The reviewed record of science sign in
Pith

arxiv: 2607.06557 · v1 · pith:NTQCW6BJ · submitted 2026-07-07 · quant-ph

Quantum Channel Polynomial Processing

Tianhan Liu , Fedor Simkovic IV , Martin Leib This is my paper

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel 2026-07-08 01:50 UTCglm-5.2pith:NTQCW6BJrecord.jsonopen to challenge →

classification quant-ph PACS 03.67.Ac03.67.Lx
keywords complexityquantumframeworkqueryarbitraryerrorpolynomialsample
0
0 comments X

The pith

Stochastic circuits replace block encodings for polynomial quantum transforms

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

The paper introduces Quantum Channel Polynomial Processing (QCPP), a framework that applies arbitrary polynomial transformations of Hermitian operators to quantum states using probabilistic mixtures of simple unitary channels rather than coherent block encodings. The core mechanism encodes Hamiltonian information into sampling probabilities over controlled Pauli rotations rather than into the state of a large ancilla register, shifting resource demands from deep coherent circuits to repeated shallow-circuit executions. Each sampled circuit instance requires at most 2d controlled Pauli rotations for a degree-d polynomial, with no coherent Hamiltonian selection hardware. The paper proves that the natural Jacobi-Anger polynomial expansions for real- and imaginary-time evolution incur exponentially growing sample complexity (Theorem 1), because most roots of those polynomials lie off the real and imaginary axes. To overcome this, the authors construct factored polynomials where one factor has all roots on the imaginary or positive real axis (yielding unit sample complexity) and the remaining factor is a low-degree Chebyshev approximation. This construction achieves super-algebraic convergence of the approximation error with only polynomially bounded sample complexity (Theorem 2), establishing a tunable tradeoff: one can operate at optimal (logarithmic-in-error) query complexity with exponential sampling overhead, or increase query complexity modestly to obtain polynomial sampling overhead with super-algebraic error decay.

Core claim

The central discovery is that the sample complexity of stochastic polynomial processing is governed by the location of the polynomial's roots in the complex plane: roots on the positive real or imaginary axis contribute nothing to sample overhead, while off-axis roots cause exponential growth. By factoring an interpolating polynomial into a component with all roots on favorable axes and a low-degree Chebyshev remainder, one can keep sample complexity polynomial while achieving super-algebraic convergence. This replaces the coherent ancilla-based block encoding of QSVT with a single ancilla qubit and classical post-processing of measurement outcomes.

What carries the argument

The basic building block is a probabilistic channel that, with probability p_z, applies a pi/4 z-rotation on an ancilla qubit and, with probability (1-p_z), applies a controlled Pauli rotation drawn from the Hamiltonian's Pauli decomposition. In a partial Pauli transfer matrix picture with respect to the ancilla, this channel decomposes into a direct sum of a 1-Z subspace and an X-Y subspace. The X-Y subspace, when viewed through its determinant, is proportional to the desired polynomial factor (A_H - Re[z_i])^2 + (C_H + Im[z_i])^2. Concatenating d such blocks (one per polynomial root) and interleaving X-gates on the ancilla computes the product of determinants, which equals the full degree-

If this is right

  • QCPP could enable near-term quantum simulation of Hamiltonian dynamics on NISQ devices that lack the coherent depth for LCU-based QSVT, since each circuit instance uses only controlled Pauli rotations and a single ancilla.
  • The root-location principle for sample complexity may extend beyond time evolution to other Hamiltonian functions (ground-state preparation, thermal-state preparation, matrix inversion), provided suitable factored polynomial approximations can be constructed.
  • The permutation invariance of the desired channel under reordering of building blocks opens hardware-specific transpilation opportunities, since circuit order can be chosen to minimize gate errors on a given device topology.
  • The tunable sample-query tradeoff allows a single algorithmic framework to span the NISQ-to-fault-tolerant transition: near-term devices use more samples and shallower circuits, while fault-tolerant machines reduce samples at the cost of deeper circuits.

Load-bearing premise

The claim of seamless scaling from NISQ to fault-tolerant hardware rests on the sample complexity Gamma(p) remaining practically manageable for realistic problems. While Theorem 2 proves polynomial scaling, the paper does not provide concrete numerical estimates of Gamma for specific chemistry or simulation instances, nor does it analyze how sampling noise on real hardware interacts with the stochastic channel construction.

What would settle it

If the constant factors in the polynomial sample-complexity bound Gamma(p) turn out to be impractically large for realistic Hamiltonians and simulation timescales, the method would require too many circuit repetitions to be competitive with existing approaches.

Figures

Figures reproduced from arXiv: 2607.06557 by Fedor Simkovic IV, Martin Leib, Tianhan Liu.

Figure 1
Figure 1. Figure 1: FIG. 1. Workflow of the Quantum Channel Polynomial Pro [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
read the original abstract

We introduce a quantum algorithmic framework based on probabilistic mixtures of unitary channels that, similar to the framework of quantum singular value transformations, enables the application of arbitrary polynomials of hermitian operators onto arbitrary initial states. We show that our framework supports a flexible tradeoff between sample- and query complexity ranging from optimal query complexity, meaning logarithmic in the error, and exponentially scaling sample complexity to sub-polynomial query complexity in the error and polynomial sample complexity. Combined with the considerably lower quantum circuit complexity, compared to quantum singular value transformations with a linear combination of unitaries block encoding, we argue that our framework can be seamlessly scaled from NISQ to fault-tolerant quantum computing.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 7 minor

Summary. This manuscript introduces Quantum Channel Polynomial Processing (QCPP), a stochastic framework for implementing polynomial transformations of Hermitian operators on quantum states. The construction replaces coherent block encodings (as used in QSVT) with probabilistic mixtures of unitary channels built from controlled Pauli rotations and ancilla-mediated measurements. The core idea is that each stochastic building block corresponds to a root of the interpolating polynomial, and the determinant structure of the partial Pauli transfer matrix enables assembly of degree-d polynomials. The authors then analyze the sample-query complexity tradeoff: Theorem 1 states that Jacobi-Anger expansions incur exponential sample complexity, while Theorem 2 claims that a factored polynomial construction p = q·h achieves super-algebraic convergence with polynomially bounded sample complexity. The framework is illustrated for real- and imaginary-time evolution.

Significance. The paper addresses a genuine algorithmic gap between block-encoded QSVT (which requires coherent Hamiltonian encoding with substantial ancillary overhead) and stochastic simulation methods (which reduce circuit depth but lack a general polynomial-transformation framework). The QCPP construction is novel in combining polynomial approximation with stochastic channel sampling, and the idea that each building block maps to a polynomial root via the determinant structure is elegant. The sample-query complexity tradeoff, if correct, is a meaningful contribution. However, the central complexity claims rest on Theorems 1 and 2, whose proofs are deferred to supplementary material not included with the manuscript. Without these proofs, the significance of the contribution cannot be fully assessed.

major comments (3)
  1. Theorems 1 and 2 are the load-bearing results for the paper's central claims about exponential sample cost (Jacobi-Anger) and super-algebraic convergence with polynomial sample complexity (factored construction). Both proofs are deferred to 'supplementary material' that is not included in the manuscript. The main text provides no derivation sketch for either theorem. For Theorem 2 specifically, the key claims — that h = f/q can be approximated by a Chebyshev polynomial of degree d_h ~ log(d) (Section III.B), and that the error bound C(t²/d)^{d−log d} (Eq. 18) follows — are asserted without any argument. The derivation of the query complexity formula (Eq. 20) from the error bound is also not shown. These are not peripheral lemmas; they are the substance of the paper's contribution. At minimum, proof sketches sufficient to verify internal consistency must appear in the main text.
  2. Section III.B, Eqs. (16)–(18): the consistency between the claimed Chebyshev degree d_h ~ log(d) and the error bound C(t²/d)^{d−log d} is not established. A naive Chebyshev approximation of h = f/q on [−1,1] would yield an error rate depending on the smoothness of h and on d_q ≈ d (the degree of q). If d_h needs to be larger than log(d) for the stated bound to hold, then Γ(h) ~ exp(c·d_h) could be super-polynomial in d, and the headline tradeoff would not hold. The manuscript should either provide the argument showing that q captures enough of f's structure to make h sufficiently smooth, or qualify the claim. This is the specific unverified link that determines whether the central practical claim stands.
  3. Abstract and Conclusion: the claim that QCPP can be 'seamlessly scaled from NISQ to fault-tolerant quantum computing' is not supported by the analysis presented. Theorem 2 establishes polynomial sample complexity for the factored construction, but the actual value of Γ(p) depends on the polynomial roots and the constant C(t²/d), for which no concrete numerical estimates are provided. The ancilla measurement post-selection halves the success probability per shot (Section II.C), compounding the sample overhead. Without bounds on Γ for representative problems (e.g., chemistry Hamiltonians with O(N⁴) terms), the NISQ scalability claim is unsupported and should be either justified with estimates or removed.
minor comments (7)
  1. Eq. (2): the notation F_ρ conflates the channel F with its action on ρ. Using F(ρ) or defining F as a superoperator acting on ρ would be clearer.
  2. Section II.B, Eq. (3): the building block E is defined with p_z and θ as free parameters, but it is only later (Eqs. 6–7) that these are tied to specific polynomial roots. A forward reference or brief note that E will be specialized to E(z_i) would help the reader.
  3. Section II.C: the reference to 'Fig. ??' appears to be a broken cross-reference. Also, the workflow figure (Fig. 1) is referenced before its appearance; consider reordering.
  4. Eq. (11): the sample complexity Γ(p) is defined as a product over roots, but the connection to the 'standard argument for unbiased sampling' (the √M ≫ Γ(p) condition) is stated without derivation. A one-line derivation or reference would strengthen this.
  5. Section III.B: the statement 'With this construction it is obvious that the sample complexity is polynomial in the query complexity' is too strong given that the proof is deferred. Consider softening to 'can be shown to be polynomial' with a reference to the supplementary material.
  6. The paper would benefit from a small numerical example (e.g., a table or plot showing Γ vs. d for a specific t and τ) to illustrate the tradeoff concretely, even if detailed numerics are in the supplementary material.
  7. Several recent works on randomized QSVT and stochastic Hamiltonian simulation (Refs. 31–33) are cited in the introduction but not discussed in relation to QCPP's novelty. A brief comparison clarifying what QCPP adds beyond these works would help position the contribution.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for a careful and constructive report. The referee correctly identifies that the proofs of Theorems 1 and 2 are deferred to supplementary material not included with the main text, and that this materially limits verifiability of the central claims. We agree that proof sketches must be added to the main text. We also agree that the NISQ scalability claim requires either concrete estimates or qualification. On the specific technical question about whether h = f/q is sufficiently smooth for d_h ~ log(d) to yield the stated error bound, we provide the argument below and will incorporate it into the revised manuscript.

read point-by-point responses

  1. Referee: Theorems 1 and 2 are the load-bearing results... Both proofs are deferred to supplementary material not included... No derivation sketch for either theorem... For Theorem 2 specifically, the key claims — that h = f/q can be approximated by a Chebyshev polynomial of degree d_h ~ log(d), and that the error bound C(t²/d)^{d−log d} follows — are asserted without any argument. The derivation of the query complexity formula (Eq. 20) from the error bound is also not shown.

    Authors: The referee is correct that the proofs of Theorems 1 and 2 are not present in the main text and that this is a significant gap. We will add proof sketches for both theorems to the revised main text. Here we outline the key steps to demonstrate internal consistency. For Theorem 1, the argument proceeds by analyzing the root structure of the truncated Jacobi-Anger expansion. The sample complexity Γ(p) = |a_d| ∏_i (ℜ[z_i] + √(1 + ℑ[z_i]²)) depends on the real and imaginary parts of each root z_i. Roots lying on the positive real axis contribute a factor of O(1) to Γ; roots on the imaginary axis contribute a factor of O(|z_i|). The proof shows that for the truncated Jacobi-Anger expansion of degree d, the majority of roots are neither on the positive real axis nor on the imaginary axis, and each such root contributes a factor that grows exponentially in |z_i|. Since the roots of the truncated expansion grow with d, the product grows at least as exp(c·d) for constants c_real, c_imag independent of d. An exponentially growing upper bound is established by direct computation of the root locations. For Theorem 2, the argument is as follows. The polynomial q is chosen so that f/q is analytic in a Bernstein ellipse E_ρ with parameter ρ > 1 that depends on d only through the ratio t²/d (or τ²/d). Specifically, the roots of q are placed at z = ±it/d_q (real time) or z = d_q/τ (imaginary time), which cancels the dominant singularity structure of f = exp(−itz) (or exp(−τ(z+1))) near the endpoints of [−1,1]. After this cancellation, the function h = f/q has a Chebyshev coefficient sequence that decays geometrically with rate depending on ρ. The degree-d_h Chebyshev truncation error is then O(ρ^{−d_h}). Setting d_h = ⌈log(d)⌉ and using the fact that ρ depends on d only through t²/d (so revision: no

  2. Referee: Section III.B, Eqs. (16)–(18): the consistency between the claimed Chebyshev degree d_h ~ log(d) and the error bound C(t²/d)^{d−log d} is not established. A naive Chebyshev approximation of h = f/q on [−1,1] would yield an error rate depending on the smoothness of h and on d_q ≈ d. If d_h needs to be larger than log(d) for the stated bound to hold, then Γ(h) ~ exp(c·d_h) could be super-polynomial in d, and the headline tradeoff would not hold.

    Authors: This is the most important technical objection, and we agree that the manuscript does not currently establish the link the referee identifies. We will add the argument to the revised main text. The key point is that q is not an arbitrary degree-d_q polynomial; it is specifically constructed to cancel the endpoint singularities of f. For real-time evolution, f(z) = exp(−itz) is entire, but its Chebyshev coefficients decay slowly (as Bessel functions J_n(t), which decay only for n ≫ t). The polynomial q_real(z) = (1 − itz/d_q)^{d_q} / (1 + t²/d_q²)^{d_q/2} has all roots at z = ±id_q/t. When d_q is chosen proportional to d, the function h = f/q has the property that its Chebyshev coefficients decay geometrically with a rate ρ^{-1} where ρ depends on d_q/t. Concretely, h is analytic in a Bernstein ellipse E_ρ with semi-major axis ρ = 1 + O(t²/d_q). The Chebyshev truncation error at degree d_h is then O(ρ^{−d_h}) = O((1 + t²/d)^{−d_h}). Setting d_h = ⌈log d⌉ gives an error of O((t²/d)^{log d}), which is super-algebraic. The total error ‖f − q·h‖_∞ ≤ ‖q‖_∞ · ‖f/q − h‖_∞, and ‖q‖_∞ is bounded by a constant independent of d (by construction of the normalization in Eqs. 16–17). The combined error is then C(t²/d)^{d − log d} as stated, where the exponent d − log d arises because deg(q) = d_q = d − d_h = d − log d. The sample complexity Γ(h) depends on the roots of h, which are the Chebyshev nodes of the degree-d_h truncation. Since d_h = O(log d), Γ(h) = exp(O(d_h)) = exp(O(log d)) = poly(d). Combined with Γ(q) = 1 (roots on imaginary/positive real axis), the total Γ(p) = Γ(q)·Γ(h) = poly(d). We will add this derivation to Section III.B. revision: no

  3. Referee: Abstract and Conclusion: the claim that QCPP can be 'seamlessly scaled from NISQ to fault-tolerant quantum computing' is not supported by the analysis presented. Theorem 2 establishes polynomial sample complexity for the factored construction, but the actual value of Γ(p) depends on the polynomial roots and the constant C(t²/d), for which no concrete numerical estimates are provided. The ancilla measurement post-selection halves the success probability per shot, compounding the sample overhead. Without bounds on Γ for representative problems (e.g., chemistry Hamiltonians with O(N⁴) terms), the NISQ scalability claim is unsupported and should be either justified with estimates or removed.

    Authors: We agree that the claim as stated is not supported by the analysis in the current manuscript. The word 'seamlessly' overstates what we have shown. Theorem 2 establishes polynomial sample complexity in the query complexity d, but we have not provided concrete numerical estimates of Γ(p) for specific problem instances, nor have we analyzed the compounding effect of the 1/2 success probability per ancilla measurement on the total sample overhead for realistic Hamiltonians. We will revise the abstract and conclusion to qualify the claim. Specifically, we will replace 'seamlessly scaled from NISQ to fault-tolerant quantum computing' with a more measured statement such as: 'The framework supports a tradeoff between circuit depth and sample complexity that may be suitable for near-term and early fault-tolerant devices, though a quantitative resource analysis for specific applications such as chemistry Hamiltonians is left for future work.' We will also add a brief discussion of the 1/2 post-selection factor and its effect on the total sample count. Providing concrete Γ estimates for O(N⁴)-term chemistry Hamiltonians would require a separate numerical study that is beyond the scope of the current theoretical contribution, and we will state this explicitly. revision: no

Circularity Check

0 steps flagged

No significant circularity; core constructions are self-contained, with deferred proofs as the main risk

full rationale

The paper's main derivation chain is largely self-contained. The stochastic building block (Eq. 3), the determinant-based polynomial assembly (Eqs. 8-10), and the sample complexity formula Γ(p) (Eq. 11) are derived from first principles without circular dependencies. The factored construction p = q·h (Section III.B) is a design choice where q is chosen to have Γ(q) = 1 by placing roots on axes (Eqs. 16-17), which is a legitimate construction rather than a circular definition. The key claims in Theorems 1 and 2 (exponential sample cost for Jacobi-Anger, super-algebraic convergence for factored polynomials) are stated with proofs deferred to supplementary material. This deferral is a correctness/completeness concern, not a circularity concern: the claims are not defined in terms of their own conclusions, nor do they reduce to their inputs by construction. The d_h ~ log(d) assertion and the error bound C(t²/d)^{d−log d} (Eq. 18) are substantive mathematical claims that could be true or false independently of the framework's definitions. The only minor self-referential element is that Theorem 1's insight (roots off-axis cause exponential growth) is used to motivate the construction in Section III.B, but this is standard mathematical reasoning (using a negative result to guide a positive construction), not circularity. No fitted parameters are relabeled as predictions, and no self-citation chain is load-bearing for the central results. The derivation is self-contained against external benchmarks (comparison with QSVT, standard Chebyshev approximation theory). Score 2 reflects the minor concern that key proofs are unavailable for verification, but this is a completeness risk rather than a circularity risk.

Axiom & Free-Parameter Ledger

4 free parameters · 4 axioms · 2 invented entities

The axiom ledger reveals a framework built on standard quantum information primitives (Pauli decompositions, transfer matrices) with two key domain assumptions: the convex Pauli decomposition of H and the unbiased sampling argument. The convergence theorems rest on supplementary proofs not available in the main text. No ad hoc entities are introduced; the building blocks are constructed from known operations. The main gap is the unspecified convergence constants C(t²/d).

free parameters (4)
  • θ_i (rotation angles) = arctan(-1/Re[z_i]) per Eq. 6
    Each building block's rotation angle is determined by the real part of the corresponding polynomial root z_i. These are derived from the interpolating polynomial, not fitted to data.
  • p_zi (sampling probabilities) = Re[z_i]/(Re[z_i] + sqrt(1+Im[z_i]^2)) per Eq. 7
    Probability of applying the Z-rotation vs. controlled Pauli rotation, determined by polynomial root. Not a free fitting parameter but a derived quantity.
  • d (polynomial degree / query complexity) = Not specified; tradeoff parameter
    The degree d is the key algorithmic parameter controlling the sample-query tradeoff. Its optimal value depends on the target function, error tolerance, and sample budget (Eq. 20).
  • C(t²/d), C(τ²/d) (convergence constants) = Unspecified
    The constant functions in the super-algebraic convergence bounds (Eqs. 18-19) are not explicitly evaluated, making the practical tightness of the bounds unclear.
axioms (4)
  • domain assumption Hamiltonian decomposability: H = λ Σ p_g g where g are signed Pauli operators and p_g form a convex combination
    §II.A states this 'without loss of generality.' While many Hamiltonians can be decomposed into Pauli operators, the assumption that coefficients are positive and sum to 1 (convex combination) constrains the class of Hamiltonians or requires rescaling.
  • standard math Commutativity of C_H and A_H superoperators with respect to the same H
    §II.A uses [C_H, A_H]=0 to factor the polynomial channel. This is a standard identity for commutator/anti-commutator superoperators.
  • domain assumption Unbiased sampling argument: √M >> Γ(p) suffices for small additive error
    §III uses a 'standard argument for unbiased sampling' to derive sample complexity. The formal conditions under which the sampling is truly unbiased (e.g., independence across shots, no systematic errors) are not discussed.
  • domain assumption Super-algebraic convergence of Chebyshev-based h polynomials
    Theorem 2 states super-algebraic convergence for the factored polynomial construction. The proof is deferred to supplementary material. The result depends on properties of the quotient function f/q being well-approximated by Chebyshev polynomials.
invented entities (2)
  • QCPP building block (stochastic channel E) independent evidence
    purpose: Probabilistic mixture of Z-rotation and controlled Pauli rotations implementing a single polynomial root
    The building block is constructed from standard quantum operations (Pauli rotations, controlled gates) and its action is derived analytically in the transfer matrix picture (Eqs. 4-5). It is falsifiable by implementing the circuit and checking the channel action.
  • Factored polynomial construction p(x)=q(x)h(x) independent evidence
    purpose: Reduce sample complexity by separating roots with Γ=1 (q) from Chebyshev approximation (h)
    The construction makes a falsifiable prediction: specific polynomial forms (Eqs. 16-17) should yield sample complexity Γ(q)=1. This can be verified numerically.

pith-pipeline@v1.1.0-glm · 13783 in / 3441 out tokens · 527175 ms · 2026-07-08T01:50:17.544872+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

35 extracted references · 35 canonical work pages · 7 internal anchors

  1. [1]

    oracle calls

    This means we are ultimately implementing, on the computational register, a probabilistic mixture with equal probabilities of two channels: ρ→ 1 2   1 0 ZA(θ d). . .A(θ 1)ZA(θ d). . .A(θ 1) 1 0 | {z } P ± 1 0 ZB(θ d). . .B(θ 1)ZB(θ d). . .B(θ 1) 1 0 | {z } ∝F   ρ.(10) For ease of illustration we have presented above a version of the expression wit...

    work page

  2. [2]

    T. J. Yoder, G. H. Low, and I. L. Chuang, Fixed-point quantum search with an optimal number of queries, Phys. Rev. Lett.113, 210501 (2014)

    work page 2014

  3. [3]

    L. K. Grover, Fixed-point quantum search, Phys. Rev. Lett.95, 150501 (2005)

    work page 2005

  4. [4]

    P. W. Shor, Polynomial-time algorithms for prime factor- ization and discrete logarithms on a quantum computer, SIAM Journal on Computing26, 1484–1509 (1997)

    work page 1997

  5. [5]

    D. W. Berry, A. M. Childs, and R. Kothari, Hamiltonian simulation with nearly optimal dependence on all param- eters, in2015 IEEE 56th Annual Symposium on Founda- tions of Computer Science(IEEE, 2015) p. 792–809

    work page 2015

  6. [6]

    G. H. Low and I. L. Chuang, Hamiltonian Simulation by Qubitization, Quantum3, 163 (2019)

    work page 2019

  7. [7]

    A. W. Harrow, A. Hassidim, and S. Lloyd, Quantum al- gorithm for linear systems of equations, Physical Review Letters103, 10.1103/physrevlett.103.150502 (2009)

    work page doi:10.1103/physrevlett.103.150502 2009

  8. [8]

    J. M. Martyn, Z. M. Rossi, A. K. Tan, and I. L. Chuang, Grand unification of quantum algorithms, PRX Quan- tum2, 040203 (2021)

    work page 2021

  9. [9]

    L. N. Trefethen,Approximation Theory and Approxima- tion Practice, Extended Edition(SIAM-Society for Indus- trial and Applied Mathematics, Philadelphia, PA, USA, 2019)

    work page 2019

  10. [10]

    L. R. Scott,Numerical Analysis(Princeton University Press, USA, 2011)

    work page 2011

  11. [11]

    Y. Dong, L. Lin, and Y. Tong, Ground-state prepara- tion and energy estimation on early fault-tolerant quan- tum computers via quantum eigenvalue transformation of unitary matrices, PRX Quantum3, 040305 (2022)

    work page 2022

  12. [12]

    C.-F. Chen, M. J. Kastoryano, and A. Gily´ en, An effi- cient and exact noncommutative quantum gibbs sampler (2023), arXiv:2311.09207 [quant-ph]. 8

    work page arXiv 2023

  13. [13]

    Gily´ en, Y

    A. Gily´ en, Y. Su, G. H. Low, and N. Wiebe, Quantum singular value transformation and beyond: exponential improvements for quantum matrix arithmetics, inPro- ceedings of the 51st Annual ACM SIGACT Symposium on Theory of Computing, STOC ’19 (ACM, 2019) p. 193–204

    work page 2019

  14. [14]

    A. M. Childs and N. Wiebe, Hamiltonian simulation us- ing linear combinations of unitary operations, Quantum Information and Computation12, 10.26421/qic12.11-12 (2012)

    work page doi:10.26421/qic12.11-12 2012

  15. [15]

    G. H. Low and N. Wiebe, Hamiltonian simulation in the interaction picture (2019), arXiv:1805.00675 [quant-ph]

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  16. [16]

    Explicit Quantum Circuits for Block Encodings of Certain Sparse Matrices

    D. Camps, L. Lin, R. V. Beeumen, and C. Yang, Explicit quantum circuits for block encodings of certain sparse matrices (2023), arXiv:2203.10236 [quant-ph]

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  17. [17]

    G. H. Low, V. Kliuchnikov, and L. Schaeffer, Trading t gates for dirty qubits in state preparation and unitary synthesis, Quantum8, 1375 (2024)

    work page 2024

  18. [18]

    Kikuchi, C

    Y. Kikuchi, C. Mc Keever, L. Coopmans, M. Lubasch, and M. Benedetti, Realization of quantum signal pro- cessing on a noisy quantum computer, npj Quantum In- formation9, 10.1038/s41534-023-00762-0 (2023)

    work page doi:10.1038/s41534-023-00762-0 2023

  19. [19]

    Nibbi and C

    M. Nibbi and C. B. Mendl, Block encoding of matrix product operators, Physical Review A110, 10.1103/physreva.110.042427 (2024)

    work page doi:10.1103/physreva.110.042427 2024

  20. [20]

    F. D. Chiara, M. Nibbi, Y. Shen, and R. V. Beeumen, Efficient lcu block encodings through dicke states prepa- ration (2025), arXiv:2507.20887 [quant-ph]

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  21. [21]

    W. A. Simon, C. M. Gustin, K. Serafin, A. Ralli, G. R. Goldstein, and P. J. Love, Ladder operator block- encoding (2025), arXiv:2503.11641 [quant-ph]

    work page arXiv 2025

  22. [22]

    D. Liu, W. Du, L. Lin, J. P. Vary, and C. Yang, An effi- cient quantum circuit for block encoding a pairing hamil- tonian (2024), arXiv:2402.11205 [nucl-th]

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  23. [23]

    Babbush, C

    R. Babbush, C. Gidney, D. W. Berry, N. Wiebe, J. Mc- Clean, A. Paler, A. Fowler, and H. Neven, Encoding elec- tronic spectra in quantum circuits with linear t complex- ity, Phys. Rev. X8, 041015 (2018)

    work page 2018

  24. [24]

    J. Lee, D. W. Berry, C. Gidney, W. J. Huggins, J. R. Mc- Clean, N. Wiebe, and R. Babbush, Even more efficient quantum computations of chemistry through tensor hy- percontraction, PRX Quantum2, 030305 (2021)

    work page 2021

  25. [25]

    D. S. Abrams and S. Lloyd, Simulation of many-body fermi systems on a universal quantum computer, Phys. Rev. Lett.79, 2586 (1997)

    work page 1997

  26. [26]

    M. Suzuki, Decomposition formulas of exponential op- erators and Lie exponentials with some applications to quantum mechanics and statistical physics, Journal of Mathematical Physics26, 601 (1985)

    work page 1985

  27. [27]

    A. M. Childs, Y. Su, M. C. Tran, N. Wiebe, and S. Zhu, Theory of trotter error with commutator scaling, Physi- cal Review X11, 10.1103/physrevx.11.011020 (2021)

    work page doi:10.1103/physrevx.11.011020 2021

  28. [28]

    Campbell, Random compiler for fast hamiltonian sim- ulation, Phys

    E. Campbell, Random compiler for fast hamiltonian sim- ulation, Phys. Rev. Lett.123, 070503 (2019)

    work page 2019

  29. [29]

    qSWIFT: High-order randomized compiler for Hamiltonian simulation

    K. Nakaji, M. Bagherimehrab, and A. Aspuru-Guzik, qswift: High-order randomized compiler for hamiltonian simulation, arXiv preprint arXiv:2302.14811 (2023)

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  30. [30]

    qSHIFT: An Adaptive Sampling Protocol for Higher-Order Quantum Simulation

    S. Lee and S. Choi, qshift: An adaptive sampling proto- col for higher-order quantum simulation, arXiv preprint arXiv:2604.26263 (2026)

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  31. [31]

    Chen, H.-Y

    C.-F. Chen, H.-Y. Huang, R. Kueng, and J. A. Tropp, Concentration for random product formulas, PRX Quan- tum2, 040305 (2021)

    work page 2021

  32. [32]

    P. Zeng, J. Sun, L. Jiang, and Q. Zhao, Simple and high- precision hamiltonian simulation by compensating trot- ter error with linear combination of unitary operations, PRX Quantum6, 10.1103/prxquantum.6.010359 (2025)

    work page doi:10.1103/prxquantum.6.010359 2025

  33. [33]

    Randomised composite linear-combination-of-unitaries: its role in quantum simulation and observable estimation

    J. Sun and P. Zeng, Randomised composite linear- combination-of-unitaries: its role in quantum simula- tion and observable estimation (2025), arXiv:2506.15658 [quant-ph]

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  34. [34]

    X. Wang, Y. Zhang, S. Hazra, T. Li, C. Shao, and S. Chakraborty, Randomized quantum singular value transformation (2025), arXiv:2510.06851 [quant-ph]

    work page arXiv 2025

  35. [35]

    G. H. Low, T. J. Yoder, and I. L. Chuang, Methodology of resonant equiangular composite quantum gates, Physical Review X6, 10.1103/physrevx.6.041067 (2016)

    work page doi:10.1103/physrevx.6.041067 2016