Pith. sign in

REVIEW 4 major objections 5 minor 14 references

Quantum Mini-Apps for Engineering Applications: A Case Study

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read The variational Poisson solver studied here fails to recover the correct solution even on a noise-free 4-qubit simulator, and the paper attributes the failure to the Rayleigh-quotient cost function it inherited from prior work.

desk verdict A useful, honest case study whose central negative claim about the Sato cost function is underdetermined by the single 4-qubit experiment it rests on. read the letter →

arxiv 2411.12920 v1 pith:BARMBONF submitted 2024-11-19 quant-ph cs.ET

Horia Mărgărit , Amanda Bowman , Krishnageetha Karuppasamy , Alberto Maldonado-Romo , Vardaan Sahgal , Brian J. McDermott This is my paper · ORCID
classification quant-phcs.ET MSC 81P6865N06
keywords variationalquantumalgorithmPoissonequationNISQhardwarecostfunctionbarrenplateaustensornetworkansatzsoftwarearchitecturecircuitdepth
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

Implementing a variational quantum algorithm (VQA) for the one-dimensional Poisson equation as a case study, the paper reports that the cost function inherited from an earlier method does not steer the ansatz to the true solution. The authors first reduced circuit depth, replaced boundary-condition encodings with sparse Pauli operators and V-chains, adopted tensor-network ansatzes, and switched to gradient-free optimizers; despite all this, 4-qubit runs on a noise-free QASM simulator still produced truncated solutions. They conclude that the Rayleigh-quotient cost function from Eq. 4 may not be viable for NISQ Poisson solvers, and advocate alternative cost functions and hybrid quantum-classical optimization. If this negative result holds, progress on quantum PDE solvers depends on redesigning the objective, not only on quieter hardware or shallower circuits.

What carries the argument

The central object is the variational Rayleigh-quotient cost function of Eq. 4, $E_0 \le E_h(r^*,\theta^*) = \min_{r,\theta} \frac{1}{2} r^2 \langle \psi(\theta)| A | \psi(\theta)\rangle - r \langle f,\psi(\theta)| X\otimes I^{\otimes n} | f,\psi(\theta)\rangle$, which maps the Poisson equation to a minimization over a scaling factor $r$ and ansatz parameters $\theta$. The paper tests this cost function under a modular VQA stack where circuit construction is decoupled from transpilation, with sparse Pauli operators or V-chains for boundary conditions, tensor-network ansatzes (MPS and TTN) to mitigate barren plateaus, and gradient-free optimizers. The machinery that carries the argument is the pairing of this specific cost function with these components: the authors' ablations are meant to isolate what survives once gate counts and noise are reduced, and the finding that truncation persists is what incriminates the cost function.

What would settle it

Re-run the 4-qubit periodic-boundary problem of Figure 4 on a noiseless simulator using Eq. 4 with a classically exact, fully expressive state-vector ansatz and a known closed-form Poisson solution; if the optimizer reaches the exact solution, the cost function is viable and the truncation observed here is an artifact of the ansatz or optimizer, whereas if the global minimum is still wrong, the cost function itself is the culprit.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is negative: after addressing every implementation limitation it identified in the earlier Poisson VQA—high gate depth, inefficient boundary-condition encoding, and barren-plateau-prone ansatzes—the algorithm's solution remained truncated on a noiseless simulator. The paper therefore claims that Eq. 4, the minimum-potential-energy Rayleigh quotient of the earlier method, is not a reliable cost function for ensuring convergence to the correct Poisson solution on near-term hardware; the observed instability of the cost landscape corroborates this. The authors position this as evidence that deep circuits with complex ansatzes can fail for algorithmic reasons independent of hardware noise, and that practical quantum advantage for PDEs in the NISQ era will require new cost functions, shallow problem-specific ansatzes, and hybrid classical-quantum optimization.

Load-bearing premise

The load-bearing premise is that the paper's implementation faithfully reproduces the earlier method, so the persistent truncation is caused by the cost function itself rather than by the chosen ansatz, optimizer, encoding, or initialization choices.

Editorial extensions

If this is right

  • If the cost function in Eq. 4 is not viable, then gate-count reduction alone cannot rescue this class of variational Poisson solvers on NISQ hardware.
  • Researchers should benchmark candidate cost functions on noiseless simulators against known analytical solutions before investing in hardware-specific optimization.
  • Tensor-network ansatzes, especially MPS, can mitigate barren plateaus and achieve higher fidelity than TTN on limited-connectivity hardware; TTN gains depend on all-to-all connectivity.
  • Gradient-free optimizers can escape barren plateaus, but only at the price of extra re-initializations and iterations, so total runtime must be part of the comparison.
  • A modular architecture that separates logical circuit construction from physical transpilation is what makes these ablation comparisons practical and should be part of reporting experimental quantum algorithm studies.

Reading between the lines

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

  • Editorial inference: the paper's evidence is not yet decisive; without a baseline run of the original algorithm unchanged or a classically exact state-vector ansatz, the observed truncation could be traced to ansatz expressibility, initialization, or optimizer failure rather than to Eq. 4 itself.
  • Editorial inference: if the cost-function flaw is real, it likely transfers to other linear PDE solvers built on the same minimum-potential-energy Rayleigh quotient, so the conclusion is broader than Poisson's equation.
  • Editorial inference: a cheap falsification is to minimize Eq. 4 with a fully expressive ansatz and a known analytical solution; if the exact solution is reached, the paper's central negative claim would be refuted.
  • Editorial inference: the 1D 4-qubit setting is too small to test the claimed curse-of-dimensionality advantage, so a next step is to test alternative cost functions on higher-dimensional grids while keeping the same modular pipeline.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper presents a case study of a variational quantum algorithm (VQA) for solving the 1D Poisson equation, building on the Rayleigh-quotient approach of Sato et al. The authors describe a modular software architecture separating logical circuit construction from transpilation, and report experiments that replace boundary-condition encoding (via V-chains or sparse Pauli operators), ansatz (MPS, TTN, custom MPS), and optimizer (gradient-free methods such as Nelder-Mead) relative to Sato et al. Their main empirical findings are reduced circuit depth (Figure 3), fidelity comparisons on the IBM Osaka hardware (Table 1), and an observed truncated solution on a noiseless 4-qubit QASM simulator (Figure 4), which they interpret as evidence that the Sato et al. cost function (Eq. 4) may not be viable for ensuring convergence. The paper concludes with recommendations for alternative cost functions and shallow, problem-specific ansatze, as well as a software engineering discussion of factories, context managers, and backend-agnostic circuit construction.

Significance. If the negative result about Eq. 4 were rigorously established, it would be a useful caution for the VQA-for-PDE community and could redirect research toward alternative cost functions. The software architecture contribution is modular and potentially reusable, and the circuit-depth and fidelity data points are of interest to practitioners. However, the empirical basis for the central claim is currently too thin: a single noiseless 4-qubit run, no baseline against the original Sato algorithm, no error bars, and no ablation isolating the cost function from the many other modified components. The paper also misattributes a polylogarithmic convergence claim to HHL, which weakens the motivational framing. As a case study, the paper is within the journal's scope, but the central negative claim needs substantially more evidence before it can be accepted.

major comments (4)
  1. [Results, Figure 4] The central negative claim—that the Sato et al. cost function of Eq. 4 'may not be a viable method for ensuring convergence'—rests on a single noiseless 4-qubit simulation. The Methods section states that, relative to Sato et al., the authors replaced the boundary-condition encoding, the ansatz, and the optimizer. Each of these changes (V-chains/sparse Pauli operators, tensor-network ansatze, gradient-free optimizers) can independently cause convergence failure. To support the claim, the authors must hold all other algorithmic choices fixed and vary only the cost function, or run the original Sato pipeline (original encoding, ansatz, optimizer) and show that it succeeds on the same problem while their modified pipeline fails. Without such a baseline or ablation, the evidence does not discriminate between 'Eq. 4 is flawed' and 'the authors' modified pipeline is flawed.'
  2. [Methods, Eq. (4)] Equation (4) contains the term ⟨f, ψ(θ)| X ⊗ I⊗n |f, ψ(θ)⟩, which is not a well-formed expectation value: X appears to be a single-qubit Pauli operator, the state |f, ψ(θ)⟩ is not defined as a normalized vector on which such an operator acts, and the notation does not match a standard inner product. This ambiguity matters because, if the printed equation reflects an implementation error, the observed truncation in Figure 4 would be explained without indicting the cost-function idea. The authors should rewrite Eq. 4 with a clear definition of all operators and states, and specify the exact operator whose expectation value is measured.
  3. [Background, PDE Scaling] The claim that 'the number of iterations for VQAs to converge scales polylogarithmically with the number of grid points' is attributed to Ref. [5], Harrow, Hassidim, and Lloyd's HHL algorithm. HHL is a non-variational quantum algorithm for linear systems and does not make a claim about VQA iteration counts. This is a load-bearing motivational claim for the paper's framing, and it should either be corrected with an appropriate reference or removed.
  4. [Methods, Table 1] Table 1 reports fidelity percentages for four ansatz types on IBM Osaka hardware, and the text states that 'MPS ansatze showed consistently higher fidelity.' No shot counts, number of circuit executions, error-mitigation procedures, or statistical uncertainties are reported. With differences of a few percent, one cannot judge whether the observed ordering is meaningful. The authors should provide error bars or raw counts, or weaken the claim to a tendency observed under the specific experimental conditions.
minor comments (5)
  1. [Throughout] The manuscript contains several typographical errors, including 'netwok' and 'consistantly' in the section on avoiding barren plateaus, and a stray 'M' at the start of the first paragraph of the Introduction; these should be corrected.
  2. [Background, Eq. (3)] Equation (3) is written as E0 = ⟨ψ*|H|ψ*⟩ ≤ min_{|ψ⟩} ...; this is mathematically incorrect, because the ground-state energy is the minimum of the Rayleigh quotient, not an upper bound. The equality chain should be fixed to state E0 = min_{|ψ⟩} ⟨ψ|H|ψ⟩/⟨ψ|ψ⟩ = min_{|ψ⟩} ⟨ψ|H|ψ⟩ under the normalization assumption.
  3. [Methods, Eq. (4)] The variational parameter r is used in Eq. (4) without definition; the authors later refer to a 'variational scaling parameter r,' but its range, its role in the optimization, and how it is updated should be specified in the Methods section.
  4. [Results, Figure 4] The text describes the solution as 'truncated' but provides no quantitative error metric, such as the L2 distance to the finite-difference solution. Adding such a metric would make the negative result more informative and comparable across different cost functions and ansatze.
  5. [Architecture] The paper claims that the modular architecture enables 'ablation studies,' but no ablation study is actually reported; the results compare different pipelines as a whole rather than isolating individual components. A concrete example of an ablation (e.g., fixing ansatz and optimizer while varying the cost function) would strengthen both the software-engineering and the scientific claims.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's negative claim about the Sato cost function is an empirical observation, not a derivation forced by the paper's own equations or by self-citation.

full rationale

The paper's central negative claim is that the Sato et al. cost function (Eq. 4) 'may not be a viable method for ensuring convergence to the correct solution,' based on a noise-free 4-qubit QASM simulation shown in Figure 4. This is an empirical observation about a specific implementation, not a result derived from an input that already contains the conclusion. No parameter is fitted to data and then renamed as a prediction; no quantity in Eq. 4 is defined in terms of the paper's own outputs; and the references are to external prior work (Sato et al. [12], Cerezo et al. [14], etc.) rather than to a load-bearing chain of the present authors' own unpublished theorems. The strongest concern about the paper is underdetermination, not circularity: the experiment simultaneously varies encoding, ansatz, and optimizer relative to Sato et al., and no baseline run of the original algorithm or ablation isolating the cost function is reported. That means the observed truncation may be an artifact of the authors' modified pipeline, so the evidence does not discriminate between 'Eq. 4 is flawed' and 'the authors' implementation is flawed.' Under the rules of this review, that is a correctness and validity risk, not a circularity, because the paper's conclusion is not equivalent to its premises by construction. Accordingly, no circular step is identified.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central claim rests on two variational parameters (r, θ) that are optimized but never reported, plus assumptions inherited from Sato et al. and the broader quantum PDE literature: the logarithmic qubit scaling, the polylogarithmic VQA convergence claim (cited to HHL, which does not support it), and the barren-plateau mitigation properties of tensor networks. No new entities are introduced.

free parameters (2)
  • Variational scaling parameter r = not reported
    Introduced in Eq. 4 (from Sato et al.) and minimized alongside θ. The value affects the balance between kinetic and potential terms in the cost function; no values or ranges are reported, making the negative result hard to reproduce.
  • Ansatz parameters θ = not reported
    All variational parameters in MPS, TTN, TTN++, and custom MPS ansatze; optimized classically. Their initialization and updates are not specified, so the convergence failure in Figure 4 could depend on these choices.
assumptions (4)
  • domain assumption Representing a grid of N points requires O(log N) qubits via amplitude encoding.
    Used in 'PDE Scaling and Quantum Computing' to motivate quantum advantage. State preparation and readout costs are not included, but the representational claim itself is standard.
  • ad hoc to paper VQA iteration count scales polylogarithmically with the number of grid points, as stated in Background.
    The paper cites Harrow et al. [5], a quantum linear systems result, as support for VQA convergence scaling. That reference does not prove anything about variational algorithm iterations; no derivation is given.
  • domain assumption The Rayleigh quotient in Eq. 4 is a valid cost function for the Poisson equation, per Sato et al. [12].
    The central negative result tests this cost function, so it is an imported premise rather than something the paper derives. If the cost function were invalid by construction the negative finding would be trivial.
  • domain assumption Tensor network ansatze mitigate barren plateaus, following Cerezo et al. [14].
    The design choice to use MPS/TTN ansatze relies on this cited theoretical result; the paper does not independently verify barren plateau behavior.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Quantum Mini-Apps for Engineering Applications: A Case Study." pith.science (2026). https://pith.science/paper/BARMBONF

@misc{pith2026241112920,
  author       = {Pith},
  title        = {Pith review of: Quantum Mini-Apps for Engineering Applications: A Case Study},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BARMBONF}},
  note         = {Machine review of arXiv:2411.12920}
}
read the original abstract

In this work, we present a case study in implementing a variational quantum algorithm for solving the Poisson equation, which is a commonly encountered partial differential equation in science and engineering. We highlight the practical challenges encountered in mapping the algorithm to physical hardware, and the software engineering considerations needed to achieve realistic results on today's non-fault-tolerant systems.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

14 extracted references · 12 canonical work pages

  1. [5]

    Quantum algorithm for linear systems of equations

    Harrow, Aram W., Avinatan Hassidim, and Seth Lloyd. "Quantum algorithm for linear systems of equations." Physical Review Letters 103, no. 15 (2009): 150502

    work page 2009

  2. [1]

    What is an ablation study? And is there a systematic way to perform it?

    Long, Robert. "What is an ablation study? And is there a systematic way to perform it?" Stats Exchange URL (2020-06-11): https://stats.stackexchange.com/q/380233

    work page 2020

  3. [2]

    A note on the complexity of solving Poisson’s equation for spaces of bounded mixed derivatives

    Bungartz, Hans-Joachim, and Michael Griebel. "A note on the complexity of solving Poisson’s equation for spaces of bounded mixed derivatives." Journal of Complexity 15, no. 2 (1999): 167-199

    work page 1999

  4. [3]

    The finite element method for elliptic problems

    Ciarlet, Philippe G. The finite element method for elliptic problems. Society for Industrial and Applied Mathematics, 2002

    work page 2002

  5. [4]

    Quantum algo- rithm implementations for beginners

    Adedoyin, Adetokunbo, John Ambrosiano, Petr Anisi- mov, William Casper, Gopinath Chennupati, Car- leton Coffrin, Hristo Djidjev et al. "Quantum algo- rithm implementations for beginners." arXiv preprint arXiv:1804.03719 (2018)

    arXiv 2018

  6. [6]

    Quantum mechanical comput- ers

    Feynman, Richard P . "Quantum mechanical comput- ers." Found. Phys. 16, no. 6 (1986): 507-532

    work page 1986

  7. [7]

    Press release at https://www- 03.ibm.com/press/us/en/pressrelease/49661.wss (2016)

    IBM Makes Quantum Computing Avail- able on IBM Cloud to Accelerate In- novation. Press release at https://www- 03.ibm.com/press/us/en/pressrelease/49661.wss (2016)

    work page 2016

  8. [8]

    Quantum computing in the NISQ era and beyond

    Preskill, John. "Quantum computing in the NISQ era and beyond." Quantum 2 (2018): 79

    work page 2018

Show all 14 references
  1. [9]

    Quantum Data Encoding: A Com- parative Analysis of Classical-to-Quantum Mapping Techniques and Their Impact on Machine Learning Accuracy

    Rath and Date. "Quantum Data Encoding: A Com- parative Analysis of Classical-to-Quantum Mapping Techniques and Their Impact on Machine Learning Accuracy" arXiv preprint arXiv:2311.10375 (2023)

    2023 arXiv

  2. [10]

    Determining eigenstates and thermal states on a quantum com- puter using quantum imaginary time evolution

    Motta, M., Sun, C., Tan, A.T.K. et al. "Determining eigenstates and thermal states on a quantum com- puter using quantum imaginary time evolution." Nat. Phys. 16 (2020): pp. 205–210

    2020

  3. [11]

    Stability of the Trot- ter–Suzuki decomposition

    Ish Dhand and Barry C Sanders. "Stability of the Trot- ter–Suzuki decomposition" J. Phys. A: Math. Theor. 47 (2014): 265206

    2014

  4. [12]

    Variational quantum al- gorithm based on the minimum potential energy for solving the Poisson equation

    Sato, Yuki, Ruho Kondo, Satoshi Koide, Hideki Taka- matsu, and Nobuyuki Imoto. "Variational quantum al- gorithm based on the minimum potential energy for solving the Poisson equation." Physical Review A 104, no. 5 (2021): 052409

    2021

  5. [13]

    Ablation of artificial neural networks

    Vishnusai, Y ., Tejas R. Kulakarni, and K. Sowmya Nag. "Ablation of artificial neural networks." In Inno- vative Data Communication T echnologies and Applica- tion: ICIDCA 2019, pp. 453-460. Springer International Publishing, 2020

    2019

  6. [14]

    Cost function dependent barren plateaus in shallow parametrized quantum cir- cuits

    Cerezo, Marco, Akira Sone, Tyler Volkoff, Lukasz Cincio, and Patrick J. Coles. "Cost function dependent barren plateaus in shallow parametrized quantum cir- cuits." Nature Communications 12, no. 1 (2021): 1791. Horia M ˘arg˘arit is a veteran statistician, computer scientist, a...

    2021

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.