Demonstrating Quadratic Monte Carlo Speedup via Quantum Amplitude Estimation: Nuclear Engineering Examples

Akira Sone; Jilang Miao; Miaomiao Jin

REVIEW 2 major objections 6 minor 9 references

Quantum amplitude estimation delivers quadratic Monte Carlo speedup on nuclear engineering expectations, recovering a U-238 resonance integral to 0.03% error with far fewer oracle calls than classical sampling.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-14 09:21 UTC pith:CCFYDRHK

load-bearing objection Solid, transparent numerical demo that standard QPE-QAE recovers the expected O(1/T^{2}) scaling on two nuclear expectation values; the resonance case uses an ideal eigendecomposition surrogate that the authors flag clearly. the 2 major comments →

arxiv 2607.10772 v1 pith:CCFYDRHK submitted 2026-07-12 quant-ph physics.comp-ph

Demonstrating Quadratic Monte Carlo Speedup via Quantum Amplitude Estimation: Nuclear Engineering Examples

Jilang Miao , Miaomiao Jin , Akira Sone This is my paper

classification quant-ph physics.comp-ph

keywords quantum amplitude estimationMonte Carlo speedupquantum phase estimationresonance integralfission neutron yieldnuclear engineeringU-238Grover iterate

verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Classical Monte Carlo is the gold standard for high-fidelity reactor calculations, but its error falls only as one over the square root of the number of particle histories. This paper shows that quantum amplitude estimation can improve that scaling to one over the number of oracle calls, a quadratic advantage. Using quantum phase estimation, the authors encode two nuclear quantities as amplitudes: a simple mean fission-neutron yield and the U-238 resonance integral under a 1/E slowing-down spectrum. In both cases the squared error falls as 1/T squared, matching theory. For the resonance integral they recover 120.65 barns versus the true 120.69 barns (0.03 percent relative error) with 14 phase-estimation qubits, a precision that would classically require tens of millions of samples. The work therefore supplies a concrete demonstration that quantum speedup for Monte Carlo-type nuclear integrals is already visible on present-day simulators and points toward hardware-ready circuits for larger resonance problems.

Core claim

QPE-based quantum amplitude estimation applied to two nuclear-engineering expectation values produces squared error that scales as O(1/T^{2}) with oracle calls T, confirming the theoretical quadratic improvement over classical Monte Carlo's O(1/N). For the U-238 resonance integral the method reaches approximately 0.03 percent relative error at m=14 phase-estimation qubits.

What carries the argument

Quantum amplitude estimation via quantum phase estimation of the Grover iterate Q: the target expectation is encoded as the success probability μ = sin^{2} heta of an ancilla, Q rotates the good/bad subspace by 2 heta, and QPE extracts heta (hence μ) from the eigenphase after T = 2^m - 1 applications of Q.

Load-bearing premise

The resonance-integral result rests on an exact matrix eigendecomposition of the Grover operator rather than a fully compiled gate-level circuit, because state-preparation decomposition already becomes prohibitive at nine system qubits.

What would settle it

Compile an explicit gate-level circuit for the same 256-bin Breit-Wigner resonance integral (or a smaller but still multi-controlled version) and measure whether the observed squared-error scaling remains O(1/T^{2}) once state-preparation cost and realistic noise are included.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

Nuclear Monte Carlo integrals that today require tens of millions of particle histories can in principle be estimated to the same precision with a few tens of thousands of coherent oracle calls.
Once fault-tolerant arithmetic circuits for Breit-Wigner amplitudes exist, the same QAE pipeline can be extended to n ≥ 18 energy bins and multiple resonances.
Iterative amplitude-estimation variants can be substituted for full QPE to reduce circuit depth on near-term noisy hardware while retaining the quadratic scaling.
The same encoding of a 1/E spectrum and energy-dependent cross section applies immediately to other resonance integrals and reaction-rate tallies used in reactor physics.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

If the state-preparation bottleneck can be removed by quantum arithmetic rather than multi-controlled rotations, the demonstrated three-order-of-magnitude sample-count advantage becomes a realistic target for early fault-tolerant machines.
The same QAE template can be reused for other continuum integrals in nuclear data evaluation (Doppler-broadened resonances, unresolved-resonance averages) without redesigning the amplification step.
Noise-aware iterative QAE would let current intermediate-scale devices test the quadratic scaling on the simpler fission-yield toy problem before full resonance circuits become available.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

Solid, transparent numerical demo that standard QPE-QAE recovers the expected O(1/T^{2}) scaling on two nuclear expectation values; the resonance case uses an ideal eigendecomposition surrogate that the authors flag clearly.

read the letter

This is a clean methods note that takes the known QAE quadratic speedup and shows it works, in ideal simulation, on two nuclear-engineering expectations: a four-outcome fission-yield mean and a U-238 resonance integral under a 1/E spectrum. That domain application is the actual novelty; the algorithm and the error bound are textbook Brassard/Montanaro.

What they do well is straightforward. The toy problem is a full gate-level Qiskit circuit with an explicit Ry-tree state prep; the squared-error plots sit on the analytical mode-error curve and give the clean slope -2 versus classical -1. For the resonance integral they discretize into 256 bins, build the 512-dimensional Grover operator analytically, and run QPE via eigendecomposition + FFT. At m=14 they recover IR to ~0.03 % relative error and show that matching that MSE classically would take ~3 000 imes more samples. The math is standard, the classical baselines are side-by-side, and they never hide the surrogate.

The soft spot is exactly the one they name: multi-controlled state preparation for n≥9 is already prohibitive on the simulator, so the more interesting example never produces a hardware-ready circuit. Noise and realistic prep cost could erase the practical edge at this scale; they point to future arithmetic circuits and iterative QAE. No code is shipped. Those are real limitations on impact, not on the ideal-case claim that is actually asserted.

Who it is for: people already working quantum-for-nuclear or looking for concrete QAE encodings outside finance. It is not a theoretical advance and does not claim to be. The citation pattern is appropriate (Brassard, Montanaro, finance demos, the one prior nuclear preprint). I would send it to referees; it is honest enough and useful enough for that niche that a serious editor should not desk-reject it. Engage if the nuclear-quantum intersection is on your radar; otherwise file it as a clean confirmation note.

Referee Report

2 major / 6 minor

Summary. The manuscript applies standard QPE-based quantum amplitude estimation (QAE) to two nuclear-engineering expectation values: a four-outcome mean fission-neutron yield and the U-238 6.674 eV resonance integral under a 1/E spectrum. The toy problem is realized as a gate-level Qiskit circuit; the resonance-integral case is obtained by exact eigendecomposition of the 512×512 Grover operator to avoid state-preparation decomposition cost. In both cases the squared error is shown to follow the theoretical O(1/T²) scaling with oracle calls T (Eq. 3), in contrast to classical Monte Carlo O(1/N). At m=14 the resonance integral is recovered to ~0.03% relative error, corresponding to a query-complexity speedup ratio N/T of order 10³ relative to classical sampling at the same MSE.

Significance. The work is a clean numerical confirmation of the known Brassard–Montanaro QAE convergence theorem on two physically motivated nuclear integrands. Strengths include (i) an explicit gate-level Qiskit construction for the discrete yield problem that matches the analytical mode-error formula, (ii) side-by-side classical MC baselines on the same log-log plots, and (iii) transparent disclosure that the resonance-integral demonstration uses an eigendecomposition surrogate. The paper does not claim a new algorithm or a hardware-ready circuit for the larger problem; its value is as a domain-specific demonstration that the ideal-case quadratic query advantage is attainable for reactor-physics expectation values. That is a useful, if incremental, contribution for the nuclear-engineering / quantum-computing interface.

major comments (2)

[§II U-238 Resonance Integral / Conclusions] §II (U-238) and Conclusions: the resonance-integral result that underpins the abstract’s 0.03% / N/T≈3016 claim is obtained from an exact eigendecomposition of Q, not from a compiled circuit. While the paper states that the surrogate is “mathematically identical” under ideal simulation, the reported speedup ratio compares classical sample count N only to the number of oracle applications T=2^m−1. It does not account for the O(2^{n_x}) multi-controlled Ry cost of the state-preparation operator A that is bypassed by the surrogate. Because the central claim is a “demonstration” of quadratic Monte Carlo speedup for a nuclear quantity, the manuscript should either (a) supply a gate-complexity accounting that shows the query advantage survives realistic A cost, or (b) rephrase the abstract, results paragraph, and speedup ratio so that they are explicitly limited to the ideal query model and do
[Eq. (3), Figs. 2–3] Eq. (3) and Figs. 2–3: the plotted “QAE analytical mode error” and the circuit/eigendecomposition points coincide almost perfectly, which is expected only when |ψ⟩=A|0⟩ lies exactly in the two-dimensional eigenspace of Q and the mode of ŷ is used. The paper should state explicitly that the reported MSE is the squared mode-estimation error (infinite-shot limit of the most-probable register outcome), not the mean-squared error over the full QPE distribution. Without that clarification a reader may over-interpret the empirical points as shot-noise-limited performance rather than the ideal mode error already guaranteed by the theory.

minor comments (6)

[Abstract / Introduction] Abstract and Introduction: “quadratic speedup for Monte Carlo-type expectation values” is standard terminology, but a one-sentence reminder that the speedup is in the number of oracle calls (query complexity), not wall-clock time or total gate count, would prevent over-reading by a nuclear-engineering audience.
[Fig. 1] Fig. 1 caption: the circuit is shown for m=4; it would help to note that the controlled-Q^{2^k} blocks for m>4 are obtained by matrix powering rather than circuit repetition, as stated in the text.
[Eq. (2)] Eq. (2): the absolute-value form ˆθ=π|ŷ/2^m−1/2| is correct for the −e^{±2iθ} eigenvalues, but a brief remark that both peaks φ± map to the same µ would aid reproducibility.
[References] Reference [6] is cited as “arXiv preprint (2024)” without an identifier; please supply the arXiv number or DOI.
[Figs. 2–3] Notation: the same symbol T is used for oracle calls and (implicitly) for temperature nowhere, but the classical sample count is N; a short glossary or consistent “T_QAE vs N_MC” labeling on the dual axes of Figs. 2–3 would improve readability.
[§II] C_phys≈1578 b is given as a classically precomputed constant; stating the precise numerical value used (or the quadrature rule) would make the IR conversion fully reproducible.

Circularity Check

0 steps flagged

No circularity: numerical confirmation of known QAE O(1/T^{2}) scaling on nuclear integrands, with no fitted parameters or self-definitional reductions.

full rationale

The paper applies the standard QPE-based quantum amplitude estimation algorithm of Brassard et al. (and Montanaro’s Monte Carlo speedup theorem) to two nuclear-engineering expectation values. The claimed O(1/T^{2}) squared-error scaling is the known theoretical bound (Eq. 3) and is recovered numerically from the mode of the QPE register; it is not derived from, nor forced by, any quantity fitted to the target data. For the fission-yield toy problem the state-preparation circuit is fully gate-level and the decoded mode matches the analytical mode-error formula. For the U-238 resonance integral the 512-dimensional Grover operator is formed analytically and eigendecomposed, which the authors explicitly state is mathematically identical to the ideal circuit; the physical conversion constant C_phys is classically precomputed and multiplies the estimated amplitude µ. No free parameters are adjusted to produce the reported 0.03 % relative error, no uniqueness theorem is imported from the authors’ prior work, and no known empirical pattern is merely renamed. The only limitation is the acknowledged absence of a hardware-ready state-preparation circuit for n ≥ 9, which is an implementation caveat, not a circularity. Consequently the derivation chain is self-contained against external benchmarks and the circularity score is zero.

Axiom & Free-Parameter Ledger

3 free parameters · 4 axioms · 0 invented entities

The central claim rests on standard QAE theory plus concrete nuclear encodings. Free parameters are problem discretizations chosen by hand; they affect approximation quality of the continuous integral but do not enter the claimed asymptotic scaling. No new physical entities are postulated.

free parameters (3)

Nbins (nx=8)
Number of energy bins (256) chosen by hand for the resonance-integral discretization; controls the quality of the continuous-integral approximation but not the QAE scaling itself.
energy window [E0 ± 10Γ]
Integration limits around the 6.674 eV resonance chosen by hand.
toy yield probabilities P=(0.1,0.2,0.3,0.4)
Simplified four-outcome fission-neutron distribution chosen for the minimal benchmark.

axioms (4)

standard math QAE mode-estimation error bound |µ̂−µ| ≤ π|sin 2θ|/T + π²/T² (Brassard et al.)
Invoked to claim O(1/T²) squared-error scaling; used throughout the results sections and figures.
standard math Grover iterate Q rotates the good/bad subspace by angle 2θ per application
Core of amplitude amplification; used to justify phase estimation of θ.
domain assumption 1/E slowing-down spectrum and single-level Breit-Wigner parameters for the U-238 6.674 eV resonance
Taken from nuclear data (Mughabghab) and used to define P(x) and f(x) for the resonance integral.
domain assumption Ideal noiseless statevector simulation (no decoherence, perfect gates)
All numerical results are obtained on a Qiskit statevector simulator or by exact eigendecomposition; noise is deferred to future work.

pith-pipeline@v1.1.0-grok45 · 11230 in / 2744 out tokens · 56680 ms · 2026-07-14T09:21:38.184806+00:00 · methodology

0 comments

read the original abstract

We demonstrate quantum amplitude estimation (QAE) as a route to quadratic speedup for Monte Carlo-type expectation values in nuclear engineering. Using QPE-based QAE, we study two examples: a discrete fission-neutron-yield expectation and a U-238 resonance integral under a $1/E$ slowing-down spectrum. The toy problem is implemented as a gate-level Qiskit circuit, while the resonance-integral example is simulated through an exact eigendecomposition of the Grover operator to avoid state-preparation decomposition bottlenecks. In both cases, the squared error scales as $O(1/T^2)$ with the number of oracle calls $T$, compared with the classical Monte Carlo scaling $O(1/N)$. For the U-238 example, QAE recovers the resonance integral to approximately $0.03%$ relative error with $m=14$ phase-estimation qubits.

Figures

Figures reproduced from arXiv: 2607.10772 by Akira Sone, Jilang Miao, Miaomiao Jin.

**Figure 1.** Figure 1: QPE circuit for the ν¯ estimation problem, m = 4 ancilla qubits (QPE register, top) and n = 3 system qubits (bottom, including ancilla). State prep A loads P(ν) via an Ry tree and f(ν) via controlled Ry rotations on the ancilla; controlled-Q 2 k blocks amplify the good amplitude; QFT† maps the phase to the register outcome. 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 8 10 9 N (classical samples) 10 13 10 11 10 9… view at source ↗

**Figure 3.** Figure 3: shows the MSE in µˆ versus T = 2 m − 1 oracle calls (QAE) and versus classical sample count N (MC sampling from P(x)). The quadratic convergence advantage is again clearly visible. At m = 14 (T = 16,383 oracle calls) the estimate µˆ = 0.07645 gives IR = 120.65 b, a 0.03% error. The MSE in µˆ at m = 14 is 6.86 × 10−10, while matching this with classical MC requires N = σ 2 f /MSE ≈ 49,400,000 samples, givin… view at source ↗

discussion (0)

Reference graph

Works this paper leans on

9 extracted references

[1]

Challenges and Prospects for Whole- Core Monte Carlo Analysis,

W. R. MARTIN, “Challenges and Prospects for Whole- Core Monte Carlo Analysis,”Nuclear Engineering and Technology,44,2, 151–160 (2012)

2012
[2]

Quantum Amplitude Amplification and Estimation,

G. BRASSARD et al., “Quantum Amplitude Amplification and Estimation,”Contemporary Mathematics,305, 53–74 (2002)

2002
[3]

Quantum speedup of Monte Carlo methods,

A. MONTANARO, “Quantum speedup of Monte Carlo methods,”Proceedings of the Royal Society A,471,2181, 20150301 (2015)

2015
[4]

Quantum risk analysis,

S. WOERNER and D. J. EGGER, “Quantum risk analysis,” npj Quantum Information,5, 15 (2019)

2019
[5]

Option Pricing using Quan- tum Computers,

N. STAMATOPOULOS et al., “Option Pricing using Quan- tum Computers,”Quantum,4, 291 (2020)

2020
[6]

Monte Carlo particle trans- port on quantum computers,

N. OLIVIER and M. NOW AK, “Monte Carlo particle trans- port on quantum computers,”arXiv preprint(2024)

2024
[7]

Quantum computing with Qiskit,

A. JA V ADI-ABHARI et al., “Quantum computing with Qiskit,”arXiv preprint(2024)

2024
[8]

M. A. NIELSEN and I. L. CHUANG,Quantum Compu- tation and Quantum Information, Cambridge University Press (2000)

2000
[9]

S. F. MUGHABGHAB,Atlas of Neutron Resonances: Res- onance Parameters and Thermal Cross Sections. Z=1–100, Elsevier, 5 ed. (2006)

2006

[1] [1]

Challenges and Prospects for Whole- Core Monte Carlo Analysis,

W. R. MARTIN, “Challenges and Prospects for Whole- Core Monte Carlo Analysis,”Nuclear Engineering and Technology,44,2, 151–160 (2012)

2012

[2] [2]

Quantum Amplitude Amplification and Estimation,

G. BRASSARD et al., “Quantum Amplitude Amplification and Estimation,”Contemporary Mathematics,305, 53–74 (2002)

2002

[3] [3]

Quantum speedup of Monte Carlo methods,

A. MONTANARO, “Quantum speedup of Monte Carlo methods,”Proceedings of the Royal Society A,471,2181, 20150301 (2015)

2015

[4] [4]

Quantum risk analysis,

S. WOERNER and D. J. EGGER, “Quantum risk analysis,” npj Quantum Information,5, 15 (2019)

2019

[5] [5]

Option Pricing using Quan- tum Computers,

N. STAMATOPOULOS et al., “Option Pricing using Quan- tum Computers,”Quantum,4, 291 (2020)

2020

[6] [6]

Monte Carlo particle trans- port on quantum computers,

N. OLIVIER and M. NOW AK, “Monte Carlo particle trans- port on quantum computers,”arXiv preprint(2024)

2024

[7] [7]

Quantum computing with Qiskit,

A. JA V ADI-ABHARI et al., “Quantum computing with Qiskit,”arXiv preprint(2024)

2024

[8] [8]

M. A. NIELSEN and I. L. CHUANG,Quantum Compu- tation and Quantum Information, Cambridge University Press (2000)

2000

[9] [9]

S. F. MUGHABGHAB,Atlas of Neutron Resonances: Res- onance Parameters and Thermal Cross Sections. Z=1–100, Elsevier, 5 ed. (2006)

2006