear for elementary identities and derivative bounds (e.g., χ'_u,u(0)=A_uu), but these are direct mathematical facts whose stated assumptions do not include the target learnability claim; they are not uniqueness theorems or fitted values, so they do not constitute circularity. The one genuine circular reduction is in the advertised SPAM robustness: rather than proving that calibrated depolarizing SPAM preserves the heavy-row and displacement guarantees, the paper postulates those guarantees as Assumptions 1–2 and then derives the robustness corollary from them. Thus the partial-circularity score is 6, not higher, because the main no-SPAM learning theorem stands independently of the SPAM assumptions.

Axiom & Free-Parameter Ledger

0 free parameters · 5 axioms · 0 invented entities

The protocol introduces no new physical entities; the no-jump operator K0 and the Legendre kernel are mathematical tools. The free parameters of the problem are the sparsity budget M0 and strength bound Γ, which are inputs, not fitted values. The threshold constants are analytic universal constants fixed by the proof, not data-fitted. The main external assumptions are the Markovian Lindblad model and the no-control access model; the SPAM robustness claim additionally requires two assumptions stated in Section V.

axioms (5)
  • domain assumption The unknown generator is exactly a time-independent Lindbladian with Pauli-Kossakowski form (Eq. S7) and A = A† ⪰ 0.
    Introduction and Eq. (1)/(S7). The semigroup property e^{tL} is required for the short-time expansion and for the no-jump domination argument.
  • domain assumption The learner is given a sparsity budget M0 ≥ |SH| + |SD|² and a strength bound Γ ≥ Γ*.
    Problem 1 and Eq. (S13). If M0 underestimates the true sparsity, the heavy-row threshold ξ is too large and the ε-accuracy guarantee fails.
  • domain assumption Access model is restricted to product Pauli preparations, single forward evolutions e^{tL}, and product Pauli measurements, with no ancillas or mid-circuit control.
    Problem 1, items (1)-(4) and the main text near Eq. (3). This defines the in-situ, control-free setting the protocol targets.
  • standard math The no-jump domination inequality e^{tL}(ρ) ⪰ e^{tK0}ρ e^{tK0†} (Lemma 11) holds via Dyson expansion.
    Proved in Supplementary Section II.E using complete positivity and the Dyson series; standard in open quantum systems; not an empirical assumption.
  • ad hoc to paper Calibrated SPAM satisfies Assumptions 1 and 2 (heavy-row candidate guarantee preserved, target displacements preserved with probability ζ1).
    Supplementary Section V.A. These are stated, not derived, and are needed for the advertised 'provably robust to calibrated SPAM errors' claim in the abstract.

pith-pipeline@v1.3.0-alltime-deepseek · 34260 in / 25917 out tokens · 239454 ms · 2026-08-01T03:47:38.260767+00:00 · methodology

0 comments
read the original abstract

Quantum devices are open systems whose dynamics interleave coherent evolution with dissipation, and benchmarking, error mitigation, and error correction all rest on a faithful model of both. Existing characterization protocols either assume prior knowledge of the interaction and noise structure, or demand ancillas, entangled probes, or mid-circuit control, or capture only the Pauli-diagonal part of the noise. Here, we present a protocol that reconstructs an arbitrary sparse Markovian generator, including every Hamiltonian together with the jump operator coefficients, using only product Pauli state preparation, single uninterrupted forward evolutions, and product Pauli measurements. Given a sparsity budget $M_0$ and a strength bound $\Gamma$ of the Lindbladian, every coefficient is learned to precision $\epsilon$ from $\widetilde{O}(\Gamma^2M_0^2/\epsilon^4)$ experiments and $\widetilde{O}(\Gamma M_0^2/\epsilon^2)$ total evolution time, with both supports identified from data without locality assumptions. The protocol runs at a logarithmic number of positive evolution times on a hardware clock lattice and is provably robust to calibrated state-preparation and measurement errors.

Figures

Figures reproduced from arXiv: 2607.23044 by Taiqi Zhou, Weiyuan Gong.

Figure 1
Figure 1. Figure 1: (Left) Learning k-local Lindbladians. (Right) Learning random sparse Lindbladians. out of every response preserve the scalings of the number of shots and the total evolution time. Numerical simulations.—We validate the complete Lind￾bladian reconstruction protocol by numerical simulations in [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗

discussion (0)

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

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Learning Arbitrary Lindbladians from Time Evolution

    quant-ph 2026-07 accept novelty 7.0

    Arbitrary Lindbladians of strength ≤Λ are learned entrywise to error ε with Õ(Λ²/ε²) ancilla-free, control-free experiments and Õ(Λ/ε²) total evolution time.

Reference graph

Works this paper leans on

92 extracted references · 14 linked inside Pith · cited by 1 Pith paper

  1. [1]

    Lindblad, On the generators of quantum dynamical semi- groups, Communications in Mathematical Physics48, 119 (1976)

    G. Lindblad, On the generators of quantum dynamical semi- groups, Communications in Mathematical Physics48, 119 (1976)

    1976

  2. [2]

    Gorini, A

    V . Gorini, A. Kossakowski, and E. C. G. Sudarshan, Completely positive dynamical semigroups of n-level systems, Journal of Mathematical Physics17, 821 (1976)

    1976

  3. [3]

    Breuer and F

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

    2002

  4. [4]

    M. M. Wolf, J. Eisert, T. S. Cubitt, and J. I. Cirac, Assessing non-Markovian quantum dynamics, Physical Review Letters 101, 150402 (2008)

    2008

  5. [5]

    T. S. Cubitt, J. Eisert, and M. M. Wolf, The complexity of relating quantum channels to master equations, Communications in Mathematical Physics310, 383 (2012)

    2012

  6. [6]

    Rivas, S

    Á. Rivas, S. F. Huelga, and M. B. Plenio, Quantum non- Markovianity: characterization, quantification and detection, Reports on Progress in Physics77, 094001 (2014)

    2014

  7. [7]

    Preskill, Quantum computing in the NISQ era and beyond, Quantum2, 79 (2018)

    J. Preskill, Quantum computing in the NISQ era and beyond, Quantum2, 79 (2018)

    2018

  8. [8]

    Lloyd, Universal quantum simulators, Science273, 1073 (1996)

    S. Lloyd, Universal quantum simulators, Science273, 1073 (1996)

    1996

  9. [9]

    I. M. Georgescu, S. Ashhab, and F. Nori, Quantum simulation, Reviews of Modern Physics86, 153 (2014)

    2014

  10. [10]

    Altman, K

    E. Altman, K. R. Brown, G. Carleo, L. D. Carr, E. Demler, C. Chin, B. DeMarco, S. E. Economou, M. A. Eriksson, K.-M. C. Fu,et al., Quantum simulators: Architectures and opportunities, PRX Quantum2, 017003 (2021)

    2021

  11. [11]

    Kraft, M

    T. Kraft, M. K. Joshi, W. Lam, T. Olsacher, F. Kranzl, J. Franke, L. K. Joshi, R. Blatt, A. Smerzi, D. S. França, B. Verm- ersch, B. Kraus, C. F. Roos, and P. Zoller, Bounded-error quantum simulation via Hamiltonian and Lindbladian learning, arXiv:2511.23392 (2025)

    arXiv 2025

  12. [12]

    C. L. Degen, F. Reinhard, and P. Cappellaro, Quantum sensing, Reviews of Modern Physics89, 035002 (2017)

    2017

  13. [13]

    B. M. Terhal, Quantum error correction for quantum memories, Reviews of Modern Physics87, 307 (2015)

    2015

  14. [14]

    G. Q. AI, Suppressing quantum errors by scaling a surface code logical qubit, Nature614, 676 (2023)

    2023

  15. [15]

    G. Q. AI, Quantum error correction below the surface code threshold, Nature638, 920 (2025)

    2025

  16. [16]

    Temme, S

    K. Temme, S. Bravyi, and J. M. Gambetta, Error mitigation for short-depth quantum circuits, Physical Review Letters119, 180509 (2017)

    2017

  17. [17]

    Li and S

    Y . Li and S. C. Benjamin, Efficient variational quantum simulator incorporating active error minimization, Physical Review X7, 021050 (2017)

    2017

  18. [18]

    Z. Cai, R. Babbush, S. C. Benjamin, S. Endo, W. J. Huggins, Y . Li, J. R. McClean, and T. E. O’Brien, Quantum error mitiga- tion, Reviews of Modern Physics95, 045005 (2023)

    2023

  19. [19]

    Y . Kim, A. Eddins, S. Anand, K. X. Wei, E. Van Den Berg, S. Rosenblatt, H. Nayfeh, Y . Wu, M. Zaletel, K. Temme,et al., Evidence for the utility of quantum computing before fault toler- ance, Nature618, 500 (2023)

    2023

  20. [20]

    Van Den Berg, Z

    E. Van Den Berg, Z. K. Minev, A. Kandala, and K. Temme, Prob- abilistic error cancellation with sparse Pauli-Lindblad models on noisy quantum processors, Nature Physics19, 1116 (2023)

    2023

  21. [21]

    H.-Y . Hu, M. Ma, W. Gong, Q. Ye, Y . Tong, S. T. Flammia, and S. F. Yelin, Ansatz-free Hamiltonian learning with Heisenberg- limited scaling, PRX Quantum6, 040315 (2025)

    2025

  22. [22]

    I. L. Chuang and M. A. Nielsen, Prescription for experimental determination of the dynamics of a quantum black box, Journal of Modern Optics44, 2455 (1997)

    1997

  23. [23]

    Poyatos, J

    J. Poyatos, J. I. Cirac, and P. Zoller, Complete characterization of a quantum process: the two-bit quantum gate, Physical Review Letters78, 390 (1997)

    1997

  24. [24]

    S. T. Merkel, J. M. Gambetta, J. A. Smolin, S. Poletto, A. D. Córcoles, B. R. Johnson, C. A. Ryan, and M. Steffen, Self- consistent quantum process tomography, Physical Review A87, 062119 (2013)

    2013

  25. [25]

    Blume-Kohout, J

    R. Blume-Kohout, J. K. Gamble, E. Nielsen, K. Rudinger, J. Mizrahi, K. Fortier, and P. Maunz, Demonstration of qubit operations below a rigorous fault tolerance threshold with gate set tomography, Nature communications8, 14485 (2017)

    2017

  26. [26]

    Nielsen, J

    E. Nielsen, J. K. Gamble, K. Rudinger, T. Scholten, K. Young, and R. Blume-Kohout, Gate set tomography, Quantum5, 557 (2021)

    2021

  27. [27]

    O’Donnell and J

    R. O’Donnell and J. Wright, Efficient quantum tomography, inProceedings of the forty-eighth annual ACM symposium on Theory of Computing(2016) pp. 899–912

    2016

  28. [28]

    J. Haah, A. W. Harrow, Z. Ji, X. Wu, and N. Yu, Sample-optimal tomography of quantum states, inProceedings of the forty-eighth annual ACM symposium on Theory of Computing(2016) pp. 913–925

    2016

  29. [29]

    Eisert, D

    J. Eisert, D. Hangleiter, N. Walk, I. Roth, D. Markham, R. Parekh, U. Chabaud, and E. Kashefi, Quantum certification and benchmarking, Nature Reviews Physics2, 382 (2020)

    2020

  30. [30]

    Elben, S

    A. Elben, S. T. Flammia, H.-Y . Huang, R. Kueng, J. Preskill, B. Vermersch, and P. Zoller, The randomized measurement tool- box, Nature Reviews Physics5, 9 (2023)

    2023

  31. [31]

    M. P. da Silva, O. Landon-Cardinal, and D. Poulin, Practical characterization of quantum devices without tomography, Physi- cal Review Letters107, 210404 (2011)

    2011

  32. [32]

    C. E. Granade, C. Ferrie, N. Wiebe, and D. G. Cory, Robust online Hamiltonian learning, New Journal of Physics14, 103013 (2012)

    2012

  33. [33]

    Wiebe, C

    N. Wiebe, C. Granade, C. Ferrie, and D. G. Cory, Hamiltonian learning and certification using quantum resources, Physical Review Letters112, 190501 (2014)

    2014

  34. [34]

    Wang, D.-L

    S.-T. Wang, D.-L. Deng, and L.-M. Duan, Hamiltonian tomogra- phy for quantum many-body systems with arbitrary couplings, New Journal of Physics17, 093017 (2015)

    2015

  35. [35]

    J. Wang, S. Paesani, R. Santagati, S. Knauer, A. A. Gentile, N. Wiebe, M. Petruzzella, J. L. O’brien, J. G. Rarity, A. Laing, et al., Experimental quantum Hamiltonian learning, Nature Physics13, 551 (2017)

    2017

  36. [36]

    Bairey, I

    E. Bairey, I. Arad, and N. H. Lindner, Learning a local hamil- tonian from local measurements, Physical Review Letters122, 020504 (2019)

    2019

  37. [37]

    Qi and D

    X.-L. Qi and D. Ranard, Determining a local Hamiltonian from a single eigenstate, Quantum3, 159 (2019)

    2019

  38. [38]

    T. J. Evans, R. Harper, and S. T. Flammia, Scalable bayesian Hamiltonian learning, arXiv:1912.07636 (2019)

    Pith/arXiv arXiv 1912

  39. [39]

    Anshu, S

    A. Anshu, S. Arunachalam, T. Kuwahara, and M. Soleimani- far, Sample-efficient learning of interacting quantum systems, Nature Physics17, 931 (2021)

    2021

  40. [40]

    Zubida, E

    A. Zubida, E. Yitzhaki, N. H. Lindner, and E. Bairey, 6 Optimal short-time measurements for hamiltonian learning, arXiv:2108.08824 (2021)

    Pith/arXiv arXiv 2021

  41. [41]

    Bakshi, A

    A. Bakshi, A. Liu, A. Moitra, and E. Tang, Learning quantum Hamiltonians at any temperature in polynomial time, inPro- ceedings of the 56th Annual ACM Symposium on Theory of Computing(2024) pp. 1470–1477

    2024

  42. [42]

    J. Haah, R. Kothari, and E. Tang, Optimal learning of quantum Hamiltonians from high-temperature Gibbs states, in2022 IEEE 63rd Annual Symposium on Foundations of Computer Science (FOCS)(IEEE, 2022) pp. 135–146

    2022

  43. [43]

    A. Gu, L. Cincio, and P. J. Coles, Practical Hamiltonian learning with unitary dynamics and Gibbs states, Nature Communications 15, 312 (2024)

    2024

  44. [44]

    Bakshi, A

    A. Bakshi, A. Liu, A. Moitra, and E. Tang, Structure learning of Hamiltonians from real-time evolution, in2024 IEEE 65th Annual Symposium on Foundations of Computer Science (FOCS) (IEEE, 2024) pp. 1037–1050

    2024

  45. [45]

    M. Ma, S. T. Flammia, J. Preskill, and Y . Tong, Learning k- body Hamiltonians via compressed sensing, arXiv:2410.18928 (2024)

    Pith/arXiv arXiv 2024

  46. [46]

    Stilck França, L

    D. Stilck França, L. A. Markovich, V . V . Dobrovitski, A. H. Werner, and J. Borregaard, Efficient and robust estimation of many-qubit hamiltonians, Nature Communications15, 311 (2024)

    2024

  47. [47]

    Rouzé and D

    C. Rouzé and D. Stilck França, Learning quantum many-body systems from a few copies, Quantum8, 1319 (2024)

    2024

  48. [48]

    W. Yu, J. Sun, Z. Han, and X. Yuan, Robust and efficient Hamil- tonian learning, Quantum7, 1045 (2023)

    2023

  49. [49]

    Hangleiter, I

    D. Hangleiter, I. Roth, J. Fuksa, J. Eisert, and P. Roushan, Ro- bustly learning the Hamiltonian dynamics of a superconducting quantum processor, Nature Communications15, 9595 (2024)

    2024

  50. [50]

    Guo, Y .-K

    S.-A. Guo, Y .-K. Wu, J. Ye, L. Zhang, Y . Wang, W.-Q. Lian, R. Yao, Y .-L. Xu, C. Zhang, Y .-Z. Xu,et al., Hamiltonian learn- ing for 300 trapped ion qubits with long-range couplings, Sci- ence Advances11, 4713 (2025)

    2025

  51. [51]

    Huang, Y

    H.-Y . Huang, Y . Tong, D. Fang, and Y . Su, Learning many-body Hamiltonians with Heisenberg-limited scaling, Physical Review Letters130, 200403 (2023)

    2023

  52. [52]

    Giovannetti, S

    V . Giovannetti, S. Lloyd, and L. Maccone, Quantum metrology, Physical Review Letters96, 010401 (2006)

    2006

  53. [53]

    Giovannetti, S

    V . Giovannetti, S. Lloyd, and L. Maccone, Advances in quantum metrology, Nature Photonics5, 222 (2011)

    2011

  54. [54]

    Demkowicz-Dobrza´nski, J

    R. Demkowicz-Dobrza´nski, J. Kołody´nski, and M. Gu¸ tua, The elusive heisenberg limit in quantum-enhanced metrology, Nature Communications3, 1063 (2012)

    2012

  55. [55]

    S. Zhou, M. Zhang, J. Preskill, and L. Jiang, Achieving the heisenberg limit in quantum metrology using quantum error correction, Nature Communications9, 78 (2018)

    2018

  56. [56]

    Zhao, Learning the structure of any Hamiltonian from mini- mal assumptions, inProceedings of the 57th Annual ACM Sym- posium on Theory of Computing(2025) pp

    A. Zhao, Learning the structure of any Hamiltonian from mini- mal assumptions, inProceedings of the 57th Annual ACM Sym- posium on Theory of Computing(2025) pp. 1201–1211

    2025

  57. [57]

    S. D. Sinha and Y . Tong, Improved Hamiltonian learning and sparsity testing through bell sampling, arXiv:2509.07937 (2025)

    Pith/arXiv arXiv 2025

  58. [58]

    Castaneda and N

    J. Castaneda and N. Wiebe, Hamiltonian Learning via Shadow Tomography of Pseudo-Choi States, Quantum9, 1700 (2025)

    2025

  59. [59]

    S. Liu, X. Wu, and M. Y . Niu, Optimal and robust in- situ quantum hamiltonian learning through parallelization, arXiv:2510.07818 (2025)

    arXiv 2025

  60. [60]

    Dutkiewicz, T

    A. Dutkiewicz, T. E. O’Brien, and T. Schuster, The advantage of quantum control in many-body Hamiltonian learning, Quantum 8, 1537 (2024)

    2024

  61. [61]

    Chen and J

    Z. Chen and J. Li, Lower bounds for Hamiltonian parameter learning from time evolution, arXiv:2509.20665 (2025)

    Pith/arXiv arXiv 2025

  62. [62]

    M. Shin, J. Lee, and C. Oh, Heisenberg-limited hamiltonian learning without short-time control, arXiv:2604.27838 (2026)

    Pith/arXiv arXiv 2026

  63. [63]

    Zhou and W

    T. Zhou and W. Gong, Optimal ansatz-free Hamiltonian learning in situ, arXiv:2606.19486 (2026)

    Pith/arXiv arXiv 2026

  64. [64]

    Boulant, T

    N. Boulant, T. F. Havel, M. A. Pravia, and D. G. Cory, Robust method for estimating the lindblad operators of a dissipative quantum process from measurements of the density operator at multiple time points, Phys. Rev. A67, 042322 (2003)

    2003

  65. [65]

    Howard, J

    M. Howard, J. Twamley, C. Wittmann, T. Gaebel, F. Jelezko, and J. Wrachtrup, Quantum process tomography and linblad estimation of a solid-state qubit, New Journal of Physics8, 33 (2006)

    2006

  66. [66]

    G. O. Samach, A. Greene, J. Borregaard, M. Christandl, J. Bar- reto, D. K. Kim, C. M. McNally, A. Melville, B. M. Niedzielski, Y . Sung, D. Rosenberg, M. E. Schwartz, J. L. Yoder, T. P. Or- lando, J. I.-J. Wang, S. Gustavsson, M. Kjaergaard, and W. D. Oliver, Lindblad tomography of a superconducting quantum processor, Physical Review Applied18, 064056 (2022)

    2022

  67. [67]

    Emerson, R

    J. Emerson, R. Alicki, and K. ˙Zyczkowski, Scalable noise es- timation with random unitary operators, Journal of Optics B: Quantum and Semiclassical Optics7, S347 (2005)

    2005

  68. [68]

    Knill, D

    E. Knill, D. Leibfried, R. Reichle, J. Britton, R. B. Blakestad, J. D. Jost, C. Langer, R. Ozeri, S. Seidelin, and D. J. Wineland, Randomized benchmarking of quantum gates, Physical Review A77, 012307 (2008)

    2008

  69. [69]

    Magesan, J

    E. Magesan, J. M. Gambetta, and J. Emerson, Scalable and ro- bust randomized benchmarking of quantum processes, Physical Rev. Lett.106, 180504 (2011)

    2011

  70. [70]

    Emerson, M

    J. Emerson, M. Silva, O. Moussa, C. Ryan, M. Laforest, J. Baugh, D. G. Cory, and R. Laflamme, Symmetrized char- acterization of noisy quantum processes, Science317, 1893 (2007)

    2007

  71. [71]

    J. J. Wallman and J. Emerson, Noise tailoring for scalable quan- tum computation via randomized compiling, Phys. Rev. A94, 052325 (2016)

    2016

  72. [72]

    Erhard, J

    A. Erhard, J. J. Wallman, L. Postler, M. Meth, R. Stricker, E. A. Martinez, P. Schindler, T. Monz, J. Emerson, and R. Blatt, Char- acterizing large-scale quantum computers via cycle benchmark- ing, Nature Communications10, 5347 (2019)

    2019

  73. [73]

    Helsen, I

    J. Helsen, I. Roth, E. Onorati, A. Werner, and J. Eisert, General framework for randomized benchmarking, PRX Quantum3, 020357 (2022)

    2022

  74. [74]

    S. T. Flammia and J. J. Wallman, Efficient estimation of pauli channels, ACM Transactions on Quantum Computing1, 10.1145/3408039 (2020)

    doi:10.1145/3408039 2020

  75. [75]

    S. T. Flammia and R. O’Donnell, Pauli error estimation via population recovery, Quantum5, 549 (2021)

    2021

  76. [76]

    Harper, W

    R. Harper, W. Yu, and S. T. Flammia, Fast estimation of sparse quantum noise, PRX Quantum2, 010322 (2021)

    2021

  77. [77]

    O’Donnell and S

    R. O’Donnell and S. Sharma, Spam tolerance for pauli error estimation, arXiv:2510.00230 (2026)

    Pith/arXiv arXiv 2026

  78. [78]

    D. S. França, T. Möbus, C. Rouzé, and A. H. Werner, Learning and certification of local time-dependent quantum dynamics and noise, arXiv:2510.08500 (2025)

    arXiv 2025

  79. [79]

    Bairey, C

    E. Bairey, C. Guo, D. Poletti, N. H. Lindner, and I. Arad, Learn- ing the dynamics of open quantum systems from their steady states, New Journal of Physics22, 032001 (2020)

    2020

  80. [80]

    Pastori, T

    L. Pastori, T. Olsacher, C. Kokail, and P. Zoller, Characterization and verification of Trotterized digital quantum simulation via Hamiltonian and Liouvillian learning, PRX Quantum3, 030324 (2022)

    2022

Showing first 80 references.