Pith. sign in

REVIEW 3 major objections 5 minor 61 references

Characterizing Pauli Propagation via Operator Complexity

T0 review · 3 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 Operator entropy sets Pauli truncation cost in spin dynamics

desk verdict The OSE truncation bound is a real result, but it does not cover the algorithm actually run; sequential truncation in Algorithm 1 is not controlled by the one-shot theorem. read the letter →

arxiv 2510.22311 v3 pith:WPF2AP3K submitted 2025-10-25 quant-ph cond-mat.stat-mech

Yuguo Shao , Song Cheng , Zhengwei Liu This is my paper · ORCID
classification quant-phcond-mat.stat-mech MSC 81P6882B2081-08
keywords PaulipropagationHeisenbergpicturesimulationOperatorStabilizerRényientropyTop-KtruncationmodelTrotterizationcomplexityquantummany-body
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 the cost of Pauli-propagation simulation—evolving an observable in the Heisenberg picture and keeping only the K largest Pauli coefficients—is controlled by an intrinsic measure of operator complexity, the Operator Stabilizer Rényi entropy (OSE). It proves an a priori bound: the truncation error decays with the OSE, and it gives an explicit formula for the K needed to achieve a target accuracy. It also proves that for the 1D Heisenberg model with Jz=0, the number of nonzero Pauli coefficients generated from a local operator grows only quadratically in the number of Trotter steps, which explains why aggressive truncation works there. Numerical tests on XXZ chains show high accuracy with small K in the free regime and performance competitive with tensor-network time evolution in the interacting case. The paper acknowledges the bound is loose in practice, but positions OSE as the resource measure that determines when observable-centric simulation beats entanglement-based methods.

What carries the argument

The central object is the Operator Stabilizer Rényi entropy (OSE) S_α(O), defined from the ℓ_α quasi-norm of the squared Pauli coefficients of an operator; it quantifies how spread out the operator is over the Pauli basis. The argument rides on two mechanisms: the tail-bound inequality (Theorem 1), which converts OSE into a concrete Top-K budget via a Markov-type tail estimate, and an invariant subspace argument for the Jz=0 Heisenberg chain (Theorem 2), which shows that the evolving operator stays in a span of words like Z_l, X_i Z...Y_j, bounding the support by O(s^2). The numerical algorithm uses the closed-form conjugation rule for Pauli words under exp(−i w P τ) and a bucket-based appro

What would settle it

Compute, for an n-qubit chain (e.g., n=50) and the Néel state, the exact expectation value ⟨O⟩_ρ(t) of the staggered magnetization and compare with the truncated observable's expectation ⟨Ô⟩_ρ(t) while also computing ‖O−Ô‖_{Pauli,2} exactly. If there exists a time t and a K such that ‖O−Ô‖_{Pauli,2} is below ε (so Theorem 1's K prescription is satisfied) but |⟨O⟩−⟨Ô⟩| is much larger than ε (e.g., by a factor 2^{n/2}), then the paper's accuracy claim is not explained by the theorem. Conversely, if |⟨O⟩−⟨Ô⟩| stays small even when ‖O−Ô‖_{Pauli,2} is large, the proxy is empirically valid for this

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is the inequality that links the tail of the squared Pauli coefficients of an observable to its OSE: for 0<α<1, ln Δ(K) ≤ (1−α)/α (S_α(O) − ln K) + ln(α/(1−α)). This implies that keeping K ≥ exp(S_α(O)) (2α/((1−α)ε^2))^{α/(1−α)} Pauli terms guarantees the normalized Pauli-2 truncation error is at most ε. The companion structural result is that for the Trotterized 1D Heisenberg chain with Jz=0, the Heisenberg-evolved local operator U_s^† Z_l U_s has only O(s^2) nonzero Pauli coefficients, so the exact support remains polynomially small rather than exponential. The paper reads these two facts together as an explanation for why the Top-K Pauli-pro

Load-bearing premise

The paper assumes that the normalized Pauli-2 truncation error bound (which Theorem 1 controls) is a faithful proxy for the error in the expectation value ⟨O⟩_ρ(t) that the method actually reports; for a generic pure state this proxy can fail by an exponentially large factor, so nothing in the proof forces the numerical accuracy at small K.

Editorial extensions

If this is right

  • If Theorem 1 is right, simulation cost for local observables is set by operator complexity (OSE) rather than state entanglement, so problems with low OSE should be classically simulable even when entanglement grows.
  • The K prescription in Eq. (13) gives a parameter-free a priori way to pick the truncation budget before running the simulation, once S_α(O) is known or estimated.
  • Theorem 2 shows a concrete family—1D Heisenberg at Jz=0—where Pauli propagation is guaranteed to be polynomially efficient in Trotter steps, a benchmark where operator-centric methods should beat matrix-product-state time evolution.
  • The bound separates cleanly from Trotter error, so using higher-order Trotter formulas or exact evolution would leave OSE as the dominant limiting factor.
  • The framework suggests analogous resource measures for other operator bases or for open-system dynamics, where support growth may be similarly tame.

Reading between the lines

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

  • Editorial inference: The proved bound controls the normalized Pauli-2 distance between O and Ô, not directly the expectation-value error |⟨O⟩−⟨Ô⟩| for a generic state; for pure n-qubit states the Cauchy–Schwarz factor can be 2^{n/2}, so the numerical accuracy at K≈2^12 is not fully explained by Theorem 1 alone—additional structure (e.g., the low support of the Néel state) must be doing work.
  • Editorial extension: Since the Jz=0 theorem gives exact support O(s^2), a fixed-accuracy simulation up to time T with Trotter step count N should cost O(N^3 n) overall once K is matched to the support; extrapolating this to late times yields polynomial scaling that the paper does not spell out.
  • Editorial testable extension: For the interacting Jz≠0 case, one could measure S_1/2(O) at several system sizes and check whether the minimal K to reach a fixed ε tracks exp(S_α) as Eq. (13) predicts; the paper only reports loose overestimates, so this is a direct way to probe the tightness of the bound.
  • Editorial connection: The same tail-argument technique should extend to other truncation rules, such as Hamming-weight truncation; the paper gives a weight-truncation bound for local-scrambling ensembles but does not link it to OSE, leaving a natural unification open.
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

3 major / 5 minor

Summary. The paper introduces a Pauli-propagation algorithm for simulating real-time spin dynamics in the Heisenberg picture, with a Top-K truncation strategy that keeps only the K largest Pauli coefficients at each step. The main theoretical results are: (i) Theorem 1, which bounds the one-shot Top-K truncation error in normalized Pauli-2 distance in terms of the Operator Stabilizer Rényi entropy (OSE); (ii) Theorem 2, which states that for the 1D Heisenberg model with J_z=0, the number of nonzero Pauli coefficients of a Trotter-evolved local operator grows at most quadratically in the number of Trotter steps. The paper benchmarks the method on 1D XXZ chains, reporting high accuracy with small K in the free regime and competitive performance against TDVP in the interacting case. The authors claim that OSE governs the resource cost of Pauli propagation.

Significance. If the central claim were fully established, the paper would make a useful contribution: a clean inequality linking truncation error to OSE, a structure theorem for the J_z=0 Heisenberg chain, and numerical evidence that Pauli propagation can outperform tensor networks in certain regimes. The proof of Theorem 1 is a neat analytic bound, and Theorem 2, if completed, provides a concrete example of operator compressibility. However, the paper currently contains a substantial gap between the proved one-shot truncation bound and the repeated-truncation algorithm actually run, as well as a mismatch between the norm in which the bound is stated and the expectation-value error used in benchmarks. These issues are load-bearing for the advertised claim that OSE governs the truncation error of the method.

major comments (3)
  1. [Algorithm 1; Section III] Lemma 2 and Theorem 1 bound a single, final truncation of the exact evolved operator O: they keep the K largest coefficients of O, renormalize according to Eq. (10), and control ||O−Ô||_{Pauli,2}. Algorithm 1 instead truncates after every elementary conjugation inside every Trotter step (the inner loop over i=1..N_H), i.e., N·N_H truncations with no intermediate renormalization. The final operator is therefore not the one-shot Top-K projection of the true O0, and no error-propagation bound is supplied. To justify the central claim that OSE governs the error of the method, the authors would need a bound of the form E_{s−1} ≤ E_s + δ_s with δ_s controlled by the OSE of the intermediate (truncated) operators, plus a contraction argument. As written, the proved inequality concerns a milder procedure than the one benchmarked.
  2. [Section IV, Eq. (13)] The K prescription in Eq. (13) is derived from ||O−Ô||_{Pauli,2} ≤ ε. The numerical benchmarks report |⟨O⟩_ρ − ⟨Ô⟩_ρ| for a pure Néel state. For a pure n-qubit state the two quantities are related only with a factor up to 2^{n/2}: |tr{(O−Ô)ρ}| ≤ 2^{n/2} ||O−Ô||_{Pauli,2}. With n=50 this factor is about 3×10^7, so a target ε=10^{-3} in Eq. (13) does not by itself imply small expectation-value error. The observed high accuracy at K=2^{12}–2^{19} may be true, but it is not a consequence of Theorem 1; a separate argument controlling the observable error is needed.
  3. [Appendix E] The proof of Theorem 2 hinges on the claim that the subspace span{Z_l, J_{l,m}, H_{l,m}, K_{l,m}, L_{l,m}} is invariant under U_{j,j+1}(θ) and V_{j,j+1}(θ). The text verifies the action on Z_l and says the other cases 'can be verified similarly.' Since the conclusion O(s^2) depends on closure for all five operator types under both gates, the proof is incomplete as written. A compact table of the conjugation action (or a structural argument) should be supplied.
minor comments (5)
  1. [Appendix B] Lemma 1(4) states convexity of exp(Sα), but the proof concludes 'we obtain exp(Sα(·)) is a concave function'. This is a typo; the argument actually proves convexity.
  2. [Fig. 2 caption] The caption uses 'PORE' where the text uses 'OSE'. Please correct.
  3. [Eq. (2)] The Trotter error is O(τ^2) per step, so the accumulated error is O(N τ^2)=O(t τ). The text writes O(τ^2) at Eq. (2) without stating the accumulated scaling; this should be clarified, especially since the exact reference is taken at finite τ.
  4. [Fig. 1 caption] The 'exact' reference curves are said to be obtained using TDVP without bond-dimension truncation. Please clarify whether this means a converged TDVP reference or exact diagonalization; the current wording is ambiguous.
  5. [Fig. 3 and Theorem 2] Figure 3(a) shows the number of Pauli words growing roughly linearly in time for J_z=0, while Theorem 2 gives an O(s^2) upper bound. The authors may wish to comment on this discrepancy (e.g., the linear regime may dominate at the simulated times).

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: Theorem 1 is an analytic tail bound, Theorem 2 is a direct structural proof; the theorem/algorithm mismatch is an applicability gap, not circularity.

full rationale

The central derivation (Lemma 2 + Theorem 1, Sec. III, Eqs. 10–13) is a self-contained analytic inequality between the tail sum Δ(K) and the OSE Sα(O) for the same exact Pauli-coefficient distribution. No parameter is fitted to data and then renamed as a prediction; the bound is a mathematical consequence of sorted coefficients and does not use the algorithm's truncated output as an input. The OSE definition is cited to external work [49], and the heavy use of current-author citations [42,48] is contextual for the Pauli-propagation framework, not load-bearing for Theorem 1 or 2. Theorem 2 is proven explicitly in Appendix E via an invariant subspace and light cone, not imported from a self-citation. I considered the skeptic's objection that Algorithm 1 applies Top-K truncation after every elementary gate while Theorem 1 bounds only a single final truncation; this is a real correctness/applicability gap (and the expectation-value metric gap noted in the reader's take is also real), but it is not a circular reduction: the theorem is not equivalent to its input by construction, and no equation in the paper conflates the one-shot truncation with the repeated-truncation output as a definitional identity. The paper itself labels the K prescription as a 'loose theoretical estimate' (Sec. IV), further showing no fitted-input-called-prediction claim. Therefore, no enumerated circularity step is present.

Assumptions & free parameters 2 free parameters · 6 assumptions · 0 invented entities

The core inequality Thm. 1 uses only standard ℓ_p norm facts. The practical and structural claims additionally rest on: exactness of the Trotterized reference, the unproved subspace invariance, the norm-rescaling heuristic, and the equivalence between approximate and exact Top-K selection. None of these are fitted parameters, but they are assumptions the reader buys without paying for them.

free parameters (2)
  • Top-K budget K = 2^12 (Jz=0), 2^19 (Jz=0.5)
    Chosen by hand in the simulations; the Theorem 1 prescription gives K>8.7e8 at Jz=0 and K>1.1e12 at Jz=0.5, orders of magnitude larger than what is used, so the practical K is not derived from the theory.
  • Fig. 3 fit parameters (a,b,n) = a=1959.2, b=2.87 (exp); n=4.75 (poly)
    Descriptive fits to the Jz=0.5 Pauli-term growth curve; quoted to support the growth-trend statement but not load-bearing for the theorems.
assumptions (6)
  • standard math Standard inequalities: Markov/Minkowski and the definition of Sα(O) as an ℓ_p norm of Pauli coefficients
    Used in proof of Thm. 1 and Lemma 1; uncontroversial but assumed without comment.
  • domain assumption First-order Trotterization has error O(τ²) and is neglected in the accuracy discussion
    Eq. (2) states the Trotter decomposition; Sec. IV says Trotter errors are disregarded, so claimed accuracy refers only to truncation.
  • domain assumption The system starts in the Néel product state and only the staggered magnetization is considered
    The numerical benchmarks and cost claims are tied to this choice; the theory does not give expectation-error bounds for general initial states.
  • ad hoc to paper The subspace span{Z_l, J_{l,m}, H_{l,m}, K_{l,m}, L_{l,m}} is invariant under conjugation by the Trotter gates
    Appendix E says the action 'can be verified similarly' but does not prove it; this is the key structural input to Theorem 2.
  • ad hoc to paper Norm rescaling after truncation preserves the accuracy of expectation values
    Eq. (8) and Lemma 2 use rescaling to enforce Hilbert–Schmidt norm conservation; no proof that this is optimal or that it controls observable expectation error.
  • ad hoc to paper Bucket-based approximate Top-K selection behaves like exact Top-K for the theoretical bound
    Appendix A describes an approximate Top-K with magnitude buckets, while Lemma 2/Thm 1 assume exact top-K coefficients.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Characterizing Pauli Propagation via Operator Complexity." pith.science (2026). https://pith.science/paper/WPF2AP3K

@misc{pith2026251022311,
  author       = {Pith},
  title        = {Pith review of: Characterizing Pauli Propagation via Operator Complexity},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WPF2AP3K}},
  note         = {Machine review of arXiv:2510.22311}
}
abstract

Pauli-propagation simulation represents observables in the Pauli basis and evolves their coefficients in the Heisenberg picture. Its efficiency depends on whether the evolving operator can be accurately compressed by retaining only a limited number of Pauli terms. In this work, we bridge operator complexity and the resource cost of Pauli-propagation methods by proving that the truncation error is governed by the Operator Stabilizer R\'enyi entropy (OSE) $\mathcal{S}^\alpha(O)$. Our a priori bounds quantify how OSE controls the compressibility of the evolving operator and give explicit prescriptions for the Top-$K$ budget required to achieve a target accuracy. As an analytic test case, we prove that for the 1D Heisenberg model at $J_z=0$, the number of non-zero Pauli coefficients generated from a local operator grows at most quadratically with the number of Trotter steps. We then benchmark the Top-$K$ Pauli propagation on XXZ Heisenberg chains. The numerical results show high accuracy with a small truncation number $K$ in the free regime ($J_z=0$) and competitive performance against tensor-network methods, such as TDVP, in the interacting case ($J_z=0.5$). These results position OSE as a resource measure for Pauli-propagation methods.

Figures

Figures reproduced from arXiv: 2510.22311 by the authors.

Figure 1
Figure 1. Time-evolution results for the one-dimensional [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. α = 1/2 and α → 1(Shannon) OSE over time. Where the blue line denotes the entropy of the Pauli Z opera￾tor with time evolution on the open boundary, and the orange line denotes the entropy corresponding to the Pauli Z opera￾tor with time evolution at the center of the one-dimensional chain. In [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Growth behavior of Pauli words over time for dif [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Distribution of the squared Pauli coefficients as a [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Pauli weight distribution of the Pauli words coefficients as a function of time for different values of [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

61 extracted references · 10 linked inside Pith

  1. [1]

    J. I. Cirac and P. Zoller, Goals and opportunities in quan- tum simulation, Nature physics8, 264 (2012)

    2012

  2. [2]

    For a Clifford circuitU, we haveS α(U OU†) = S α(O)

  3. [3]

    For0< α <0.5,S α(O)is concave, i.e., for anyP i λi = 1withλ i ≥0, we haveS α(P i λiOi)≥P i λiS α(Oi)

  4. [4]

    We now establish the following error bound for the Top-Ktruncation: Lemma 2.LetO= P P∈{I,X,Y,Z} ⊗n cP P, wherec P = 1 2n tr{P O}

    For0.5≤α <1, exp(S α(O))is convex, i.e., for any P i λi = 1withλ i ≥0, we have exp(S α(P i λiOi))≤ P i λiexp(S α(Oi)). We now establish the following error bound for the Top-Ktruncation: Lemma 2.LetO= P P∈{I,X,Y,Z} ⊗n cP P, wherec P = 1 2n tr{P O}. The approximated observable is given by: bO= ∥O∥2q 2n PK i=1 c′2 i KX i=1 c′ iP ′ i ,(10) where{c ′ i}and{P ...

    1959

  5. [5]

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

    2014

  6. [6]

    Polkovnikov, K

    A. Polkovnikov, K. Sengupta, A. Silva, and M. Vengalat- tore, Colloquium: Nonequilibrium dynamics of closed in- teracting quantum systems, Reviews of Modern Physics 83, 863 (2011)

    2011

  7. [7]

    Eisert, M

    J. Eisert, M. Friesdorf, and C. Gogolin, Quantum many- body systems out of equilibrium, Nature Physics11, 124 (2015)

    2015

  8. [8]

    Miessen, P

    A. Miessen, P. J. Ollitrault, F. Tacchino, and I. Taver- nelli, Quantum algorithms for quantum dynamics, Na- ture Computational Science3, 25 (2023)

    2023

Show all 61 references
  1. [9]

    A. J. Daley, I. Bloch, C. Kokail, S. Flannigan, N. Pearson, M. Troyer, and P. Zoller, Practical quantum advantage in quantum simulation, Nature607, 667 (2022)

    2022

  2. [10]

    Fauseweh, Quantum many-body simulations on digi- tal quantum computers: State-of-the-art and future chal- lenges, Nature Communications15, 2123 (2024)

    B. Fauseweh, Quantum many-body simulations on digi- tal quantum computers: State-of-the-art and future chal- lenges, Nature Communications15, 2123 (2024)

    2024

  3. [11]

    A. M. Childs, J. Leng, T. Li, J.-P. Liu, and C. Zhang, Quantum simulation of real-space dynamics, Quantum6, 860 (2022)

    2022

  4. [12]

    H.-J. Liao, K. Wang, Z.-S. Zhou, P. Zhang, and T. Xiang, Simulation of ibm’s kicked ising experiment with projected entangled pair operator, arXiv preprint arXiv:2308.03082 (2023)

    2023 arXiv

  5. [13]

    Weimer, A

    H. Weimer, A. Kshetrimayum, and R. Or´ us, Simulation methods for open quantum many-body systems, Reviews of Modern Physics93, 015008 (2021)

    2021

  6. [14]

    D’Alessio, Y

    L. D’Alessio, Y. Kafri, A. Polkovnikov, and M. Rigol, From quantum chaos and eigenstate thermalization to statistical mechanics and thermodynamics, Advances in Physics65, 239 (2016)

    2016

  7. [15]

    Gogolin and J

    C. Gogolin and J. Eisert, Equilibration, thermalisation, and the emergence of statistical mechanics in closed quantum systems, Reports on Progress in Physics79, 056001 (2016)

    2016

  8. [16]

    Vasseur and J

    R. Vasseur and J. E. Moore, Nonequilibrium quantum dynamics and transport: from integrability to many- body localization, Journal of Statistical Mechanics: The- ory and Experiment2016, 064010 (2016)

    2016

  9. [17]

    T. Mori, T. N. Ikeda, E. Kaminishi, and M. Ueda, Thermalization and prethermalization in isolated quan- tum systems: a theoretical overview, Journal of Physics B: Atomic, Molecular and Optical Physics51, 112001 (2018)

    2018

  10. [18]

    D. A. Abanin, E. Altman, I. Bloch, and M. Serbyn, Col- loquium: Many-body localization, thermalization, and entanglement, Reviews of Modern Physics91, 021001 (2019)

    2019

  11. [19]

    Swingle, Unscrambling the physics of out-of-time- order correlators, Nature Physics14, 988 (2018)

    B. Swingle, Unscrambling the physics of out-of-time- order correlators, Nature Physics14, 988 (2018)

    2018

  12. [20]

    X. Mi, P. Roushan, C. Quintana, S. Mandra, J. Mar- shall, C. Neill, F. Arute, K. Arya, J. Atalaya, R. Bab- bush,et al., Information scrambling in quantum circuits, Science374, 1479 (2021)

    2021

  13. [21]

    Zanardi and N

    P. Zanardi and N. Anand, Information scrambling and chaos in open quantum systems, Physical Review A103, 062214 (2021)

    2021

  14. [22]

    Waintal, M

    X. Waintal, M. Wimmer, A. Akhmerov, C. Groth, B. K. Nikolic, M. Istas, T. ¨O. Rosdahl, and D. Var- jas, Computational quantum transport, arXiv preprint arXiv:2407.16257 (2024)

    2024 arXiv

  15. [23]

    Shevtsov and X

    O. Shevtsov and X. Waintal, Numerical toolkit for elec- tronic quantum transport at finite frequency, Physical Review B—Condensed Matter and Materials Physics87, 085304 (2013)

    2013

  16. [24]

    R. P. Feynman, Simulating physics with computers, in Feynman and computation(cRc Press, 2018) pp. 133– 153

    2018

  17. [25]

    Z. Feng, Z. Liu, F. Lu, and N. Wang, Quon classical sim- ulation: Unifying clifford, matchgates and entanglement, arXiv preprint arXiv:2505.07804 (2025)

    2025 arXiv

  18. [26]

    B. Kang, C. Zhao, Z. Liu, X. Gao, and S. Choi, 2d quon language: Unifying framework for cliffords, matchgates, and beyond, arXiv preprint arXiv:2505.06336 (2025)

    2025 arXiv

  19. [27]

    X. Xu, S. Benjamin, J. Sun, X. Yuan, and P. Zhang, A herculean task: Classical simulation of quantum comput- ers, arXiv preprint arXiv:2302.08880 (2023)

    2023 arXiv

  20. [28]

    Biamonte and V

    J. Biamonte and V. Bergholm, Tensor networks in a nut- shell, arXiv preprint arXiv:1708.00006 (2017)

    2017 arXiv

  21. [29]

    Or´ us, Tensor networks for complex quantum systems, Nature Reviews Physics1, 538 (2019)

    R. Or´ us, Tensor networks for complex quantum systems, Nature Reviews Physics1, 538 (2019)

    2019

  22. [30]

    Or´ us, A practical introduction to tensor networks: Matrix product states and projected entangled pair states, Annals of physics349, 117 (2014)

    R. Or´ us, A practical introduction to tensor networks: Matrix product states and projected entangled pair states, Annals of physics349, 117 (2014)

    2014

  23. [31]

    Verstraete and J

    F. Verstraete and J. I. Cirac, Matrix product states 7 represent ground states faithfully, Physical Review B—Condensed Matter and Materials Physics73, 094423 (2006)

    2006

  24. [32]

    Paeckel, T

    S. Paeckel, T. K¨ ohler, A. Swoboda, S. R. Manmana, U. Schollw¨ ock, and C. Hubig, Time-evolution methods for matrix-product states, Annals of Physics411, 167998 (2019)

    2019

  25. [33]

    Schollw¨ ock, The density-matrix renormalization group, Reviews of modern physics77, 259 (2005)

    U. Schollw¨ ock, The density-matrix renormalization group, Reviews of modern physics77, 259 (2005)

    2005

  26. [34]

    Schollw¨ ock, The density-matrix renormalization group in the age of matrix product states, Annals of physics 326, 96 (2011)

    U. Schollw¨ ock, The density-matrix renormalization group in the age of matrix product states, Annals of physics 326, 96 (2011)

    2011

  27. [35]

    X. Qian, J. Huang, and M. Qin, Augmenting density ma- trix renormalization group with clifford circuits, Physical Review Letters133, 190402 (2024)

    2024

  28. [36]

    Haegeman, C

    J. Haegeman, C. Lubich, I. Oseledets, B. Vandereycken, and F. Verstraete, Unifying time evolution and optimiza- tion with matrix product states, Physical Review B94, 165116 (2016)

    2016

  29. [37]

    Haegeman, J

    J. Haegeman, J. I. Cirac, T. J. Osborne, I. Piˇ zorn, H. Ver- schelde, and F. Verstraete, Time-dependent variational principle for quantum lattices, Physical Review Letters 107, 070601 (2011)

    2011

  30. [38]

    X. Qian, J. Huang, and M. Qin, Clifford circuits aug- mented time-dependent variational principle, Physical Review Letters134, 150404 (2025)

    2025

  31. [39]

    ˇZnidariˇ c, Entanglement growth in diffusive systems, Communications Physics3, 100 (2020)

    M. ˇZnidariˇ c, Entanglement growth in diffusive systems, Communications Physics3, 100 (2020)

    2020

  32. [40]

    Nahum, J

    A. Nahum, J. Ruhman, S. Vijay, and J. Haah, Quantum entanglement growth under random unitary dynamics, Physical Review X7, 031016 (2017)

    2017

  33. [41]

    Erhard, M

    M. Erhard, M. Krenn, and A. Zeilinger, Advances in high-dimensional quantum entanglement, Nature Re- views Physics2, 365 (2020)

    2020

  34. [42]

    Aharonov, X

    D. Aharonov, X. Gao, Z. Landau, Y. Liu, and U. Vazi- rani, A polynomial-time classical algorithm for noisy ran- dom circuit sampling, inProceedings of the 55th Annual ACM Symposium on Theory of Computing(2023) pp. 945–957

    2023

  35. [43]

    Schuster, C

    T. Schuster, C. Yin, X. Gao, and N. Y. Yao, A polynomial-time classical algorithm for noisy quantum circuits, arXiv preprint arXiv:2407.12768 (2024)

    2024 arXiv

  36. [44]

    Gao and L

    X. Gao and L. Duan, Efficient classical simula- tion of noisy quantum computation, arXiv preprint arXiv:1810.03176 (2018)

    2018 arXiv

  37. [45]

    Y. Shao, F. Wei, S. Cheng, and Z. Liu, Simulating noisy variational quantum algorithms: A polynomial approach, Physical Review Letters133, 120603 (2024)

    2024

  38. [46]

    Angrisani, A

    A. Angrisani, A. Schmidhuber, M. S. Rudolph, M. Cerezo, Z. Holmes, and H.-Y. Huang, Classically es- timating observables of noiseless quantum circuits, arXiv preprint arXiv:2409.01706 (2024)

    2024

  39. [47]

    Angrisani, A

    A. Angrisani, A. A. Mele, M. S. Rudolph, M. Cerezo, and Z. Holmes, Simulating quantum circuits with arbi- trary local noise using pauli propagation, arXiv preprint arXiv:2501.13101 (2025)

    2025 arXiv

  40. [48]

    Fontana, M

    E. Fontana, M. S. Rudolph, R. Duncan, I. Rungger, and C. C ˆ ırstoiu, Classical simulations of noisy varia- tional quantum circuits, npj Quantum Information11, 1 (2025)

    2025

  41. [49]

    Y. Shao, Z. Chen, Z. Wei, and Z. Liu, Diagnosing quan- tum circuits: Noise robustness, trainability, and express- ibility, arXiv preprint arXiv:2509.11307 (2025)

    2025

  42. [50]

    Martinez, A

    V. Martinez, A. Angrisani, E. Pankovets, O. Fawzi, and D. Stilck Fran¸ ca, Efficient simulation of parametrized quantum circuits under nonunital noise through pauli backpropagation, Physical Review Letters134, 250602 (2025)

    2025

  43. [51]

    Zhang, Y

    R. Zhang, Y. Shao, F. Wei, S. Cheng, Z. Wei, and Z. Liu, Clifford perturbation approximation for quantum error mitigation, arXiv preprint arXiv:2412.09518 (2024)

    2024 arXiv

  44. [52]

    Dowling, P

    N. Dowling, P. Kos, and X. Turkeshi, Magic resources of the heisenberg picture, Physical Review Letters135, 050401 (2025)

    2025

  45. [53]

    Leone, S

    L. Leone, S. F. Oliviero, and A. Hamma, Stabilizer r´ enyi entropy, Physical Review Letters128, 050402 (2022)

    2022

  46. [54]

    E. Lieb, T. Schultz, and D. Mattis, Two soluble models of an antiferromagnetic chain, Annals of Physics16, 407 (1961)

    1961

  47. [55]

    Katsura, Statistical mechanics of the anisotropic linear heisenberg model, Physical Review127, 1508 (1962)

    S. Katsura, Statistical mechanics of the anisotropic linear heisenberg model, Physical Review127, 1508 (1962)

    1962

  48. [56]

    Fishman, S

    M. Fishman, S. R. White, and E. M. Stoudenmire, The ITensor Software Library for Tensor Network Calcula- tions, SciPost Phys. Codebases , 4 (2022). 8 Appendix A: Computational Cost We analyze the computational complexity of Algorithm 1 in terms of the number of qubitsn, number...

    2022

  49. [57]

    ForP=P 1 ⊗P 2, we havec P =c P1 cP2 forP 1 ∈ {I, X, Y, Z}⊗n1 and P2 ∈ {I, X, Y, Z}⊗n2

    We consider the tensor product of two observablesO=O 1 ⊗O 2, and denote the Pauli coefficients ofO,O 1 andO 2 asc,c 1 andc 2 respectively. ForP=P 1 ⊗P 2, we havec P =c P1 cP2 forP 1 ∈ {I, X, Y, Z}⊗n1 and P2 ∈ {I, X, Y, Z}⊗n2 . Therefore, we have: c2 α α = X P1∈{I,X,Y,Z} ⊗n1 X ...

  50. [58]

    For Q=U †P UwithP∈ {I, X, Y, Z}⊗n, by the property of the Clifford circuit, we haveQ∈ {I, X, Y, Z}⊗n and c′ Q =c P

    Consider the conjugation ofOby a Clifford circuitU, we denote the Pauli coefficients ofU OU † asc ′. For Q=U †P UwithP∈ {I, X, Y, Z}⊗n, by the property of the Clifford circuit, we haveQ∈ {I, X, Y, Z}⊗n and c′ Q =c P . Therefore, we have: c′2 α = c2 α.(B3) Taking into the entro...

  51. [59]

    The Pauli coefficients ofc i(O) for operatorOis defined asc i(O) = 1 2n tr{PiO}

    LetO= P i λiOi be the weighted sum of the operatorsO i. The Pauli coefficients ofc i(O) for operatorOis defined asc i(O) = 1 2n tr{PiO}. We have: ci(O) = X j λjci(Oj),(B5) wherec i(Oj) = 1 2n tr{PiOj}. Since 0< α <0.5, the functionf(x) =x 2α is concave, by Jensen’s inequality,...

  52. [60]

    On the other hand, we have: exp (Sα(O)) = X i |ci(O)|2α ! 1 1−α = X i |ci(O)|q ! 1 1−α =∥c(O)∥ r q,(B12) whereq= 2α,r= 2α 1−α andc(O) = (c 1(O), c2(O),

    Similarly, we have: ci(O) = X j λjci(Oj),(B11) wherec i(Oj) = 1 2n tr{PiOj}. On the other hand, we have: exp (Sα(O)) = X i |ci(O)|2α ! 1 1−α = X i |ci(O)|q ! 1 1−α =∥c(O)∥ r q,(B12) whereq= 2α,r= 2α 1−α andc(O) = (c 1(O), c2(O), . . . , c4n (O)). By Minkowski inequality, we ha...

  53. [61]

    [43], except that there is a normalization procedure after truncation in our method

    Proof of Theorem 3 Proof.The proof is similar to that in Ref. [43], except that there is a normalization procedure after truncation in our method. Byρ=V|0⟩ ⟨0| ⊗n V † withVsampled from a local-scrambling ensemble, we have: Eρ ⟨ bO⟩ρ − ⟨O⟩ρ 2 =E V tr n V †( bO−O)V|0⟩ ⟨0|⊗n o 2 ...

Pith tools

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