{"id":"35cc706c-3f7e-49aa-aa01-be6be0f0a66d","arxiv_id":"2411.17482","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The authors rebuild the phase-shifting circuit of their quantum multislice algorithm using Walsh functions, removing multi-controlled gates and adding an error-controlled truncation that cuts gate count by over an order of magnitude.","lead":"This paper presents an improved quantum circuit for simulating electron diffraction, replacing complex multi-control gates with simpler gates. The authors also add a truncation step that reduces the number of gates by more than ten times while keeping simulation error around one percent.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed size-improving truncation speedup rests on Eq. (17), fit to 12/14/16 qubits for one Au crystal; no evidence the Walsh spectrum remains sparse at larger N or for other specimens.","rationale":"The reader's CONDITIONAL verdict is well founded; my stress-test pass found no reason to move it. The most load-bearing assumption is the one the reader identifies: the truncation threshold formula Eq. (17) is the bridge between the demonstrated 12-16 qubit simulations and the conclusion that the gate-count reduction grows with system size. Without that extrapolation, the paper's central improvement reduces to a circuit compilation (multi-controlled gates replaced by Walsh circuits) plus a size-dependent truncation whose benefit at scale is unverified. I considered whether the Walsh circuit construction itself is unsound, but the authors verify it against the classical multislice method in Fig. 8, and the Welch et al. construction is established. I also considered the missing end-to-end resource count; that is a real limitation for quantum-advantage claims, but it is not the load-bearing point for the algorithmic gate-count claim. The empirical formula is the weakest link because the paper's own Fig. 12 shows only three sizes, and the extrapolation in Fig. 13 is a log-scale trend over a factor of 16 in Hilbert-space dimension for a single material. The proposed test—adding 18/20 qubits and a second specimen—would directly determine whether the trend holds. Until then, the verdict remains CONDITIONAL, and no adjustment is needed relative to the reader's assessment.","tokens_in":11538,"tokens_out":5907,"duration_ms":59942,"concrete_test":"Classically compute the Walsh decomposition of the Au potential and kinetic operators for 18 and 20 qubits (N=512 and 1024), apply Eq. (17) to set the truncation threshold, and count retained terms and the Eq. (16) relative error after the same multislice iteration count. If the retained term count grows faster than about 2^(0.5Q) or the error exceeds 1%, the 'speedup improves with size' claim is unsupported. Also repeat the 12/14/16-qubit truncation study for a second specimen, e.g., Si at 100 keV with the same slice configuration; if the remaining-term curve differs materially from Fig. 13, Eq. (17) is not a general parameter setting.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central scaling claim—that truncation reduces gates by more than an order of magnitude and that the speedup improves with qubit count—depends on the number of retained Walsh terms growing much more slowly than the full basis. Eq. (17) is an empirical fit to three system sizes (12, 14, 16 qubits) and a single Au specimen at 100 keV. Fig. 13 shows the remaining term count rising from 8,120 to 23,778 while the full basis grows by a factor of 16, but three points cannot establish the asymptotic exponent, and the paper does not tie the Walsh coefficient tail to any physical property that would guarantee the trend continues. If the potential's Walsh spectrum fills in at larger N, the remaining term count can approach the full N^2 basis and the order-of-magnitude gate reduction disappears. The same concern applies to other materials: atomic potentials with different screening lengths or multiple atoms per cell will have different Walsh coefficient distributions, so a threshold fitted to Au need not keep the stated 1% error.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents an improved quantum algorithm for the multislice method used in transmission electron microscopy simulations. The authors replace the multi-controlled phase-shifting gates of their previous quantum circuit by a Walsh-Hadamard decomposition, so that the phase-shifting operator is implemented with single-qubit phase gates and CNOT gates only. They then propose a truncation of small Walsh coefficients to reduce the gate count at the cost of a controlled approximation error. The improved exact circuit is verified against the previous quantum algorithm and the classical multislice algorithm for a 100 keV electron beam on a Au specimen with 12 qubits. Truncation thresholds are tested at 12, 14, and 16 qubits, leading to an empirical formula, Eq. (17), intended to keep an average relative error near 1% while reducing the number of gates by more than an order of magnitude. The authors conclude that the speedup from truncation improves as the number of qubits increases.","tokens_in":11703,"tokens_out":5084,"duration_ms":47792,"significance":"If the central claims hold, this is a meaningful step toward making the quantum multislice algorithm more hardware-friendly: the replacement of multi-controlled gates by one- and two-qubit gates is a concrete improvement, and the exact version of the improved circuit is a clean application of the known diagonal-unitary construction by Welch et al. The truncation idea is interesting and the paper provides simulation evidence, including a direct comparison with the classical multislice result. However, the claimed size-improving truncation speedup rests on an empirical fit to three system sizes and a single specimen, and the error metric used for the 1% claim is a full-state average rather than an observable-specific error; these points limit the generality of the conclusions. The paper is therefore of moderate significance pending a strengthening of the scaling evidence.","major_comments":[{"comment":"The threshold formula and the conclusion that the truncation speedup improves with qubit number rest on three simulated sizes (12, 14, and 16 qubits) and a single Au specimen at 100 keV. The paper offers no physical argument that the Walsh coefficient tail remains sparse as N grows or for other materials; if the tail fills in at larger N, the remaining-term percentage could increase and the order-of-magnitude gate reduction would disappear. This is a load-bearing extrapolation for the central claim, so it needs either a theoretical bound on the coefficient decay or validation on additional system sizes and materials.","section":"Sec. 4, Eq. (17), Figs. 12-13"},{"comment":"The 1% error claim refers to the average relative error over the full wave function defined in Eq. (16), not to errors in the extracted electron diffraction pattern or other observables. The truncation threshold is tuned to this particular metric, and the paper asserts rather than justifies that this metric is the right proxy for the application. The statement that the error is 'controllable' should be specified with respect to the metric actually controlled.","section":"Sec. 4, Eq. (16)"},{"comment":"The complexity statements ('complexity advantage', 'speedup effect may increase') are not accompanied by a formal accounting that includes state preparation, measurement repetitions, and compilation of the improved circuit to hardware-native gates. Figure 15 reports raw gate counts from the simulation; without the omitted overheads, the asymptotic comparison is not quantified and the quantum-advantage claim is not fully supported.","section":"Sec. 4, Sec. 5"},{"comment":"The reported 'original' term numbers (65,010; 261,057; 1,046,709 for 12, 14, and 16 qubits) appear to count Walsh terms across all 16 potential slices, whereas a single multislice iteration uses only one potential operator. The relationship between these term counts, the per-iteration gate count, and the total simulation gate count is not made explicit, which makes the reported reduction factor ambiguous.","section":"Sec. 4, Figs. 13 and 15"}],"minor_comments":[{"comment":"Several equations contain garbled mathematical expressions (for example, Eq. (17) is printed as '3128 102 V nτ −=×'); the manuscript needs a careful proofreading pass so that all formulas are readable.","section":"Equations (4), (6), (14), (17)"},{"comment":"There are typos such as 'multisilce' in the abstract and 'It can be proven' in Sec. 3; these should be corrected to 'multislice' and 'proved'.","section":"Abstract, Sec. 3"},{"comment":"The caption lists several numbers without a clear legend; a small table giving the original term count, remaining term count, truncation threshold, and error for each qubit count would improve readability.","section":"Fig. 13"},{"comment":"The discussion of initial state preparation cites Ref. [35] for O(N) state preparation and Ref. [36] for sparse states, but these costs are not included in the later complexity comparison; please clarify their role in the overall algorithm.","section":"Sec. 2"},{"comment":"The statement that a phase-shifting gate without control qubits 'will result in a phase shift for half of the basis states' is imprecise: it applies to a single-qubit rotation on one register qubit, and the correspondence to a Walsh function should be stated explicitly.","section":"Sec. 3"}],"recommendation":"major_revision","confidential_remarks":"The exact circuit replacement is a genuine improvement and is verified by simulation, so the core of the paper is defensible. The main risk is the extrapolation from three simulated sizes and one specimen to a general scaling claim; this needs to be either substantiated with more data or softened. The error-metric issue is also important for the 1% claim. I recommend major revision rather than rejection, as these points can be addressed within the manuscript's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Joe, quick take on this one. The paper does something genuinely useful: it replaces the multi-controlled phase-shifting gates in the authors' earlier quantum multislice algorithm with a Walsh-decomposition circuit using only one- and two-qubit gates, extending a construction from Welch et al. (which they cite) from 1D to the 2D potential and kinetic operators of the multislice step. In exact form the improved circuit is verified against both their earlier quantum circuit and the classical multislice method, and the gate count is comparable, so no hidden overhead reappears. That part is solid and publishable.\n\nThe second contribution, truncation of the Walsh expansion, is more fragile. The authors show an order-of-magnitude gate reduction at roughly 1% average relative error for 12, 14, and 16 qubits on one Au specimen at 100 keV. Their threshold formula, Eq. (17), is an empirical fit to those three sizes, not a derived bound, and the stress-test note is right: nothing in the paper ties the sparsity of the Walsh spectrum to any physical invariant, so extrapolating to larger N or other materials is a claim, not a result. The same caveat applies to their 'speedup improves with size' statement — it rests on three points. This should be tempered in any revision.\n\nThree smaller soft spots. The introduction says the classical cost increases exponentially because of repeated FFTs; that is sloppy. Per slice it's O(N log N), and the exponential growth is in the number of states/configurations, not in the FFT itself. The error metric in Eq. (16) is a full-state average, which can miss localized errors in a diffraction pattern. And the resource comparison leaves out state preparation, sampling to estimate probabilities, and compilation overhead; the paper mentions sampling but does not fold it into the complexity claim. None of this invalidates the exact circuit, which is the real deliverable.\n\nWho should read this: people working on quantum simulation for electron microscopy or on diagonal-unitary circuit synthesis. I'd bring it to a reading group as an example of a well-scoped circuit optimization with a clear empirical component. The exact construction deserves a serious referee; the truncation claims need revision, not rejection. I'd accept it for peer review and push the authors to rewrite the scaling claims more carefully, correct the FFT statement, and ideally test at least one other material or qubit size before claiming asymptotic behavior.","headline":"Solid modest result: a 2D Walsh-based phase-shifting circuit for the multislice method that removes multi-controlled gates and demonstrably works in simulation; truncation scaling claims are empirical and over-reaching.","tokens_in":12266,"tokens_out":3077,"would_cite":true,"duration_ms":27410,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["03.67.Lx","61.05.jm"],"model":"deepseek-v4-flash","headline":"This paper shows that phase-shifting circuits in the quantum multislice algorithm can be rebuilt with only one- and two-qubit gates, and that truncating small Walsh terms cuts the gate count by over an order of magnitude at about 1…","keywords":["quantum algorithm","multislice method","Walsh-Hadamard transform","phase-shifting circuit","truncation optimization","electron diffraction","quantum simulation","transmission electron microscopy"],"falsifier":"Run the truncated circuit for a 20- or 24-qubit simulation of the same gold specimen, or for another material such as silicon or an oxide, using the threshold formula (17); the claim fails if the remaining-term percentage stops decreasing with qubit count or if the average relative error of the electron density exceeds about 1 percent.","tokens_in":11271,"feed_emoji":"🔬","tokens_out":10285,"duration_ms":82548,"temperature":0.7,"pith_summary":"The paper improves the authors' earlier quantum algorithm for the multislice method, which simulates how high-energy electrons scatter and diffract through a specimen in transmission electron microscopy. The improvement replaces every multi-controlled phase-shifting gate with a cascade of one- and two-qubit gates built from the Walsh-Hadamard expansion of the potential and propagator operators. It then shows that truncating small Walsh coefficients cuts the total gate count by more than an order of magnitude while keeping the average relative error near 1 percent, verified by classical simulation of 12-, 14-, and 16-qubit circuits for 100 keV electrons in gold. If that Walsh sparsity persists at larger sizes and for other materials, the result makes a practical quantum speedup for electron-diffraction simulations substantially more plausible.","feed_headline":"Walsh transform yields 10x smaller quantum diffraction circuit","feed_subtitle":"One- and two-qubit gates only, with error near one percent, for easier near-term hardware.","key_machinery":"The central object is the phase-shifting quantum circuit built from the Walsh-Hadamard expansion of a diagonal phase operator $U = e^{if(x)}$. Each retained Walsh basis term is implemented by one phase-shifting gate, and the CNOT gates connecting neighboring terms are ordered by Gray code so that successive Walsh indices differ by one bit. The authors extend the one-dimensional construction of [34] to two dimensions by reshaping the $N\\times N$ potential matrix into a vector of length $N^2$, which matches the qubit encoding with $n$ qubits per coordinate. A relative truncation threshold $\\tau$ removes Walsh coefficients below $\\tau$ times the maximum coefficient, and the empirical formula (17) sets the potential threshold as a function of $n$ while the kinetic threshold is fixed to keep all non-zero terms. This machinery is what converts a circuit with many multi-controlled gates into a comparable-size circuit of one- and two-qubit gates and then shrinks it further at controlled cost.","core_discovery":"The paper's central claim is that the phase-shifting part of the quantum multislice algorithm can be rebuilt exactly using the Walsh-Hadamard expansion of the diagonal phase operator, so that no multi-controlled gates are needed. Each Walsh basis term corresponds to a single phase-shifting gate, and the CNOT gates between them follow Gray-code ordering, giving a circuit whose one- and two-qubit gate count is comparable to the old multi-controlled circuit. The paper further claims that truncating Walsh coefficients below a threshold reduces the total gate count by more than an order of magnitude while holding the average relative error around 1 percent, with the remaining-term percentage falling as the qubit number grows. It supports these claims with classical simulations of 12-, 14-, and 16-qubit circuits for 100 keV electrons in a gold specimen, comparing the exact and truncated circuits against the classical multislice result.","pith_inferences":["Reader extension: the threshold formula is fitted to 12-, 14-, and 16-qubit runs for gold at 100 keV; testing other crystals, temperatures, and energies would show whether Walsh-coefficient sparsity is a general property of projected atomic potentials or particular to this specimen.","Reader extension: the kinetic operator's Walsh spectrum is described as having almost all coefficients zero, which suggests an analytic description of its support; deriving one could replace the empirical threshold fit with a rigorous gate-count bound as the qubit number grows.","Reader extension: because truncation acts on the operators rather than on the wave function, the same Walsh-truncation idea transfers to any diagonal-unitary simulation whose generator is sparse in the Walsh basis, such as other scattering or spectral-phase problems."],"forward_implications":["The phase-shifting circuit is now implementable with only one- and two-qubit gates, removing the multi-controlled-gate compilation overhead that threatened the earlier algorithm's advantage on real hardware.","Using the paper's threshold formula, the total gate count is reduced by more than one order of magnitude while the average relative error stays around 1 percent.","The retained fraction of Walsh terms decreases as the qubit count grows from 12 to 16, so the truncation benefit is expected to strengthen, not weaken, for larger simulations.","The classical preprocessing step, the fast Walsh-Hadamard transform, costs only $O(N \\log N)$ additions and is reusable across slices and input states, so it does not become a new bottleneck."],"supporting_citations":[{"why":"previous quantum multislice algorithm whose phase-shifting circuit is the baseline being improved and compared against.","marker":"[9]"},{"why":"provides the diagonal-unitary circuit construction via Walsh expansion that the paper extends to two dimensions.","marker":"[34]"},{"why":"defines the Walsh-Hadamard transform and basis functions used to expand and truncate the operators.","marker":"[38]"},{"why":"documents the compilation overhead of multi-controlled gates, motivating the circuit reconstruction.","marker":"[33]"},{"why":"supplies the quantum-circuit simulator used for the classical verification runs.","marker":"[41]"},{"why":"gives the atomic potential parameters used to build the gold specimen test case.","marker":"[43]"}],"fun_headline_variants":["Quantum multislice without multi-controlled gates","Walsh expansion reduces quantum diffraction circuit gates","1% error, 10x fewer gates for quantum diffraction","Walsh-Hadamard circuits cut quantum multislice gates","Gray-code CNOTs enable compact quantum multislice"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The gate-count reduction depends on the potential's Walsh expansion remaining sparse as the simulation grows and for materials other than the single gold specimen at 100 keV, since the paper's threshold formula is fitted to only three qubit counts (12, 14, and 16).","fun_headline_variants_meta":{"raw":{"variants":["Quantum multislice without multi-controlled gates","Walsh expansion reduces quantum diffraction circuit gates","1% error, 10x fewer gates for quantum diffraction","Walsh-Hadamard circuits cut quantum multislice gates","Gray-code CNOTs enable compact quantum multislice"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000631,"raw_usage":{"total_tokens":2870,"prompt_tokens":854,"completion_tokens":2016,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":470,"completion_tokens_details":{"reasoning_tokens":1937}},"tokens_in":470,"tokens_out":2016,"duration_ms":27600,"temperature":1.0,"reasoning_tokens":1937,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T12:04:16.281651+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the truncated circuit for a 20- or 24-qubit simulation of the same gold specimen, or for another material such as silicon or an oxide, using the threshold formula (17); the claim fails if the remaining-term percentage stops decreasing with qubit count or if the average relative error of the electron density exceeds about 1 percent.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"previous quantum multislice algorithm whose phase-shifting circuit is the baseline being improved and compared against."},{"cited_title":"Welch, D","cited_arxiv_id":null,"evidence_quote":"provides the diagonal-unitary circuit construction via Walsh expansion that the paper extends to two dimensions."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"defines the Walsh-Hadamard transform and basis functions used to expand and truncate the operators."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"gives the atomic potential parameters used to build the gold specimen test case."}],"review_version":1}