REVIEW 3 major objections 5 minor 30 references
Variational quantum algorithm for the Poisson equation based on the banded Toeplitz systems
T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash
Pith's one-line read A variational quantum Poisson solver decomposes A and A² into just 5 and 6 terms in one dimension, and 4d+1 and 12d² in d dimensions.
desk verdict The 1D decomposition is a real but modest improvement; the d-dimensional claims rest on a false matrix identity and don't hold as written. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The mechanism is the unit circulant shift L (the cyclic permutation matrix with ones just below the diagonal and in the top-right corner), its inverse $L^{{−1}}$, and the standard embedding of a Toeplitz matrix into a 2n×2n circulant matrix C_{2n}. Because L is diagonalized by the quantum Fourier transform, expectation values of Toeplitz pieces become Hadamard-test measurements of diagonal phases; the sparse boundary corrections M1–M5 are evaluated by extended Bell measurements. The paper's decomposition expresses the tridiagonal Poisson matrix as a short combination of such circulant shifts plus sparse corner terms.
What would settle it
Compute the first diagonal entry of 2I − L − $L^{{−1}}$ + diag(1,0,…,0,1) for any n≥2: the L and $L^{{−1}}$ shifts contribute no diagonal terms, while the corner matrix adds 1 to the (1,1) entry, giving 3 instead of A′'s 2, so the equality fails already at n=2 and the derived 12d² count would need a corrected derivation.
Extended reading notes
Core claim
The central claim is that the structural features of the discrete Poisson matrix—constant diagonals away from boundary corrections—let one express the cost function E(θ)=⟨ψ(θ)|A²|ψ(θ)⟩−|⟨b|A|ψ(θ)⟩|² without ever forming A in the Pauli basis. A is written as a banded Toeplitz matrix plus a handful of sparse boundary matrices, and A² likewise; each Toeplitz expectation is evaluated by doubling the matrix into a circulant that the quantum Fourier transform diagonalizes. The paper's advertised figures are 5 terms for ⟨b|Ã|ψ⟩ and 6 for ⟨ψ|ò|ψ⟩ in 1D with unified boundary conditions, and 4d+1 and 12d² in d dimensions with Dirichlet boundary conditions.
Load-bearing premise
The d-dimensional term counts depend on the identity A′ = 2I − L − $L^{{−1}}$ + M5 with M5 = diag(1,0,…,0,1); if that identity is false, the 4d+1 and 12d² counts collapse.
Editorial extensions
If this is right
- For the 1D Poisson equation with unified boundary conditions, ⟨b|Ã|ψ⟩ is evaluated with 5 terms and ⟨ψ|ò|ψ⟩ with 6, so roughly half the quantum circuits of the 7/15-item baseline.
- In d dimensions the counts are 4d+1 and 12d², compared with 4d+1 and (4d+1)²−(4d+1) in the baseline, a saving that grows quadratically with d.
- The counts do not depend on the grid size n, so refining the discretization adds no new decomposition terms, only wider QFT circuits.
- The same circuit templates extend to Hermitian sparse coefficient matrices whose diagonals are constant (a_{i,i+c}=a_c), including K-banded Toeplitz systems with K∈O(polylog n), giving variational linear-solver and matrix-vector-multiplication routines.
- Numerical simulation for the 1D Dirichlet case with n=8 reaches a fidelity above 0.99, indicating the optimized ansatz state converges to the exact solution.
Reading between the lines
- Beyond the paper: the term count is a proxy for measurement cost only if each term needs comparable shots; the practical saving depends on how the measurement budget is shared among the terms, which the paper does not analyze.
- The Toeplitz/circulant embedding used here is generic, so the same 5/6 decomposition should transfer to other constant-diagonal discretizations (e.g., convection-diffusion or heat equations), with only the boundary correction matrices changed.
- The d-dimensional count 12d² rests on the algebraic identity A′=2I−L−L^{−1}+M5 in Section 3.2; a reader who verifies that identity directly will find the constant term needs checking before the multidimensional circuit count can be taken as settled.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proposes a variational quantum algorithm for solving discretized Poisson equations by decomposing the coefficient matrix A and its square A^2 into linear combinations of banded Toeplitz matrices and sparse correction matrices. The authors claim five and six decomposition terms for the one-dimensional Poisson equation with different boundary conditions, and 4d+1 and 12d^2 terms for the d-dimensional Poisson equation with Dirichlet boundary conditions, which are claimed to improve on the decomposition counts in Phys. Rev. A 108, 032418 (2023). They also construct Hadamard-test and extended-Bell-measurement circuits for estimating the cost function, provide a numerical simulation for the one-dimensional Dirichlet case, and extend the approach to K-banded Toeplitz matrices.
Significance. If the decomposition counts were correct, the paper would offer a practically useful reduction in the number of quantum circuits needed to evaluate the variational cost function for Poisson-type problems. The explicit circuit constructions and the numerical demonstration for the one-dimensional Dirichlet case are useful contributions. However, the d-dimensional results rest on an algebraic identity that is false as written, and the one-dimensional unified-boundary term count appears inconsistent with the displayed decomposition. Since the claimed advantage over Ref. [24] is built directly on these decompositions, the central quantitative claims are not established by the manuscript.
major comments (3)
- [Section 3.2, decomposition A' = 2I − L − L^{-1} + M5] This identity is false for the circulant shift L defined in Section 3.1. Direct computation gives (2I − L − L^{-1})_{1,n} = (2I − L − L^{-1})_{n,1} = −1 and diagonal entries 2. Adding M5 = diag(1,0,…,0,1) leaves the corner off-diagonal entries at −1 and changes the first and last diagonal entries to 3, whereas A' has 0 at the corners and 2 on the diagonal. The correct correction is the anti-diagonal matrix E = |e_1⟩⟨e_n| + |e_n⟩⟨e_1|, not a diagonal matrix. Because M5 is the object that is spread over the d tensor factors in A^{(d)}, the subsequent decomposition of M5 into (1/2)\tilde X − (1/2)\tilde X\tilde Z, and every term count derived from it, including 4d+1 and 12d^2, is unsupported as written.
- [Section 3.2, decomposition of A^{(d)} and (A^{(d)})^2] Even if the correction matrix were replaced by the correct anti-diagonal E, the decomposition M5 = (1/2)\tilde X − (1/2)\tilde X\tilde Z is specific to the diagonal M5 and does not apply to E. The operators appearing in the crossed terms of A^{(d)} and (A^{(d)})^2 would change, so the sets M1 and M2 and the resulting counts of 4d+1 and 12d^2 would need to be re-derived. The claimed d-dimensional advantage over Ref. [24] therefore cannot be taken from the current manuscript.
- [Section 3.1 and end of Section 3.1, count of six terms for \tilde A^2] The statement that the number of decomposition terms of ⟨ψ|\tilde A^2|ψ⟩ is 6 for different boundary conditions is inconsistent with the displayed formula \tilde A^2 = T_n^2 − (4c+1−c^2)M_2 − (4d+1−d^2)M_3 + cM_4 + dM_5. In the Dirichlet limit c=d=0 this reduces to T_n^2 − M_1 and the count 6 = 5+1 is plausible. For nonzero c,d, however, there are four sparse matrices in addition to T_n^2; if T_n^2 is expanded in its five displayed components, the total is nine, and if T_n^2 is counted as a single term the total is five. Neither gives six. The claimed count therefore does not cover the 'different boundary conditions' case as stated.
minor comments (5)
- [Throughout] The word 'Toeplitz' is consistently misspelled as 'Teoplitz', including in the title, abstract, and keywords.
- [Abstract and Section 4] The complexity expression 'O(ploy log n)' should read 'O(poly log n)'.
- [Section 3.3 and Figure 5 caption] The numerical simulation validates only the one-dimensional Dirichlet case; no numerical evidence is provided for the unified-boundary or d-dimensional claims.
- [Section 3.2, Figure 3 caption] The caption contains the typo 'Hadmard test'; it should be 'Hadamard test'.
- [General] Equations are not numbered, which makes it difficult to refer to specific claims; a numbered display for the key decomposition identities would substantially improve the presentation.
Circularity Check
No circularity: the term counts follow from explicit linear-combination decompositions of A and A^2, with Ref. [24] used only as an external comparison; the Sec. 3.2 premise appears algebraically wrong, but that is a correctness issue, not circularity.
full rationale
The derivation chain is self-contained. Section 3.1 decomposes the one-dimensional matrix as A' = T_n^1 and A'^2 = T_n^2 - M1, with T_n^k evaluated via circulant embedding (Eqs. (1)-(2)) and M1 handled by explicit state constructions; the counts '5' and '6' are then obtained by counting the displayed expectation terms, not by assuming the conclusion. The unified-boundary case similarly decomposes A~ and A~^2 as T_n^1 - cM2 - dM3 and T_n^2 - (4c+1-c^2)M2 - (4d+1-d^2)M3 + cM4 + dM5, again by direct matrix algebra. Section 3.2 extends the same expansion to d dimensions and counts 4d+1 and 12d^2 from the sum over tensor factors; the comparison with Ref. [24] is an external benchmark, not an input to the derivation, and there is no parameter fitting or prediction-from-fit. I found no self-citation used as load-bearing evidence: Refs. [21], [24], and [27] are prior independent works. The one substantive defect I observed is mathematical rather than circular: the asserted identity 'A' = 2I - L - L^{-1} + M5' appears false with L the circulant shift defined in Sec. 3.1, because the wrap-around entries of L and L^{-1} are not cancelled by the diagonal M5, which would make the later d-dimensional counts unsupported as written. Correctness errors of this kind do not make a derivation circular, so the circularity score remains 0.
Assumptions & free parameters
assumptions (4)
- domain assumption The finite-difference discretization of the Poisson equation with boundary conditions yields a symmetric coefficient matrix A.
- domain assumption There exists an efficient unitary Ub that prepares the state |b⟩.
- standard math The expectation of a Toeplitz matrix T can be computed by embedding it in a 2n×2n circulant matrix and using the quantum Fourier transform.
- ad hoc to paper The correction matrix M5 in the d-dimensional case can be decomposed as 1/2 X̃ − 1/2 X̃ Ẑ with X̃ and Ẑ unitaries.
Cite this review
Pith. "Pith review of Variational quantum algorithm for the Poisson equation based on the banded Toeplitz systems." pith.science (2026). https://pith.science/paper/7LEV5YGI
@misc{pith2026250414828,
author = {Pith},
title = {Pith review of: Variational quantum algorithm for the Poisson equation based on the banded Toeplitz systems},
year = {2026},
howpublished = {\url{https://pith.science/paper/7LEV5YGI}},
note = {Machine review of arXiv:2504.14828}
}
abstract
For solving the Poisson equation it is usually possible to discretize it into solving the corresponding linear system $Ax=b$.Variational quantum algorithms (VQAs) for the discreted Poisson equation have been studied before. We give a VQA based on the banded Toeplitz systems for solving the Poisson equation with respect to the structural features of matrix $A$. In detail, we decompose the matrix $A$ and $A^2$ into a linear combination of the corresponding banded Toeplitz matrix and sparse matrices with only a few non-zero elements. For the one-dimensional Poisson equation with different boundary conditions and the $d$-dimensional Poisson equation with Dirichlet boundary conditions, the number of decomposition terms is less than the work in [Phys. Rev. A 108, 032418 (2023)]. Based on the decomposition of the matrix, we design quantum circuits that evaluate efficiently the cost function.Additionally, numerical simulation verifies the feasibility of the proposed algorithm. In the end, the VQAs for linear systems of equations and matrix-vector multiplications with $K$-banded Teoplitz matrix $T_n^K$ are given, where $T_n^K\in R^{n\times n}$ and $K\in O({\rm ploy}\log n)$.
Figures
Figures from the paper (2 more)
Reference graph
Works this paper leans on
- [24]
-
[1]
P. W. Shor, Algorithms for quantum computation: Discrete logar ithms and factoring, in Proceedings of the 35th Annual Symposium on Foundations of Comp uter Science (IEEE, Piscataway, NJ, 1994), pp. 124–134
work page 1994
-
[2]
L. K. Grover, Quantum mechanics helps in searching for a needle in a haystack, Phys. Rev. Lett. 79, 325 (1997)
1997
-
[3]
A. W. Harrow, A. Hassidim, and S. Lloyd, Quantum Algorithm for Lin ear Systems of Equations, Phys. Rev. Lett. 103, 150502 (2009)
work page 2009
-
[4]
J. Biamonte, P. Wittek, N. Pancotti, P. Rebentrost, N. Wiebe an d S. Lloyd, Quantum machine learning, Nature 549, 195–202 (2017)
work page 2017
-
[5]
Preskill, Quantum computing in the NISQ era and beyond, Quant um 2, 79 (2018)
J. Preskill, Quantum computing in the NISQ era and beyond, Quant um 2, 79 (2018)
2018
- [6]
- [7]
Show all 30 references
-
[8]
S. Endo, S. C. Benjamin and Y. Li, Practical quantum error mitiga tion for near-future applications, Phys. Rev. X 8, 031027 (2018)
2018
-
[9]
Peruzzo, J
A. Peruzzo, J. McClean, P. Shadbolt, M.-H. Yung, X.-Q. Zhou, P. J. Love, A. Aspuru- Guzik, and J. L. O’Brien, A variational eigenvalue solver on a photonic quantum processor, Nat. Commun. 5, 4213 (2014)
2014
-
[10]
Farhi, J
E. Farhi, J. Goldstone and S. Gutmann, A quantum approximate optimization algorithm, arXiv:1411.4028 (2014)
2014 arXiv
-
[11]
G. K. Batchelor, An Introduction to Fluid Dynamics, (Cambridge University, New York, 2000)
2000
-
[12]
Meyn, Control Techniques for Complex Networks (Cambridg e University, New York, 2007)
S. Meyn, Control Techniques for Complex Networks (Cambridg e University, New York, 2007). 19
2007
-
[13]
S. Meyn, R. L. Tweedie, and P. W. Glynn, Markov Chains and Stoc hastic Stability, (Cam- bridge University, New York, 2009)
2009
-
[14]
Engel and R
E. Engel and R. M. Dreizler, Density Functional Theory: An Adv anced Course (New York: Springer, 2011)
2011
-
[15]
G. E. Forsythe, W. R. Wasow, and W. Nachbar, Finite-differenc e methods for partial differential equations, Phys. Today 14(4), 58 (1961)
1961
-
[16]
Bravo-Prieto, R
C. Bravo-Prieto, R. LaRose, M. Cerezo, Y. Subasi, L. Cincio, a nd P. Coles, Variational quantum linear solver: A hybrid algorithm for linear systems, Bull. Am. Phys. Soc. 65 (2020)
2020
-
[17]
Huang, K
H.-Y. Huang, K. Bharti, and P. Rebentrost, Near-term quant um algorithms for linear systems of equations with regression loss functions, New J. Phys. 23(11), 113021 (2021)
2021
-
[18]
A. G. Rattew, S. Hu, M. Pistoia, R. Chen and S. Wood, A domain-a gnostic, noise-resistant, hardware-efficient evolutionary variational quantum eigensolver, arXiv:1910.09694, 2019
1910 arXiv
-
[19]
X. Xu, J. Sun, S. Endo, Y. Li, S. C. Benjamin, and X. Yuan, Varia tional algorithms for linear algebra, Sci. Bull. 66, 2181 (2021)
2021
-
[20]
Patil, Y
H. Patil, Y. Wang and P. S. Krsti´ c, Variational quantum linear s olver with a dynamic ansatz, Phys. Rev. A 105(1), 012423 (2022)
2022
-
[21]
Liu, Y.-S
H.-L. Liu, Y.-S. Wu, L.-C. Wan, S.-J. Pan, S.-J. Qin, F. Gao and Q.-Y . Wen, Variational quantum algorithm for the Poisson equation, Phys. Rev. A 104(2), 022418 (2021)
2021
-
[22]
Gnanasekaran and A
A. Gnanasekaran and A. Surana, Efficient Variational Quantum Linear Solver for Struc- tured Sparse Matrices, arXiv:2404.16991 (2024)
2024 arXiv
-
[23]
Y. Sato, R. Kondo, S. Koide, H. Takamatsu and N. Imoto, Varia tional quantum algorithm based on the minimum potential energy for solving the Poisson equat ion, Phys. Rev. A 104(5), 052409 (2021)
2021
-
[25]
S. S. Zhou and J. B. Wang, Efficient quantum circuits for dense c irculant and circulant like operators, R. Soc. Open Sci. 4(5), 160906 (2017)
2017
-
[26]
R. M. Gray, Toeplitz and circulant matrices: A review, Found. Tr ends Commun. Inf. Theory 2(3), 155 (2006)
2006
-
[27]
Kondo, Y
R. Kondo, Y. Sato, S. Koide, S. Kajita and H. Takamatsu, Comp utationally efficient quantum expectation with extended bell measurements, Quantum 6, 688 (2022)
2022
-
[28]
Gao and Y.-G
S. Gao and Y.-G. Yang, A quantum algorithm for Toeplitz matrix-v ector multiplication, Chin. Phys. B 32(10), 100309 (2023)
2023
-
[29]
Maslov, Linear depth stabilizer and quantum Fourier transfo rmation circuits with no auxiliary qubits in finite-neighbor quantum architectures, Phys
D. Maslov, Linear depth stabilizer and quantum Fourier transfo rmation circuits with no auxiliary qubits in finite-neighbor quantum architectures, Phys. Re v. A 76(5), 052310 (2007)
2007
-
[30]
S. C. Xue, Y. Liu, Y. Wang, P. Y. Zhu, C. Guo and J. J. Wu, Variat ional quantum process tomography of unitaries, Phys. Rev. A 105(3), 032427 (2022). 20
2022
Reviewed August 16, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.