Pith. sign in

REVIEW 3 major objections 4 minor 5 cited by

Optimal quantum simulation of linear non-unitary dynamics

T0 review · 3 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read A near-exact exponential kernel makes non-unitary quantum simulation query-optimal.

desk verdict Real algorithmic advance in LCHS, but the abstract's 'optimal in all parameters' is an overstatement; the core theorems deserve serious refereeing. read the letter →

arxiv 2508.19238 v2 pith:D2MDFKWN submitted 2025-08-26 quant-ph

Guang Hao Low , Rolando D. Somma This is my paper
classification quant-ph MSC 81P68 PACS 03.67.Lx
keywords non-unitarydynamicslinearcombinationofHamiltoniansimulationblock-encodingquerycomplexityquantumlogarithmicnormexponentialdecayapproximationtrapezoidalrule
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

This paper presents a quantum algorithm for simulating time evolution generated by bounded, time-dependent operators -A whose Hermitian part L=(A+A†)/2 is positive semidefinite, generalizing Hamiltonian simulation. It proves that the evolution operator e^{-At} can be block-encoded with O(t log(1/ε)) queries to a block-encoding of A, optimal in all parameters and matching the best Hamiltonian simulation scaling. The key move is relaxing the exact Fourier representation of exponential decay to an approximate one, which bypasses a previously proven no-go result for the Linear-Combination-of-Hamiltonian-Simulation framework. A uniform trapezoidal rule is shown to converge exponentially, simplifying circuit implementation, and new lower bounds show further constant-factor improvements beyond roughly a factor of three are infeasible.

What carries the argument

The central object is the LCHS integral O_R(t) = (1/√(2π)) ∫_{-R}^{R} f̂(k) T exp(-i ∫_0^t (kL(s)+H(s))ds) dk, which approximates U0(t). The paper's advance is replacing the exact exponential-decay identity e^{-x} = (1/√(2π))∫ f̂(k)e^{-ikx}dk with an approximate version, using entire kernel functions like f2(x;γ,c) = e^c f2(x+c;γ) that only approximate e^{-x} for x≥0 but have Fourier transforms with bounded L1 norm and fast decay. A residue theorem argument on a strip in the complex plane converts the approximate scalar identity into a rigorous matrix-valued error bound.

What would settle it

Take a concrete small matrix A = L + iH with L⪰0, e.g. A = I, and numerically evaluate the two error integrals in Theorem 1 for the f̂2 kernel using the parameters prescribed by Theorem 2 at ε = 0.1; if their sum exceeds the claimed ε, the theorem's constants are wrong. For the L⪰0 condition, simulate the same LCHS construction on a 1×1 generator A = -γ with γ>0 (where L = -γ has a negative eigenvalue) and check whether the empirical error and query count still match O(t log(1/ε)) — if they degrade, the premise is load-bearing.

Watch

Extended reading notes

Core claim

For any t≥0 and A = L + iH with L⪰0, the LCHS formed from the kernel function f̂2(k;γ,c) = √(2/π) e^{c(1-ik)}/(1+k²) e^{-(k²+1)/(4γ²)} approximates the time-ordered evolution U0(t) = T exp(-∫ A(s)ds) to error ε, with normalization e^c and truncation radius R = O(log(1/ε)). Consequently, for time-independent A, e^{-At} can be block-encoded using O(t log(1/ε)) queries to the block-encoding oracle of A, which is optimal in evolution time, error, and norm. The paper also proves exponential convergence of a uniform trapezoidal discretization of the LCHS integral, gives an explicit gate-efficient circuit, and establishes query lower bounds showing the algorithm's constant factor is within about a

Load-bearing premise

The whole error analysis leans on the positive semidefiniteness of L = (A+A†)/2; if L has negative eigenvalues, the norm bounds on the time-ordered exponentials fail and the query-optimal block-encoding of e^{-At} is not preserved.

Editorial extensions

If this is right

  • Time-independent non-unitary evolution e^{-At} with L⪰0 can be block-encoded with O(t log(1/ε)) queries, matching Hamiltonian simulation complexity and improving on all prior algorithms.
  • The normalized evolved state |u(t)⟩ can be prepared with O(1/‖e^{-At}|u0⟩‖) queries to the initial-state preparation oracle, which is optimal.
  • A uniform trapezoidal rule converges exponentially fast with step size h = O(1/(‖L‖_L1 + log(1/ε))), yielding simpler circuits than previous non-uniform quadrature methods.
  • Any further improvement of the LCHS approach by more than a constant factor of approximately three is ruled out by new lower bounds based on quantum singular value transformations.
  • The method extends to time-dependent A and to inhomogeneous equations with logarithmic overhead, though the time-dependent case inherits the suboptimal scaling of general time-dependent Hamiltonian simulation.

Reading between the lines

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

  • The same approximate-exponential-kernel idea could be applied to other Laplace-transform-based quantum algorithms, potentially improving eigenvalue transformation and differential equation solvers beyond e^{-At}.
  • The kernel family f̂_{j,y} is not exhausted; further numerical or analytic search over kernel parameters may close the remaining constant-factor gap toward the QSVT lower bound.
  • If access to a block-encoding of √L is available, the LCHS approach might be adapted to achieve a square-root speedup with respect to L, a question the paper leaves open.
  • For operators where L has negative eigenvalues, the paper's shift strategy changes the success probability; a testable extension is whether a different encoding or error analysis can restore query-optimality without the L⪰0 premise.
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 / 4 minor

Summary. The paper presents a quantum algorithm for simulating time evolution generated by bounded, time-dependent operators -A with non-positive logarithmic norm, i.e. A = L + iH with L >= 0. It generalizes the LCHS framework by allowing kernel functions that only approximately reproduce exponential decay, overcoming a previous no-go result. For time-independent A, it gives a block-encoding of e^{-At} with O(t log(1/eps)) queries to the block-encoding oracle of A, and normalized state preparation with O(1/||e^{-At}|u0>||) queries. It also proves exponential convergence of a uniform trapezoidal rule, gives explicit circuit constructions, and presents lower bounds intended to show optimality up to a small constant factor, partly based on numerical polynomial-approximation computations.

Significance. The main algorithmic contribution is significant: Theorems 1-4 provide a rigorous, self-contained LCHS construction with optimal-looking scaling for non-unitary evolution under L >= 0, improving on prior polynomial or near-optimal error dependencies. The approximate-decay kernel idea is a genuine conceptual advance, and the uniform-quadrature theorem simplifies implementations. The paper also gives an explicit circuit compilation and a non-trivial search lower bound for state preparation. However, the advertised claim of being 'optimal in all parameters' goes beyond what is proven: in the Hermitian limit the method is suboptimal by a log(1/eps) factor relative to known optimal Hamiltonian simulation, and the claimed constant-factor lower bounds are empirical rather than theorems.

major comments (3)
  1. [Abstract; §1, Theorem 2; Lemma 16; §7] The claim that the query complexities are 'optimal in all parameters' is not supported and is contradicted by an admissible instance within the stated scope. Take A = iH, so L = 0 and A is anti-Hermitian. Then Theorem 2 and Lemma 16 give a block-encoding of e^{-iHt} with O(alpha_H t log(1/eps)) queries, whereas optimal Hamiltonian simulation [LC17b, LC19] achieves O(alpha_H t + log(1/eps)) queries. Thus the LCHS algorithm is not optimal in the Hamiltonian limit. The body effectively concedes this in §7, where the Hamiltonian case is set aside because it 'will not give useful lower bounds on LCHS'. The abstract and introduction should be revised to state the precise optimality claim, e.g. optimal scaling of LCHS for L >= 0, or optimal up to a logarithmic factor in the unitary limit, rather than 'optimal in all parameters'.
  2. [§7, Eq. (181)-(185), Figure 3] The claim that 'any improvement of our approach exceeding a constant factor of approximately 3 is infeasible' is based on numerical linear-programming fits, not on a theorem. The text explicitly calls the scaling min_f deg(f) ≈ 0.31 tau log(e/(31 eps)) an 'empirical scaling law', and Equation (185) is only a finite-discretization lower bound. Moreover, this QSVT reduction applies only to Hermitian A = L >= 0, and the constant-factor gap to the LCHS upper bound is read off from the fit, not derived. The paper should clearly label the factor-3 statement as a numerical conjecture, not a proven lower bound, and should not present it in the abstract as an established infeasibility result.
  3. [Theorem 5; §1.1.7] Theorem 5 is a lower bound for state preparation, with an overlap factor 1/||e^{-tL}|phi>|| and a tilde that hides K/log N. It does not provide a matching lower bound for the block-encoding query complexity O(t log(1/eps)) of e^{-At} for general non-Hermitian A. The text in §1.1.7 appropriately limits the optimality claim to the task of block-encoding e^{-At}, but even that task lacks a proven lower bound: the only candidates are the Hermitian QSVT numerics of §7 and the weak additive lower bounds mentioned in §1. The paper should therefore separate what is proven (optimal LCHS scaling, optimal state-prep query count up to log factors) from what is conjectured (block-encoding optimality in all parameters).
minor comments (4)
  1. [Theorem 1 statement] The residue condition is written as 'Res(fhat,-i) = i sqrt(2 pi)' in the plain-text version; it should be i/sqrt(2 pi). If the published PDF is correct, this is only a transcription issue, but the current text is ambiguous and should be fixed.
  2. [Figure 1 and §7] The 'QSVT lower bound' in Figure 1 is actually a fit to numerically computed optimal polynomial degrees. The label should say 'empirical QSVT fit' or 'numerically computed lower-bound fit', consistently with the caveats in the text.
  3. [Section 1.1.4] The extension to L with negative eigenvalues via a shift L + l(t)I changes the operator being block-encoded: it simulates e^{-\int l} U0(t), not U0(t), so the success probability and the block-encoding of e^{-At} are not preserved. The paragraph should state this limitation explicitly rather than only noting that the phase factor is easy to compute.
  4. [Abstract] The abstract says 'for general time-dependent operators A, we also prove that a uniform trapezoidal rule ... yields exponential convergence'. The exponential convergence theorem, Theorem 3, is proven under the same L >= 0 assumption as the rest of the LCHS construction; the abstract should not suggest a broader scope than the theorem supports.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: LCHS kernel parameters are fixed by closed-form error bounds; optimality claims rest on independent lower-bound arguments.

full rationale

The paper's central derivation is self-contained. Theorem 1 expresses the LCHS error as two kernel integrals (tail and shifted-strip integrals) via contour integration; the only operator-dependent ingredient is the norm bound ||U(t;z)|| ≤ 1 from Lemma 4, which is exactly the L ⪰ 0 (non-positive logarithmic norm) hypothesis and is not an input fitted to e^{-At}. Corollary 7 and Lemmas 8–9 evaluate these integrals in closed form for the constructed kernel f2, and Theorem 2 solves the resulting inequalities to set γ = (1/c)√(c + log((1+1/(2π))e^c/ε)) and R = 2cγ². Thus the 'predicted' e^{-At} block-encoding accuracy is derived from proven bounds, not obtained by fitting parameters to the target operator. The target enters only through the exact residue identity U(t;−i) = U0(t). The quadrature and circuit theorems add only additive error budgets. The lower bounds are either a self-contained SEARCH reduction (Theorem 5) or an independent QSVT polynomial-degree LP (Section 7); they do not use the LCHS kernel as input. The only caveat is that the abstract's 'optimal in all parameters' is stronger than what the theorems prove (for the anti-Hermitian limit L = 0, LCHS costs O(αt log 1/ε) while Hamiltonian simulation achieves O(αt + log 1/ε), as the paper effectively concedes in Section 7), but this is an overstatement of the scope of the optimality result, not circular reasoning. Self-citations (e.g., LC17b/LC19) are standard external Hamiltonian-simulation subroutines and are not used to justify the kernel construction or error analysis.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The central claim rests on standard complex analysis, known quantum algorithmic subroutines, and the domain assumption L >= 0. The listed free parameters are algorithm design choices, not fitted constants, and no new physical entities are introduced.

free parameters (4)
  • gamma (kernel width) = (1/c) sqrt(c + log((1 + 1/(2 pi)) e^c / eps_lchs)) (Theorem 2)
    Controls the accuracy of f2's approximation to exponential decay; derived from the error bound, not fitted to a target phenomenon.
  • c (kernel shift) = Any positive constant; numerically optimized in Section 6 (e.g., c about 1.7 for eps 1e-2)
    Sets the normalization alpha_fhat approximately e^c; any constant c > 0 gives a valid algorithm, and the constant factor depends on it.
  • y0 (contour strip height) = 2 c gamma^2 (chosen in Lemma 8)
    Part of the error analysis in Theorem 1; chosen to balance the two error integrals and is not a physical parameter.
  • j, y (generalized kernel order) = Numerically optimized in Section 6 for constant-factor improvement
    In the extended kernel family fj,y, j and y are tuned to minimize the query prefactor; the main theorem uses f2 = f2,1, so these are not needed for the asymptotic claim.
assumptions (6)
  • standard math Cauchy's residue theorem and contour integration for matrix-valued functions
    Used in the proof of Theorem 1 to relate the LCHS integral to U0(t).
  • standard math Logarithmic norm bound for time-ordered exponentials (Lemma 4, attributed to Soderlind)
    Provides the norm bound ||Te^{-i integral (zL+H)}|| <= e^{|Im z| integral ||L||}, which is essential for the contour error estimate.
  • standard math Trefethen-Weideman exponentially convergent trapezoidal rule (Lemma 10)
    Restated as a theorem for matrix-valued functions; used to prove exponential convergence of the uniform quadrature in Theorem 3.
  • domain assumption Quantum singular value transformation (GSLW19) and optimal Hamiltonian simulation (LC17b, LC19, BMPW24)
    Used as black-box subroutines for polynomial transformations and for simulating the unitaries U(t;k) in the LCHS sum.
  • domain assumption Block-encoding access model for A (Definition 1) and state preparation oracle
    The query complexity is measured in these oracles; this is the standard model for quantum matrix algorithms.
  • domain assumption L >= 0 (or -A has non-positive logarithmic norm)
    Needed for the norm bound in Lemma 4 and the error bound in Theorem 1; negative-eigenvalue cases require the shift in Section 1.1.4.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Optimal quantum simulation of linear non-unitary dynamics." pith.science (2026). https://pith.science/paper/D2MDFKWN

@misc{pith2026250819238,
  author       = {Pith},
  title        = {Pith review of: Optimal quantum simulation of linear non-unitary dynamics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/D2MDFKWN}},
  note         = {Machine review of arXiv:2508.19238}
}
abstract

We present a quantum algorithm for simulating the time evolution generated by any bounded, time-dependent operator $-A$ with non-positive logarithmic norm, thereby serving as a natural generalization of the Hamiltonian simulation problem. Our method generalizes the recent Linear-Combination-of-Hamiltonian-Simulation (LCHS) framework. In instances where $A$ is time-independent, we provide a block-encoding of the evolution operator $e^{-At}$ with $\mathcal{O}\big(t\log\frac{1}{\epsilon})$ queries to the block-encoding oracle for $A$. We also show how the normalized evolved state can be prepared with $\mathcal{O}(1/\|e^{-At}|{\vec{u}_0}\rangle\|)$ queries to the oracle that prepares the normalized initial state $|{\vec{u}_0}\rangle$. These complexities are optimal in all parameters and improve the error scaling over prior results. Furthermore, we show that any improvement of our approach exceeding a constant factor of approximately 3 is infeasible. For general time-dependent operators $A$, we also prove that a uniform trapezoidal rule on our LCHS construction yields exponential convergence, leading to simplified quantum circuits with improved gate complexity compared to prior nonuniform-quadrature methods.

Figures

Figures reproduced from arXiv: 2508.19238 by the authors.

Figure 1
Figure 1. Query complexity prefactor αfˆ · R versus target error ϵ for the LCHS approximating e −At based on (green) the nearly-optimal kernel function ˆfACL23 [ACL23, PJKW25] and (blue) our family of kernel functions ˆfj,y(k; γ, c). We plot upper bounds (solid lines) based on closed-form expressions that are valid for any ϵ as well as numerically bounded upper bounds for numerically optimized kernel functions with (circles) … view at source ↗
Figure 2
Figure 2. Our entire approximation f2 to exponential decay converges to limγ→∞ f2(k, γ, c) = e −|x+c|+c . [0, t] → C 2 n×2 n satisfy L ⪰ 0, ∥L∥L1 < ∞1 , ∥H∥L1 < ∞. Let the uniform strip S[−y0,0] .= {z : Im[z] ∈ [−y0, 0]} for some y0 > 1. On this strip, let the kernel function ˆf(z) decay uniformly such that lim|z|→∞ ˆf(z) = 0, let (z + i) ˆf(z) be analytic, and let the residue Res( ˆf, −i) = limz→−i(z + i) ˆf(z) = √ i 2π . Th… view at source ↗
Figure 3
Figure 3. (Left) Plot of degree n = 10 polynomial pn that is bounded on [−1, 1] and minimizes ϵ = minpn maxx∈[0,1] |e −τx−1 − pn(x)|. (Right) Plot of degree of the numerically computed pn with respect to error and τ together with a fit consistent with Θ(τ log 1 ϵ ) scaling. approximates exponential decay. In Section 3, we prove our main algorithmic result Theorem 2 that a LCHS based on the kernel function ˆf2 block-encodes U0… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Parameter choices of numerically optimized LCHS from Table [PITH_FULL_IMAGE:figures/full_fig_p033_4.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 5 Pith papers

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

  1. Optimal Quantum Eigenvalue Transformation via Linear Combinations of Hermitian Matrices

    quant-ph 2026-07 conditional novelty 7.0 of 10

    The paper derives the exact angular projection A^m = (2/N)Σ_j e^{imθ_j}T_m(Re(e^{-iθ_j}A)) and a quantum algorithm realizing matrix polynomial transforms with Θ(d) depth and optimal post-selection overhead.

  2. Transmutation based Quantum Simulation for Non-unitary Dynamics

    quant-ph 2026-01 conditional novelty 7.0 of 10

    The Kannai transform turns dissipative quantum simulation for A = L†L into a Gaussian-weighted sum of unitary wave propagators, giving query complexity Õ(√(‖A‖T log(1/ε))) and a κ^{3/2} linear-solver corollary.

  3. Explicit block-encoding for partial differential equation-constrained optimization

    quant-ph 2025-11 conditional novelty 6.0 of 10

    An explicit block-encoding bridges a quantum PDE solver and a quantum optimizer, enabling end-to-end, readout-free quantum PDE-constrained optimization with conditional speedups.

  4. Quantum Simulation of Non-Hermitian Special Functions and Dynamics via Contour-based Matrix Decomposition

    quant-ph 2025-11 unverdicted novelty 6.0 of 10

    CBMD decomposes non-Hermitian operators via contour residues to enable optimal-query quantum simulation of first-order dynamics and special functions such as Bessel and Airy evolutions without requiring diagonalizability.

  5. Circuit-Efficient Randomized Quantum Simulation of Non-Unitary Dynamics with Observable-Driven and Symmetry-Aware Designs

    quant-ph 2025-09 reject novelty 5.0 of 10

    A randomized compilation of LCHS for non-unitary dynamics, with an observable-driven variant and a symmetry-aware sampler, claims reduced ancilla and circuit depth at the cost of more repetitions.

Reference graph

Works this paper leans on

19 extracted references · 4 canonical work pages · cited by 5 Pith papers

  1. [1]

    Childs, and Lin Lin

    [ACL23] Dong An, Andrew M. Childs, and Lin Lin. Quantum algorithm for linear non- unitary dynamics with near-optimal dependence on all parameters. arXiv preprint: 2312.03916,

    arXiv

  2. [7]

    An adversary bound for quantum signal processing

    [Lan25] Lorenzo Laneve. An adversary bound for quantum signal processing. arXiv preprint: 2506.20484,

    work page arXiv

  3. [9]

    FromLinearDifferentialEquationstoUnitaries: AMoment-MatchingDila- tion Framework with Near-Optimal Quantum Algorithms.arXiv preprint: 2507.10285,

    [Li25] XiantaoLi. FromLinearDifferentialEquationstoUnitaries: AMoment-MatchingDila- tion Framework with Near-Optimal Quantum Algorithms.arXiv preprint: 2507.10285,

    arXiv

  4. [10]

    Fast quantum simulation of electronic structure by spectrum amplification.arXiv preprint arXiv:2502.15882,

    [LKB+25] Guang Hao Low, Robbie King, Dominic W Berry, Qiushi Han, A Eugene DePrince III, Alec White, Ryan Babbush, Rolando D Somma, and Nicholas C Rubin. Fast quantum simulation of electronic structure by spectrum amplification.arXiv preprint arXiv:2502.15882,

    arXiv

  5. [11]

    Quantum eigenvalue processing

    [LS24a] Guang Hao Low and Yuan Su. Quantum eigenvalue processing. In 2024 IEEE 65th Annual Symposium on Foundations of Computer Science (FOCS), pages 1051–1062. IEEE,

    work page 2024

  6. [12]

    Quantum linear system algorithm with optimal queries to initial state preparation.arXiv preprint arXiv:2410.18178,

    [LS24b] Guang Hao Low and Yuan Su. Quantum linear system algorithm with optimal queries to initial state preparation.arXiv preprint arXiv:2410.18178,

    arXiv

  7. [14]

    Quantum state preparation without coherent arithmetic

    [MGB22] Sam McArdle, András Gilyén, and Mario Berta. Quantum state preparation without coherent arithmetic. arXiv preprint: 2210.14892,

    arXiv

  8. [15]

    [OOL+25] F. W. J. Olver, A. B. Olde Daalhuis, D. W. Lozier, B. I. Schneider, R. F. Boisvert, C. W. Clark, B. R. Miller, B. V. Saunders, H. S. Cohl, M. A. McClain, and eds. NIST digital library of mathematical functions.https://dlmf.nist.gov/ Release 1.2.4 of 2025-03-15,

    work page 2025

Show all 19 references
  1. [16]

    Johnson, Amara Katabarwa, and Nathan Wiebe

    40 [PJKW25] Matthew Pocrnic, Peter D. Johnson, Amara Katabarwa, and Nathan Wiebe. Constant- factor improvements in quantum algorithms for linear differential equations. arXiv preprint: 2506.20760,

  2. [18]

    [SGS+25] Yudai Suzuki, Marek Gluza, Jeongrak Son, Bi Hong Tiang, Nelly H. Y. Ng, and Zoë Holmes. Grover’s algorithm is an approximation of imaginary-time evolution.arXiv preprint: 2507.15065,

  3. [19]

    Quantum algorithm for linear matrix equations.arXiv preprint: 2508.02822,

    [SLBB25] RolandoD.Somma, GuangHaoLow, DominicW.Berry, andRyanBabbush. Quantum algorithm for linear matrix equations.arXiv preprint: 2508.02822,

    arXiv

  4. [2013]

    Design nearly optimal quantum algorithm for linear differential equations via Lindbladians.arXiv preprint arXiv:2410.19628,

    [SGAZ24] Zhong-Xia Shang, Naixu Guo, Dong An, and Qi Zhao. Design nearly optimal quantum algorithm for linear differential equations via Lindbladians.arXiv preprint arXiv:2410.19628,

  5. [2015]

    Somma, and Nicholas C

    [KLB+25] Robbie King, Guang Hao Low, Ryan Babbush, Rolando D. Somma, and Nicholas C. Rubin. Quantumsimulationwithsum-of-squaresspectralamplification. arXiv preprint: 2505.01528,

    arXiv

  6. [2019]

    Fourier transform-based linear combination of Hamiltonian simulation

    [HA25] Xi Huang and Dong An. Fourier transform-based linear combination of Hamiltonian simulation. arXiv preprint: 2508.19596,

    arXiv

  7. [2020]

    Hamiltonian simulation in the interaction picture

    [LW18] Guang Hao Low and Nathan Wiebe. Hamiltonian simulation in the interaction picture. arXiv: 1805.00675,

    arXiv

  8. [2021]

    [LC17a] Guang Hao Low and Isaac L. Chuang. Hamiltonian simulation by uniform spectral amplification. arXiv:1707.05391,

    arXiv

  9. [2023]

    Childs, Lin Lin, and Lexing Ying

    [ACLY24] Dong An, Andrew M. Childs, Lin Lin, and Lexing Ying. Laplace transform based quantum eigenvalue transformation via linear combination of Hamiltonian simulation. arXiv: 2411.04010,

    arXiv

  10. [2024]

    [Joh15] Steven G. Johnson. Saddle-point integration of C∞ "bump" functions.arXiv preprint: 1508.04376,

    arXiv

  11. [2025]

    On the Schrödingerization method for linear non-unitary dynamics with optimal dependence on matrix queries.arXiv preprint: 2505.00370,

    [JLMY25] Shi Jin, Nana Liu, Chuwen Ma, and Yue Yu. On the Schrödingerization method for linear non-unitary dynamics with optimal dependence on matrix queries.arXiv preprint: 2505.00370,

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.