Efficient and Explicit Block Encoding of Finite Difference Discretizations of the Laplacian

Andreas Sturm; Niclas Schillo

REVIEW 3 major objections 5 minor 1 cited by

Efficient and Explicit Block Encoding of Finite Difference Discretizations of the Laplacian

T0 review · 3 major / 5 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 fully explicit quantum circuit block-encodes the finite-difference Laplacian with optimal sub-normalization when the dimension is a power of two, and its success probability is shown to scale as h^4.

desk verdict A correct, practically useful explicit block encoding of the finite-difference Laplacian with α=1 for D=1,2,4 and clean analytic success-probability scaling; the main construction holds up and the paper deserves refereeing despite a few presentation gaps. read the letter →

arxiv 2509.02429 v1 pith:2AO5TXFU submitted 2025-09-02 quant-ph

Andreas Sturm , Niclas Schillo This is my paper · ORCID

classification quant-ph MSC 65N0681P68

keywords blockencodingfinitedifferenceLaplacianquantumPDEsolverssub-normalizationfactorsuccessprobabilityscalingsingularvaluetransformationT-gatecomplexityperiodicboundaryconditions

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 gives an explicit quantum circuit that block-encodes the standard finite-difference discretization of the Laplacian with periodic boundary conditions, in any number of dimensions. The encoding is exact, not an approximation of the discrete matrix, and in dimensions that are powers of two it achieves sub-normalization factor alpha = 1, the largest possible value, so no amplitude is wasted. For other dimensions the factor is D divided by the next power of two, for example 3/4 in 3D. The authors also derive analytically that the success probability scales as h^4 times a ratio of L2 norms, and they report T-gate counts that grow logarithmically in the number of grid points. This matters because block encoding is the standard input model for quantum linear-algebra solvers of PDEs, and every lost factor in the sub-normalization multiplies the cost of downstream algorithms.

What carries the argument

The mechanism is a controlled-shift gadget: two ancilla qubits prepared in a four-way superposition select whether the grid register moves left, right, or stays, with signs arranged so that the |00> branch accumulates the three-point stencil (1,-2,1)/4. In D dimensions an additional ancilla register, transformed by a uniform Hadamard, creates an equal superposition over the D coordinate directions, and controlled shifts apply the stencil along each direction; after the inverse transform the all-zero branch contains the average of the per-direction stencils, which is exactly the scaled Laplacian divided by D and multiplied by D/D_hat. All elements are explicit Hadamard gates, controlled shift

What would settle it

Build the unitary from Theorem 2 explicitly for D = 3 and N = 4 and compare its top-left 64x64 block with (3/4) times the scaled discrete Laplacian; any mismatch falsifies the encoding claim. To test the success-probability claim, apply the circuit to a non-smooth vector such as a step function and check whether the measured post-selection probability follows the h^4 formula; a different scaling would invalidate Lemma 2 but not the exact encoding.

Watch

Extended reading notes

Core claim

Theorem 2 states that for D > 1, with D_hat the smallest power of two at least D, an explicit circuit on the grid register plus 2 + log(D_hat) ancillas is a block encoding of the scaled discrete Laplacian with alpha_D = D/D_hat. Since the scaled Laplacian already has unit spectral norm, alpha_D = 1 for D a power of two is optimal. The proof computes the action on basis vectors: a Hadamard-induced superposition over the D coordinate directions, controlled shifts that apply the one-dimensional three-point stencil along each coordinate, and an inverse Hadamard transform; the all-zero ancilla component comes out as alpha_D times the scaled Laplacian acting on the basis vector. The same calculati

Load-bearing premise

The analytic success-probability scaling assumes the exact solution is four times differentiable and that discrete grid sums can be replaced by continuous L2 integrals via Riemann approximation; if the solution is less regular, or on coarse grids, the h^4 formula need not hold, although the block-encoding identity itself does.

Editorial extensions

If this is right

For D = 1, 2, 4, 8, ... the block encoding wastes no amplitude because alpha_D = 1, so downstream quantum signal processing or phase estimation algorithms pay no extra cost from sub-normalization.
The analytic success-probability formula gives a quantitative prediction for how many repetitions are needed as the grid is refined: the probability decays quadratically in the grid width h for smooth states.
The T-gate count scales as O(log N_D), so exponentially finer grids in each dimension add only linearly more T gates.
In one dimension the new circuit improves on the previous explicit construction by a factor of 16 in success probability while using fewer ancillas and fewer T gates.
For non-power-of-two dimensions the penalty is explicit, for example alpha_3 = 3/4, allowing a user to know exactly what is lost before deciding on a formulation.

Reading between the lines

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

Because alpha_D is always at least 1/2, the worst-case amplitude penalty of this scheme is only a factor of 2, which suggests the method stays near-optimal for every dimension without further adjustment.
The h^4 success-probability scaling implies the encoding becomes more expensive for high-frequency components of the solution; a preconditioned or multilevel representation that flattens the Laplacian spectrum could turn this scaling from a limitation into a design target.
The same controlled-shift construction carries over to first-order operators, as the appendix sketches, so gradient, divergence, and wave-operator encodings plausibly inherit the same logarithmic resource profile and could be combined into a PDE solver toolkit.
Replacing the Riemann-sum approximation in Lemma 2 with exact discrete norms should yield a discrete, regularity-free success-probability bound, removing the C^4 smoothness condition from the scaling claim.

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.

Desk Editor's Note

read the letter

First thing to know: this is a correct and practically useful paper. The authors construct an explicit block encoding of the periodic finite-difference Laplacian that achieves α=1 whenever D is a power of two—a real improvement over the earlier explicit circuit from Camps et al., which encodes only −L/4 and costs an extra ancilla and a factor 16 in success probability. The 1D circuit is elementary and the D-dimensional extension via the k-register is natural. I traced the basis-state computation and it holds; the stress-test note about the missing H gates in the first arrow of Theorem 2's proof is right, but the circuit diagram is unambiguous and that's a presentation shortcut, not a flaw.

What's genuinely new: the two-ancilla α=1 1D encoding, the D/ceil_pow2(D) sub-normalization, and the analytic h^4 success-probability scaling with explicit constants. The authors state the regularity assumption (C^4) and the Riemann-sum approximation, so the scaling claim is honestly conditioned. The numerical experiments confirm their own formulas and the resource estimates show the expected O(log N) T-count. It's also nice that they sketch first-order operators in the appendix.

Soft spots, in proportion: (i) the 4x4 block form of U in Theorem 1 is asserted rather than derived—derivable, but the reader has to do the work. (ii) The comparison to [12] (Kharazi et al.) is qualitative; no resource counts or sub-normalization comparison in arbitrary D. (iii) The qualtran implementation isn't released, so the resource estimates aren't independently reproducible. (iv) Lemma 2's success-probability scaling breaks down on coarse grids or for low-regularity solutions, but that doesn't affect the exact block-encoding result. None of these undermine the central theorem.

Who should read it: anyone building quantum linear-algebra solvers for PDEs, or needing a concrete block-encoding primitive. It deserves a serious referee; the obvious asks are to release the qualtran code and add a quantitative comparison with [12]. I'd accept it for peer review.

Referee Report

3 major / 5 minor

Summary. The paper constructs explicit block-encoding circuits for the normalized finite-difference Laplacian \tilde L_{D,h} on a D-dimensional periodic grid. In one dimension (Theorem 1) a two-ancilla circuit is shown to encode \tilde L_{1,h} with sub-normalization α=1. For arbitrary D (Theorem 2) an additional \hat d = ceil(log_2 D)-qubit register is used to average over the D coordinate directions, yielding a block encoding with α_D = D/\hat D and m = 2+\hat d ancilla qubits. The authors also derive asymptotic success-probability scalings under a C^4 regularity assumption (Lemma 2), compare their construction with the earlier Camps--Lin--Van Beeumen--Yang encoding, and report Qualtran-based T-gate counts showing logarithmic scaling in N_D. An appendix adapts the construction to first-order differential operators.

Significance. If the construction is fully correct, the paper gives a practical advance: compile-ready, exact block encodings with optimal sub-normalization in the power-of-two case, explicit circuits rather than black-box oracles, and analytic scalings for success probability in terms of h, D, and solution regularity. The resource estimates (O(log N_D) T gates for fixed D) and the factor-16 improvement over the previous one-dimensional encoding are tangible. The paper's explicit basis-state proofs and self-contained numerical verification are strengths. The main reservation concerns the rigor of the proof of the D-dimensional theorem, which is currently too abbreviated in a load-bearing place.

major comments (3)

[§5, Theorem 2, Eq. (18a) and proof] The proof of Theorem 2 as printed is not a valid derivation. The first arrow writes |0⟩|0⟩|j⟩ → (1/√D̂) Σ_k |k⟩|0⟩|j⟩, omitting the H gates on the |ℓ⟩ register that are present in circuit (18a). More seriously, the later line writes the k ≥ D contribution as |k⟩|3⟩|j⟩, but the circuit's part 2 does not map |0⟩_ℓ to |3⟩_ℓ when k ≥ D; the cancellation that prevents k ≥ D terms from entering the (|0⟩,|0⟩) block is not shown. Since Theorem 2 is the central result, please supply the complete basis-state calculation, including the explicit action of part 3 on the four |ℓ⟩ states and the fate of the k ≥ D terms.
[§4, Theorem 1, proof after Eq. (15)] The proof computes only the first column of blocks and then asserts the full 4×4 block form (including the D_{1,h} and Q_{1,h} blocks) without derivation. This is not needed for the block-encoding property itself, but the coefficients of the top-left block depend on the exact unitary V implemented by part 3 of circuit (14a). Because the circuit diagram is ambiguous and the proof does not give V, the α=1 claim relies on an unstated calculation. Please add the ancilla unitary in a table or matrix form so that Eq. (15) can be checked directly.
[Abstract and §6.3] The claimed T-gate complexity 'O(log N_D)' is not established for arbitrary D. The circuit contains, for each spatial dimension, S± gates that are controlled not only by the two |ℓ⟩ qubits but also by a multi-qubit condition on the |k⟩ register (k = d). The paper does not count the cost of these multi-controlled shift operations as a function of D; Figure 10 only fits constants C_D log N for D = 1,2,3. Please either prove the T-count including the D-dependent control overhead or state the claim as 'for fixed D'.

minor comments (5)

[§6.2, Eq. (23)] The eigenvalue identity for v_D(x) = ∏_{d=0}^{D-1} sin(2π x^{(d)}) is incorrect as written: L_D v_D = -D(2π)^2 v_D, not (-1)^D(2π)^2 v_D. The constant C_D in Eq. (24) is consistent with the correct identity, so this is a typo, but it should be fixed.
[§4, Eq. (14c)] In the one-dimensional success-probability formula, the numerator uses ||L_D v||²_{L2(Ω_1)}, but the operator should be L_1 v (or L_D with D=1, which is the same, but the notation invites confusion).
[§3, Lemma 2] The statement 'Let v : Ω_D → Ω_D be a four times differentiable function' has the wrong codomain; it should be a scalar-valued function, e.g. v : Ω_D → ℝ. Also, the Riemann-sum approximation in Eq. (12) should mention that the leading h^4 behavior requires C(D,v) ≠ 0; when L_D v vanishes identically, the h^2 scaling is not meaningful.
[§2.2, Eq. (4)] The notation in Eq. (4a) is slightly nonstandard: U has the block α\tilde A in the top-left, while the text says 'block encoding' of A. This is fine once \tilde A is defined, but it would help to state explicitly that (⟨0|⊗I)U(|0⟩⊗I) = αA/||A||₂.
[§6.1, Eq. (21)] The error expression emax(v) uses both λ_{1,max} and α; for the reader it should be stated that α=1 for BE1 and α=-1/4 for BE2, so the 1/α factor compensates the scaling of the encoded matrix. This is clear from context but worth spelling out.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the block encoding is derived from explicit basis-state computation and the success-probability scalings are analytic consequences of stated regularity assumptions.

full rationale

The central claims are (i) Theorem 1: circuit (14a) is a block encoding of \tilde L_{1,h} with α=1, and (ii) Theorem 2: circuit (18a) is a block encoding of \tilde L_{D,h} with α_D=D/\hat D. Both are proved by direct basis-state computation: Theorem 1 computes U|0⟩|j⟩ and identifies the top-left block as \tilde L_{1,h}|j⟩, and Theorem 2 uses the same one-dimensional calculation per spatial dimension together with identity (9). No fitted parameter is renamed as a prediction: α_D is defined from D and \hat D, and the block-encoding relation (4) is checked explicitly in (18c). The success-probability scalings (14c) and (18d) follow analytically from Lemma 2, whose Taylor-expansion and Riemann-sum approximations are stated rather than hidden; the constants C(D,v) are derived from ||L_D v||_{L2}/||v||_{L2}, not extracted from the numerical runs. The numerical experiments then verify those analytically derived scalings, which is a consistency check, not a circular prediction. The comparison with [11] uses that paper's explicit circuit and angles; no load-bearing self-citation occurs, and the authors cite no prior work of their own. The only in-scope anomaly is a proof-writing slip in Theorem 2's first arrow, where the Hadamard gates on the |ℓ⟩ register are not written in the displayed transformation; restoring them as the circuit diagram requires yields exactly the claimed result. This is a notational/proof gap, not a circular reduction. Overall the derivation is self-contained against explicit circuit definitions and analytic estimates.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No free parameters are introduced in the theorems; alpha_D and the h^4 constants are derived rather than fit. The log-slope constants in the resource plots are descriptive fits to measured T-gate counts and do not enter the central claim. The assumptions listed are the domain restrictions (power-of-two grid, periodic boundaries, C^4 solutions for the analytic success-probability scaling) and standard background results from the block-encoding and adder literature.

assumptions (5)

domain assumption Grid size N is a power of two, N = 2^n
Used throughout to represent grid indices in binary and to implement S+ and S- with X and multi-controlled X; all theorems are stated for this domain.
domain assumption Periodic boundary conditions on the D-dimensional unit hypercube
Makes the discrete Laplacian circulant and allows the shift-based circuit; the paper explicitly leaves other boundary conditions to future work.
domain assumption The exact solution v is four times differentiable and periodic for the success-probability scaling
Lemma 2 uses Taylor's theorem with O(h^2) remainder and a Riemann-sum approximation; without this regularity the h^4 success-probability scaling does not follow.
standard math Shift operators have efficient binary adders with O(log N) T gates
The resource estimate relies on cited carry-lookahead adder and qualtran AddK results [20,21]; the block encoding itself only needs any implementation of S+ and S-.
standard math Block-encoding post-selection framework is valid
The paper uses the standard definition and measurement argument from Gilyen et al. [7], not re-derived.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Efficient and Explicit Block Encoding of Finite Difference Discretizations of the Laplacian." pith.science (2026). https://pith.science/paper/2AO5TXFU

@misc{pith2026250902429,
  author       = {Pith},
  title        = {Pith review of: Efficient and Explicit Block Encoding of Finite Difference Discretizations of the Laplacian},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2AO5TXFU}},
  note         = {Machine review of arXiv:2509.02429}
}

read the original abstract

The data input model is a fundamental component of every quantum algorithm, as its efficiency is crucial for achieving potential speed-ups over classical methods. For quantum linear algebra tasks that utilize quantum eigenvalue or singular value transformations, block encoding is the established technique for accessing matrix data. A key application of this is solving partial differential equations, where the Laplacian operator and its finite difference discretization serve as foundational examples. In this paper, we present an efficient and explicit block encoding method that enhances existing approaches in key aspects. We detail the construction of the quantum algorithm and illustrate how it leverages the unique structure of finite difference discretizations. Furthermore, we analytically derive the scaling of the sub-normalization factor and of the success probability of the block encoding with respect to the problem dimension, the grid width of the finite difference grid and the regularity of the exact solution, and we give resource estimates.

Figures

Figures reproduced from arXiv: 2509.02429 by the authors.

**Figure 1.** Examples of grids ΩD,h for the domain ΩD = [0, 1]D for dimensions D = 1 (a), D = 2 (b), and D = 3 (c). The solid blue points are the grid points xk ∈ ΩD,h. The grey circled points do not belong to the grid as the value of a function v on those points is determined by the periodic boundary condition (5b) [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗

**Figure 2.** Example of the ordering of the grid points of the vectorization of [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 3.** Heat map plot of the entries of the matrix [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗

Figures from the paper (8 more)

**Figure 4.** Figure 4: Heat map plot of the entries of the matrix [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]

**Figure 5.** Figure 5: Number of dimensions D versus sub-normalization factor αD. In particular, we have αD = 1 for D = 1 and 2 and αD = 3/4 for D = 3. transformations: |0⟩ |0⟩ |j⟩ →√ 1 Db DbX−1 k=0 |k⟩ |0⟩ |j⟩ →√ 1 Db |0⟩ |0⟩ |j (D−1)⟩. . . |j (1)⟩Le 1,h|j (0)⟩ + √ 1 Db |1⟩ |0⟩ |j (D−1)⟩. .…

**Figure 6.** Figure 6: Heat map plot of the entries of the matrix [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]

**Figure 7.** Figure 7: Heat map plot of the entries of the matrix [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]

**Figure 8.** Figure 8: Block encoding of the one dimensional, discretized Laplacian [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]

**Figure 9.** Figure 9: Number of grid points ND = N D = 2nD versus success probability (4d) of encoding the D dimensional, discretized Laplacian LD,h with the block encoding (18a) for dimensions D = 1, 2, 3, and 4 (blue, orange, green, and red markers). The functions vD are defined in (23) a…

**Figure 10.** Figure 10: Number of T gates versus grid width h (a) and versus number grid points ND (b) for block encoding LD,h with the block encoding (18). Note that we have ND = N D = 1/hD. The constants are C1 = 9, C2 = 17, C3 = 25, and C = 11. 2 0 2 5 2 10 2 15 2 20 2 25 2 30 2 35 101 10…

**Figure 11.** Figure 11: Number of T gates versus number of grid points N1 = N = 1/h for encoding L1,h with the block encoding (14) (blue markers) and (17) (red markers). L1,h. Besides the higher success probability of the block encoding (14), that we discussed in Section 6.1, we also observe…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

Moment-Structured Block Encodings of Periodic Finite-Difference Operators
quant-ph 2026-07 accept novelty 6.0 of 10

Moment order of a periodic finite-difference stencil simultaneously fixes the continuum operator, Fourier-symbol vanishing, and a closed-form optimality certificate for its shift-LCU block encoding.

Reference graph

Works this paper leans on

25 extracted references · 23 canonical work pages · cited by 1 Pith paper

[12]

Ex- plicit block encodings of boundary value problems for many-body elliptic operators

T. Kharazi, A. M. Alkadri, J.-P. Liu, K. K. Mandadapu, and K. B. Whaley. “Ex- plicit block encodings of boundary value problems for many-body elliptic operators”. Quantum 9, 1764 (2025)

work page 2025
[1]

Partial differential equations

L. C. Evans. “Partial differential equations”. Volume 19 of Graduate studies in math- ematics. American Mathematical Society. Providence, Rhode Island (2022). Second edition

work page 2022
[2]

Numerical solution of partial differential equations: Finite difference methods

G. D. Smith. “Numerical solution of partial differential equations: Finite difference methods”. Oxfordappliedmathematicsandcomputingscienceseries.ClarendonPress. Oxford (1993). 3. ed., repr. (with corr.) edition

work page 1993
[3]

Numerical partial differential equations: Finite difference methods

J. W. Thomas. “Numerical partial differential equations: Finite difference methods”. Volume 22 of Springer eBook Collection Mathematics and Statistics. Springer. New York, NY (1995). 18

work page 1995
[4]

Solving ordinary differential equations II

E. Hairer and G. Wanner. “Solving ordinary differential equations II”. Volume 14. Springer Berlin Heidelberg. Berlin, Heidelberg (1991)

work page 1991
[5]

Krylov subspace methods for linear systems

T. Sogabe. “Krylov subspace methods for linear systems”. Volume 60. Springer Nature Singapore. Singapore (2022)

work page 2022
[6]

Hamiltonian simulation by qubitization

G. H. Low and I. L. Chuang. “Hamiltonian simulation by qubitization”. Quantum3, 163 (2019)

work page 2019
[7]

Quantum singular value transformation and beyond: exponential improvements for quantum matrix arithmetics

A. Gilyén, Y. Su, G. H. Low, and N. Wiebe. “Quantum singular value transformation and beyond: exponential improvements for quantum matrix arithmetics”. In Proceed- ings of the 51st Annual ACM SIGACT Symposium on Theory of Computing. Pages 193–204. ACM Digital Library. Association for Computing Machinery (2019)

work page 2019

Show all 25 references

[8]

Efficient quantum algorithms for simulating sparse Hamiltonians

D. W. Berry, G. Ahokas, R. Cleve, and B. C. Sanders. “Efficient quantum algorithms for simulating sparse Hamiltonians”. Communications in Mathematical Physics270, 359–371 (2007)

2007
[9]

Hamiltonian simulation with nearly opti- mal dependence on all parameters

D. W. Berry, A. M. Childs, and R. Kothari. “Hamiltonian simulation with nearly opti- mal dependence on all parameters”. In 2015 IEEE 56th Annual Symposium on Foun- dations of Computer Science (FOCS 2015). Pages 792–809. Piscataway, NJ (2015). IEEE

2015
[10]

Quantum algorithm for systems of linear equations with exponentially improved dependence on precision

A. M. Childs, R. Kothari, and R. D. Somma. “Quantum algorithm for systems of linear equations with exponentially improved dependence on precision”. SIAM Journal on Computing 46, 1920–1950 (2017)

1920
[11]

Explicit quantum circuits for block encodings of certain sparse matrices

D. Camps, L. Lin, R. Van Beeumen, and C. Yang. “Explicit quantum circuits for block encodings of certain sparse matrices”. SIAM Journal on Matrix Analysis and Applications 45, 801–827 (2024)

2024
[13]

Optimal Hamiltonian simulation by quantum signal processing

G. H. Low and I. L. Chuang. “Optimal Hamiltonian simulation by quantum signal processing”. Physical review letters118, 010501 (2017)

2017
[14]

Grand unification of quantum algorithms

J. M. Martyn, Z. M. Rossi, A. K. Tan, and I. L. Chuang. “Grand unification of quantum algorithms”. PRX Quantum2, 040203 (2021)

2021
[15]

Quantum computation and quantum information

M. A. Nielsen and I. L. Chuang. “Quantum computation and quantum information”. Cambridge University Press. (2012)

2012
[16]

Quantum computing: A gentle introduction

E. Rieffel and W. Polak. “Quantum computing: A gentle introduction”. Scientific and engineering computation. The MIT Press. Cambridge, Massachusetts and London, England (2014). First MIT press paperback edition

2014
[17]

Lecture notes on quantum algorithms for scientific computation

L. Lin. “Lecture notes on quantum algorithms for scientific computation” (2022). arXiv:2201.0830

2022
[18]

Quantum error correction for quantum memories

B. M. Terhal. “Quantum error correction for quantum memories”. Reviews of Modern Physics 87, 307–346 (2015)

2015
[19]

Expressing and analyzing quantum algorithms with qual- tran

M. P. Harrigan, T. Khattar, C. Yuan, A. Peduri, N. Yosri, F. D. Malone, R. Bab- bush, and N. C. Rubin. “Expressing and analyzing quantum algorithms with qual- tran” (2024) arXiv:2409.04643

2024 arXiv
[20]

A logarithmic- depth quantum carry-lookahead adder

T. G. Draper, S. A. Kutin, E. M. Rains, and K. M. Svore. “A logarithmic- depth quantum carry-lookahead adder”. Quantum Info. Comput.6, 351–369 (2006). url: https://dl.acm.org/doi/abs/10.5555/2012086.2012090

2006
[21]

Improved quantum circuits for elliptic curve discrete logarithms

T. Häner, S. Jaques, M. Naehrig, M. Roetteler, and M. Soeken. “Improved quantum circuits for elliptic curve discrete logarithms”. In Jintai Ding and Jean-Pierre Tillich, editors, Post-Quantum Cryptography. Volume 12100 of Lecture Notes in Computer Science, pages 425–444. Sprin...

2020
[22]

Block-encoding structured matrices for data input in quantum computing

C. Sünderhauf, E. Campbell, and J. Camps. “Block-encoding structured matrices for data input in quantum computing”. Quantum8, 1226 (2024)

2024
[23]

FABLE: Fast approximate quantum circuits for block-encodings

D. Camps and R. Van Beeumen. “FABLE: Fast approximate quantum circuits for block-encodings”. In 2022 IEEE International Conference on Quantum Computing and Engineering (QCE). Page 104–113. IEEE (2022)

2022
[24]

Realization of quantum signal processing on a noisy quantum computer

Y. Kikuchi, C. McKeever, L. Coopmans, M. Lubasch, and M. Benedetti. “Realization of quantum signal processing on a noisy quantum computer”. npj Quantum Informa- tion 9, 93 (2023)

2023
[25]

Resource-efficient variational block-encoding

L. Rullkötter, S. Weber, V. M. Katukuri, C. Tutschku, and B. C. Mummaneni. “Resource-efficient variational block-encoding” (2025). arXiv:2507.17658. 20

2025

Pith tools

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

[1] [12]

Ex- plicit block encodings of boundary value problems for many-body elliptic operators

T. Kharazi, A. M. Alkadri, J.-P. Liu, K. K. Mandadapu, and K. B. Whaley. “Ex- plicit block encodings of boundary value problems for many-body elliptic operators”. Quantum 9, 1764 (2025)

work page 2025

[2] [1]

Partial differential equations

L. C. Evans. “Partial differential equations”. Volume 19 of Graduate studies in math- ematics. American Mathematical Society. Providence, Rhode Island (2022). Second edition

work page 2022

[3] [2]

Numerical solution of partial differential equations: Finite difference methods

G. D. Smith. “Numerical solution of partial differential equations: Finite difference methods”. Oxfordappliedmathematicsandcomputingscienceseries.ClarendonPress. Oxford (1993). 3. ed., repr. (with corr.) edition

work page 1993

[4] [3]

Numerical partial differential equations: Finite difference methods

J. W. Thomas. “Numerical partial differential equations: Finite difference methods”. Volume 22 of Springer eBook Collection Mathematics and Statistics. Springer. New York, NY (1995). 18

work page 1995

[5] [4]

Solving ordinary differential equations II

E. Hairer and G. Wanner. “Solving ordinary differential equations II”. Volume 14. Springer Berlin Heidelberg. Berlin, Heidelberg (1991)

work page 1991

[6] [5]

Krylov subspace methods for linear systems

T. Sogabe. “Krylov subspace methods for linear systems”. Volume 60. Springer Nature Singapore. Singapore (2022)

work page 2022

[7] [6]

Hamiltonian simulation by qubitization

G. H. Low and I. L. Chuang. “Hamiltonian simulation by qubitization”. Quantum3, 163 (2019)

work page 2019

[8] [7]

Quantum singular value transformation and beyond: exponential improvements for quantum matrix arithmetics

A. Gilyén, Y. Su, G. H. Low, and N. Wiebe. “Quantum singular value transformation and beyond: exponential improvements for quantum matrix arithmetics”. In Proceed- ings of the 51st Annual ACM SIGACT Symposium on Theory of Computing. Pages 193–204. ACM Digital Library. Association for Computing Machinery (2019)

work page 2019