{"id":"02f34c8e-2b2c-4028-8ea4-cedf561ec998","arxiv_id":"2512.17779","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"An ion-trap quantum computer correctly compared 9-bit integers with 95% output success and 69% ancilla-inclusive success.","lead":"A team ran a quantum comparator circuit on RIKEN's 20-qubit trapped-ion computer and measured high success: 95% of runs gave the correct comparison at 9-bit numbers under the standard criterion, and 69% when the ancilla qubit also had to be correct. It is a benchmark showing that a known arithmetic primitive can be executed at larger scale on current hardware than previous experiments.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Measured computational-basis statistics cannot certify coherent quantum comparison; the data are equally explained by a classical noisy reversible circuit, so the 'quantum' part of the central claim is unsupported.","rationale":"The reader's weakest assumption is correct and is the load-bearing point. The paper's strongest claim is not just that a specific circuit runs on hardware, but that it constitutes a quantum arithmetic primitive. The currently reported evidence—success counts from computational-basis measurements—is compatible with purely classical noise models. The manuscript's own note about phase coherence not surviving strengthens the concern. A single coherence-sensitive measurement (e.g., tomography on n=2) would settle this. No other issue is as central: the 100-shot statistics, the 69%/29% typo, and reference misattribution are real but do not affect the logical core as much as the missing coherence evidence. Since the identified concern is precisely why the reader issued CONDITIONAL, I recommend no change to the verdict.","tokens_in":6399,"tokens_out":7522,"duration_ms":92728,"concrete_test":"For n=2, prepare the input superposition (|00_a>+|11_a>)/√2 ⊗ |01_b> with the result qubit in |0>, run the comparator, and perform quantum state tomography on the final 6-qubit state (including the ancilla). Compute fidelity with the ideal entangled state (|00_a,01_b,0_anc,1_c>+|11_a,01_b,0_anc,0_c>)/√2 and inspect the off-diagonal density-matrix elements. If the fidelity is ≈0.5 or off-diagonals vanish, the device is dephased and the data support only a classical truth-table demonstration. Repeating at n=9 is unnecessary; n=2 suffices to test coherence of the primitive.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim, 'demonstrate reliable quantum comparison at scales far beyond those previously achieved experimentally,' has two parts: (i) the comparator truth table is reproduced with high fidelity, and (ii) the operation is a coherent quantum process. The data support (i) but not (ii). The experiment (Secs. III, IV) prepares a uniform superposition of all input basis states, applies a circuit that permutes computational basis states, and measures every qubit in the computational basis. For any such basis-permuting unitary, the final measurement statistics are exactly the statistics of a classical reversible circuit acting on uniformly sampled inputs with bit-flip noise; off-diagonal coherences never enter. The paper itself concedes this in Sec. III: 'the phase coherence, even if created in the first stage, might not survive in later stages. The relevant noises are rather the bit-flip (X) errors.' Because Z errors do not affect computational-basis statistics, the reported success rates (95% conventional, 69% ancilla-inclusive at n=9) cannot distinguish a coherent quantum comparator from an incoherent classical simulation of the same circuit. Thus the 'quantum comparison' claim, and the comparison to prior quantum arithmetic experiments, rests entirely on an unverified coherence assumption.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports execution of a reversible comparator circuit for n-bit integers (n = 3, 5, 7, 9) on the RIKEN Reimei trapped-ion quantum computer. The circuit prepares a uniform superposition over all 2n-bit input pairs with Hadamard gates, applies a ripple-carry-style comparator, and measures all qubits in the computational basis. Reported success probabilities are about 98%, 97%, 97%, and 95% under a conventional output-only criterion, and about 95%, 92%, 89%, and 69% under a stricter ancilla-inclusive criterion, with no postselection or error mitigation. The paper interprets these results as demonstrating reliable quantum comparison at scales beyond previous experiments.","tokens_in":6690,"tokens_out":10096,"duration_ms":98713,"significance":"The experiment addresses a meaningful benchmark: comparators are a basic arithmetic primitive, and linear-depth ripple circuits stress multi-qubit operations. The stricter ancilla-inclusive criterion is a useful addition, and the absence of postselection/error mitigation strengthens the practical relevance of the raw success rates. However, the central scientific claim—that the data demonstrate a quantum comparator—is not supported by the measurements as presented. The data are consistent with a classical noisy reversible circuit executed on a quantum processor; the paper itself concedes that phase coherence was not verified. If the coherence gap is addressed, or the claims are appropriately reframed, the result would be a valuable system-level benchmark for arithmetic circuits on trapped-ion hardware.","major_comments":[{"comment":"The reported observable is a histogram of computational-basis outcomes after a unitary that permutes computational basis states. For such a protocol, the output distribution is exactly that of a classical reversible circuit acting on uniformly sampled inputs with bit-flip noise; off-diagonal coherences never appear in the measured statistics. The paper itself states in Sec. III that 'the phase coherence, even if created in the first stage, might not survive in later stages' and in Sec. VI that 'an estimation of the phase is anticipated.' No state tomography, entanglement witness, phase-sensitive interference experiment, or equivalent coherence certification is reported. Therefore the title/abstract claim of 'reliable quantum comparison' is not established by the data: what is demonstrated is correct comparator truth-table statistics on a quantum processor, not a coherent quantum comparis","section":"Sec. III (Eq. (4), Fig. 2); Sec. VI"},{"comment":"All quantitative claims rest on 100 shots per n, yet no confidence intervals or statistical uncertainties are reported. At n=9, for example, the 95% and 69% point estimates have approximate Wilson 95% intervals of (88.7%, 98.4%) and (59.6%, 77.7%) respectively. The separation from the 50% and 25% random baselines is large enough that the main qualitative conclusion is robust, but quantitative comparisons such as 'more than an order-of-magnitude reduction in failure probability' and 'higher than the random baseline by a factor of 69%/29%≈3' need error bars and corrected arithmetic: the ancilla-inclusive random baseline is 25%, so 69%/25% ≈ 2.8, not 69%/29%. Please report counts, confidence intervals, and the exact baseline calculation.","section":"Sec. IV (Figs. 3 and 4)"},{"comment":"The manuscript does not provide the concrete compiled circuits used for n=5, 7, and 9 (gate counts, native-gate decompositions, qubit layout), nor device calibration data such as two-qubit gate fidelities and readout error rates. Since the entire quantitative claim is about hardware execution, these are not optional details; for example, a readout error of a few percent on the result qubit would directly contribute to the reported 5% failure at n=9. Please include the compiled circuits in a permanent repository and report the relevant calibration/readout characterization.","section":"Secs. II–IV; Data Availability"}],"minor_comments":[{"comment":"The text says the comparator circuit is 'adapted from Ref. [21]', but Ref. [21] is Gouzien et al., a cat-code architecture paper, not a comparator construction. This appears to be a citation error; Ref. [11] (Cuccaro et al.) or another comparator reference is likely intended.","section":"Sec. II"},{"comment":"The caption reads 'introduce as a stricter criterion'; it should be 'introduced as a stricter criterion'.","section":"Fig. 4 caption"},{"comment":"The text says 'JPSJ KAKENHI'; this should likely be 'JSPS KAKENHI'.","section":"Acknowledgments"}],"recommendation":"major_revision","confidential_remarks":"The coherence objection is the main risk. If the authors can supply any phase-sensitive evidence, or alternatively reframe the claims to a computational-basis benchmark of a comparator circuit on a quantum processor, the paper could be publishable as an experimental benchmark. The missing gate-level and calibration data should also be addressed before acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The real content is an n=9, 20-qubit run of a standard Cuccaro-style comparator on RIKEN's trapped-ion machine, 100 shots per bit width, no error mitigation. That is genuinely new as an experimental scale for this primitive, and the ancilla-inclusive success category is a nice, honest addition: it separates borrow-ancilla errors from output errors and shows ancilla fidelity is the bottleneck. The paper deserves credit for reporting the strict criterion and for noting the phase-coherence limitation in Sec. III.\n\nThe soft spots are clear. First, the central claim \"demonstrate reliable quantum comparison\" is overstated. The experiment prepares a uniform superposition, applies a basis-permuting unitary, and measures in the computational basis. For such a circuit, the outcome statistics are exactly those of a classical reversible circuit with bit-flip noise; off-diagonal coherence never enters. The paper itself concedes this in Sec. III, then Sec. VI says \"an estimation of the phase is anticipated\" — so the quantum part of the demonstration is still missing. That does not ruin the value as a NISQ arithmetic benchmark, but the abstract should be reworded. Second, quantitative support is thin: 100 shots per n, no confidence intervals, no gate-level circuit data, no measurement-error characterization, and data only \"available upon request.\" These are fixable. Third, there is an internal typo in Sec. IV B: the random baseline for ancilla-inclusive success is 25%, but they write 69%/29%≈3; presumably they meant 25%. Minor. Fourth, the circuit attribution looks off: Fig. 1 is said to be adapted from Ref. [21], which is a repetition cat code architecture paper, not the comparator reference (Cuccaro et al. [11]). Likely a citation error, but it needs correction.\n\nMy take: the experimental result is probably sound as a benchmark — the success rates are far above random baselines and the trend with n is sensible. But the strong claim of \"quantum comparison far beyond prior\" is not supported by the data as presented. The paper is worth a serious referee; with revisions that add error bars or a coherence witness (or at least tone down the claim), it would be a useful benchmark note. Recommendation: send it to peer review, with a clear request to fix the quantum claim, the baseline typo, and the citation.","headline":"Useful experimental benchmark for NISQ arithmetic, but the “quantum” claim is thinner than the abstract suggests — the data certify a high-fidelity reversible circuit, not coherent comparison.","tokens_in":7161,"tokens_out":2020,"would_cite":true,"duration_ms":23034,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P68"],"pacs":["03.67.Lx"],"model":"deepseek-v4-flash","headline":"A trapped-ion quantum computer ran a 9-bit comparator—a circuit deciding whether one integer is smaller than another—with 95% success and no error mitigation.","keywords":["quantum comparator","quantum arithmetic","integer comparison","trapped-ion quantum computing","ripple-borrow circuit","ancilla errors","NISQ benchmark","unitary comparison"],"falsifier":"Prepare the input registers in the superposition (|0,1> + e^{iφ}|1,1>)/√2, run the comparator, and measure the output qubit in the X basis. The ideal unitary gives an X-measurement probability that oscillates with φ; observing a φ-independent 50/50 mixture would falsify the claim that the process is coherently quantum rather than a classical truth-table computation.","tokens_in":6343,"feed_emoji":"⚛️","tokens_out":7719,"duration_ms":76001,"temperature":0.7,"pith_summary":"This paper reports the experimental implementation of a quantum comparator—a reversible circuit that decides whether one integer is less than another—on a trapped-ion quantum computer. The authors show that for bit widths up to n=9, the circuit returns the correct comparison outcome in about 95% of runs under the conventional output-only criterion, and in 69% of runs when the ancilla qubit must also be correct. No postselection or error mitigation is used. The result matters because comparison is a primitive in modular-arithmetic algorithms such as integer factoring and minimum-finding, and prior experimental comparators reached only n=2 with classical inputs. The paper argues that all-to-all connectivity and high two-qubit gate fidelities make arithmetic circuits of this depth practical on current hardware.","feed_headline":"9-bit quantum comparison runs at 95% success on ion-trap device","feed_subtitle":"Trapped-ion qubits compare 9-bit integers with 95% accuracy; ancilla flips are the main error.","key_machinery":"The load-bearing object is the ripple-borrow comparator, a reversible subtraction circuit that extracts the sign of a−b without storing the full difference. A single ancilla qubit carries the borrow bit, updated cell by cell from the least to most significant pair of input bits using majority/unmajority gates composed of CNOT and doubly-controlled-NOT operations. The comparator's depth grows linearly with n, which is precisely what makes it a demanding benchmark: errors in the ancilla accumulate across the whole cascade. The hardware feature that makes the depth tolerable is all-to-all qubit connectivity plus high two-qubit gate fidelities.","core_discovery":"The paper's central claim is that a quantum comparator—a unitarily implemented Boolean function f(a,b) that returns 1 exactly when the n-bit integer a is less than b—has been executed on real trapped-ion hardware for bit widths n=3, 5, 7, and 9. The experiment prepares a uniform superposition of all ordered integer pairs using single-qubit superposition gates, runs a ripple-borrow comparator constructed from CNOT and doubly-controlled-NOT gates with one ancilla, then measures every qubit in the computational basis. Under the conventional output-only criterion the success probabilities are about 98%, 97%, 97%, and 95%; under the stricter criterion requiring the ancilla to be correct as well,","pith_inferences":["The reported statistics are exactly those of a classical reversible truth table sampled uniformly, so a reader should not infer evidence of superposition or entanglement from them; an interference measurement would be needed to certify coherence.","If the goal is to benchmark hardware for arithmetic, the comparator could serve as a scalable stress test for decoherence along a long ancilla chain—analogous to randomized benchmarking but at algorithmic depth.","A concrete next experiment is to vary the phase of a superposition of two input pairs and look for interference in the output register; the paper's own Z-error commutativity analysis suggests such a signal may be weak or absent.","The success rates suggest that with an error-corrected or logical ancilla, the stricter criterion could approach the conventional one, making the comparator attractive as a building block for modular-exponentiation circuits."],"forward_implications":["Quantum comparison is no longer confined to n=2 with classical inputs; the same circuit family now runs with n=9 on real hardware.","The output-only success rate at n=9 (95%) is more than an order of magnitude above random (5% failure vs 50% failure), so the comparator is usable as a standalone arithmetic block.","The ancilla-inclusive success rate (69% at n=9) quantifies the cost of using the comparator inside a larger circuit where the borrow ancilla must be restored.","Ancilla-only errors are the dominant failure mode at n=7 and n=9, so future effort should target the ancilla's fidelity rather than the logical comparison path.","No error mitigation is needed, meaning the measured rates are raw hardware characteristics and can be improved by hardware advances alone."],"fun_headline_variants":["Ion-trap quantum comparator: 95% accurate on 9-bit pairs","Quantum comparator on ion-trap device: 95% success at 9 bits","Trapped-ion quantum comparator reaches 95% for 9-bit integers","9-bit quantum comparison on trapped ions: 95% success","Ion-trap quantum comparator proves 95% on 9-bit data"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing assumption is that measuring all qubits in the computational basis and seeing the correct truth table after the unitary is enough to establish a 'quantum' comparison; no tomography, entanglement witness, or interference experiment is performed, so if coherent superposition is required for the claim, the data do not establish it.","fun_headline_variants_meta":{"raw":{"variants":["Ion-trap quantum comparator: 95% accurate on 9-bit pairs","Quantum comparator on ion-trap device: 95% success at 9 bits","Trapped-ion quantum comparator reaches 95% for 9-bit integers","9-bit quantum comparison on trapped ions: 95% success","Ion-trap quantum comparator proves 95% on 9-bit data"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000222,"raw_usage":{"total_tokens":1258,"prompt_tokens":679,"completion_tokens":579,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":423,"completion_tokens_details":{"reasoning_tokens":480}},"tokens_in":423,"tokens_out":579,"duration_ms":6053,"temperature":1.0,"reasoning_tokens":480,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T15:09:21.584744+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Prepare the input registers in the superposition (|0,1> + e^{iφ}|1,1>)/√2, run the comparator, and measure the output qubit in the X basis. The ideal unitary gives an X-measurement probability that oscillates with φ; observing a φ-independent 50/50 mixture would falsify the claim that the process is coherently quantum rather than a classical truth-table computation.","supporting_citations":[],"review_version":1}