{"id":"dd929f5a-2eeb-4569-8987-0c9eeb30e1d9","arxiv_id":"2411.14638","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Low-overhead error detection without logical encoding improves long-range CNOT fidelity and establishes a record 75-qubit GHZ state with genuine multipartite entanglement on IBM superconducting processors.","lead":"Q-CTRL researchers report that lightweight error-detection checks on unencoded qubits improve long-range CNOT fidelities past 85% over up to 40 lattice sites and enable a 75-qubit GHZ state with verified entanglement, both on IBM superconducting hardware. The results argue that small QEC-style primitives can outperform alternative error-reduction strategies on current devices.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 75-qubit GME record rests on MQC fidelity >0.5, but the paper's own 80Q data (F=0.511, no GME claimed) show this criterion is not conclusive; undiagnosed phase errors can make F_mqc exceed the true fidelity.","rationale":"The paper's technical contributions—the unitary entangle-disentangle protocol for long-range CNOT gates and the sparse parity-check generation of GHZ states—are described in sufficient detail, and the experimental comparisons to the measurement-based protocol of Bäumer et al. are direct and fair. The protocol-level error detection is a legitimate QEC primitive, and the overhead analysis is clearly presented. The single most load-bearing point, however, is the certification of the headline record: a 75-qubit GHZ state with genuine multipartite entanglement. That claim depends entirely on the assertion that MQC fidelity F_mqc > 0.5 suffices. The threshold in Ref. [64] is for the true quantum state fidelity. The MQC estimate is only equal to the true fidelity when the off-diagonal coherence is real and positive; phase errors break this equality, making F_mqc an upper bound rather than a certified fidelity. The paper inadvertently demonstrates this problem in its own App. C: the 80Q state has F_mqc = 0.511(2), yet the authors decline to claim GME because the interference pattern is nearly entirely damped. If F_mqc > 0.5 were sufficient, they would have to claim GME for 80Q as well. Their reluctance shows that the criterion is not being applied consistently, and the 75Q record claim inherits the same unsupported assumption with a smaller margin (F_mqc > 0.55). The reader's conditional verdict already identifies this as the weakest assumption; my stress test confirms and sharpens it. The appropriate response is not to reject the paper's experimental work, but to require either a phase-resolved fidelity estimate or a direct stabilizer-based lower bound on the true fidelity before the GME record is accepted. Hence the verdict remains CONDITIONAL, and I recommend no change to the reader's assessment.","tokens_in":24633,"tokens_out":8577,"duration_ms":84523,"concrete_test":"Re-analyze the raw MQC parity-oscillation data for the 75-qubit run (Fig. 13): compute the complex Fourier coefficient at mode n without discarding the imaginary/sine quadrature. Estimate the off-diagonal phase theta and set a lower bound F_true = (P + 2|rho_01| cos(theta))/2 using the reported population P. If this lower bound is not > 0.5 (with uncertainty), the genuine-multipartite-entanglement claim fails. If raw files are unavailable, rerun a 75Q experiment measuring the X^⊗75 stabilizer directly (or parity oscillations at phi and phi + pi/n) to determine the real part of the GHZ coherence.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing assumption is that MQC fidelity F_mqc > 0.5 certifies genuine multipartite entanglement (Sec. III.A). Reference [64]'s threshold applies to the true state fidelity, i.e., (P + 2 Re(rho_01))/2 for the GHZ off-diagonal rho_01. The MQC estimate F_mqc = (P + C)/2 equals the true fidelity only if the off-diagonal coherence is real and positive. Systematic phase errors rotate rho_01; they do not necessarily reduce the measured parity-oscillation amplitude C, so F_mqc can exceed the true fidelity. The paper's own control is revealing: App. C/Fig. 14 reports an 80-qubit state with F_mqc = 0.511(2) (C=0.064, P=0.958) and states 'we do not claim to have prepared a state exhibiting genuine multipartite entanglement, despite its MQC fidelity at F=0.511(2).' This is a direct admission that F_mqc > 0.5 is not treated as sufficient. Nevertheless, the 75-qubit record claim uses exactly this criterion, with a margin of only ~0.05 over threshold and a substantially dephased coherence (Fig. 4(e)). Without a phase-resolved measurement or a lower bound on Re(rho_01), the 'largest reported GHZ state exhibiting GME' claim is unsupported. Secondary overclaims ('relative to any alternative error-reduction strategy') are also unsupported, but the GME record is the central falsifiable claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports two experimental demonstrations of low-overhead error detection ('QEC primitives') on IBM superconducting processors. In the first part, a 'unitary entangle-disentangle' protocol implements a long-range CNOT by unitarily preparing a linear-chain GHZ state along the path between control and target, then unitarily disentangling all but two of the intermediate qubits into a Bell pair; the disentangled qubits are measured as error-detection flags at no additional ancilla cost. The authors report average gate fidelities above 85% for teleportation distances up to 40 qubits, consistently outperforming the measurement-based protocol of Bäumer et al. on the same device, in both shot-by-shot and readout-mitigated post-processing modes. In the second part, following Mooney et al., the paper prepares GHZ states with sparse type-0 parity-check flags, combined with a deterministic error-suppression pipeline, reporting MQC fidelities above 0.78 for up to 54 qubits and above 0.55 for 75 qubits, with a maximum discarded fraction of 78% at 75 qubits. The paper claims that the 75-qubit state exhibits genuine multipartite entanglement (GME) and is the largest reported to date, and it argues that the combination of error suppression and physical-level error detection offers a favorable overhead trade-off relative to full logical encoding.","tokens_in":24849,"tokens_out":27478,"duration_ms":234857,"significance":"If upheld, the results are significant: the flag-based long-range CNOT design is elegant, the error-detection capability of the disentangled qubits costs no additional ancillas, and the sparse parity-check layout adds only one two-qubit-gate layer (43 vs. 42) to the 75-qubit GHZ circuit. The experimental methodology is careful: both CNOT protocols were run consecutively on ibm_fez to control drift, error bars are bootstrapped, the error-suppression pipeline is applied symmetrically to both arms, and the paper reports honest negative results (type-1 flags are not beneficial; error detection alone does not suffice). The overhead accounting (Table I; 27-qubit vs. encoded 24-qubit comparisons) is transparent. The two load-bearing weaknesses are the GME certification at 75 qubits, which rests on an MQC fidelity threshold that is not rigorously established as presented and is contradicted in practice by the paper's own 80-qubit control, and the 'largest reported to date' claim, which appears to conflict with a larger certified GHZ state in a cited reference.","major_comments":[{"comment":"The GME claim for the 75-qubit state rests on the statement in Sec. III.A that 'it is sufficient to show MQC fidelity F > 0.5 [64]'. The threshold of Ref. [64] applies to the true state fidelity, F_true = (P + 2 Re ρ_01)/2 with ρ_01 the GHZ off-diagonal element, whereas the measured quantity is F_mqc = (P + C)/2 with C = 2√I_n (Eq. (4)). F_mqc equals F_true only when the coherence is real and positive; a systematic phase error in ρ_01 makes F_mqc an upper bound on the true fidelity, so F_mqc > 0.5 does not by itself certify GME. This is not academic: App. C (Fig. 14) reports an 80-qubit state with F_mqc = 0.511(2) (C = 0.064(3), P = 0.958(2)), statistically well above 0.5, and explicitly declines to claim GME 'despite its MQC fidelity at F = 0.511(2)'. The 75-qubit state has similar qualitative features — F ≈ 0.55 with strongly suppressed coherence (Fig. 4(e)) and heavily damped parity oscillations (Fig. 13) — and its margin over threshold is only about 0.05. The authors should report the fitted phase of the parity oscillation and give a rigorous lower bound on Re ρ_01 satisfying Re ρ_01 > (1 − P)/2, or certify GME by an independent witness, or downgrade the 75-qubit claim.","section":"Sec. III.A, Eq. (4); App. C, Figs. 13-14"},{"comment":"The abstract and Sec. III.B describe the 75-qubit GHZ state as 'the largest reported to date' and 'a record in the published literature'. This appears to conflict with Ref. [58] (Moses et al., Phys. Rev. X 13 (2023)), which reports the preparation of a 92-qubit GHZ state with a quoted fidelity above the 0.5 GME threshold (approximately 0.75). If that fidelity estimate is accepted, the record claim is incorrect; if the authors believe the certification in Ref. [58] does not meet the standard applied here, they should say so explicitly and justify the difference. Because the record claim is a headline result, a reconciliation with all larger certified GHZ states in the cited literature is required.","section":"Abstract; Sec. III.B"}],"minor_comments":[{"comment":"The abstract claims advantages 'relative to any alternative error-reduction strategy', but the experiments benchmark against one specific alternative (the measurement-based protocol of Ref. [36]) and the authors' own no-error-detection and no-error-suppression baselines; the wording should be tempered to reflect the strategies actually tested.","section":"Abstract; Sec. I"},{"comment":"The reported gate fidelities are conditional on discarding shots requiring feedforward, a substantial fraction of shots, in addition to the error-detection discards shown in Fig. 10; reporting the total acceptance probability and the fidelity at each post-selection stage would make the 'computational gain' claim measurable.","section":"Fig. 3 caption; Sec. II.B"},{"comment":"For each flag count l, the presented MQC fidelity is the maximum over the C(f_max, l) subsets of flags; this makes the plotted curves an upper envelope, and the optimism of this selection should be acknowledged or quantified (for example, by reporting the mean or median over subsets).","section":"Sec. III.B"},{"comment":"The reason for not claiming GME for the 80-qubit state ('the interference pattern is nearly entirely damped') is qualitative; given that F_mqc = 0.511(2) is statistically above 0.5, the authors should state the quantitative criterion that separates the 75-qubit GME claim from the 80-qubit non-claim.","section":"App. C, Fig. 14"},{"comment":"The statement that the discarded fraction 'grows no higher than 78%' should be scoped to the sizes for which GME is claimed, since the 80-qubit control in App. C has a discard fraction of about 98%.","section":"Abstract; Sec. III.B"}],"recommendation":"major_revision","confidential_remarks":"The two load-bearing issues identified in the report both concern the headline claims. The GME certification question is fixable with additional analysis of the existing parity-oscillation data (fitted phase and a confidence interval on Re rho_01) or a supplementary witness, and the record claim requires explicit reconciliation with Ref. [58]. I do not see grounds for rejection: the experimental core, the CNOT comparison, and the overhead analysis are sound, but the claims as currently worded exceed what the data certify."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The real news here is the unitary entangle-disentangle protocol for long-range CNOTs. It is a genuinely new circuit, the disentangled qubits double as error-detection flags without extra ancillas, and the experimental comparison against Bäumer et al. is run in a way that makes the improvement credible: same device, consecutive runs, same error-suppression baseline, Monte Carlo process certification. The 85%+ fidelity across 40 sites is a solid result, and the appendix material on generalizing to fan-out and Toffoli gates, plus the interpolation to measurement-based circuits, is thoughtful. The GHZ half is more of a combination: sparse parity checks from Mooney et al. plus the Q-CTRL error-suppression pipeline, applied carefully, with honest reporting of discard rates and type-1 flag overhead. That is useful but not as new as the CNOT part.\n\nThe soft spot is exactly where the stress-test note points. The \"largest reported GME\" claim rests on MQC fidelity F_mqc = (P+C)/2 > 0.5. The threshold in Ref. [64] is for true state fidelity, and F_mqc equals true fidelity only when the GHZ off-diagonal coherence is real and positive. Systematic phase errors can make C larger than 2Re(rho_01), so F_mqc can exceed 0.5 while the true fidelity is below. The authors more or less concede this themselves in App. C: the 80-qubit state has F_mqc = 0.511(2), C=0.064, P=0.958, and they explicitly do not claim GME. That makes the 75-qubit claim with F_mqc about 0.55 and heavily damped coherence hard to accept as a record. They need a lower bound on the real part of the off-diagonal element, or a phase-resolved measurement, or they need to soften the claim.\n\nSecondary issue: the abstract says \"relative to any alternative error-reduction strategy,\" but the experiments only compare to the measurement-based protocol of Bäumer et al. and to no-error-detection versions of their own circuits. That is an overclaim. Citation practice is fine; they credit Mooney and Bäumer and the comparison is fair.\n\nBottom line: the CNOT protocol is ready for serious peer review after minor revision, and the GHZ claim needs either a rigorous fidelity bound or a downgrade. The paper is worth reading for the circuit design and the overhead analysis. I would send it out; a good referee will catch the GME issue but there is real content here.","headline":"The long-range CNOT protocol is a real contribution and the comparison is credible; the 75-qubit GME record claim is not supported because MQC fidelity above 0.5 does not certify genuine multipartite entanglement when phase errors are present, as the authors' own 80-qubit control shows.","tokens_in":25504,"tokens_out":5603,"would_cite":true,"duration_ms":55674,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Applying quantum error-correction primitives to unencoded qubits yields net gains on current superconducting processors: over 85% fidelity for CNOT gates across up to 40 sites, and genuine multipartite entanglement in a 75-qubit GHZ state.","keywords":["quantum error correction primitives","error detection without encoding","unitary entangle-disentangle protocol","long-range CNOT gate","GHZ state preparation","genuine multipartite entanglement","multiple-quantum coherence fidelity","superconducting processors"],"falsifier":"Run the 75-qubit preparation and measure the imaginary part of the n-th Fourier mode of the MQC parity oscillation (the term that is exactly zero for a real, positive GHZ coherence). If that imaginary part is statistically nonzero, the reported fidelity is not the true fidelity and the genuine-multipartite-entanglement claim is unsupported; equivalently, a simulation with a known systematic phase error that pushes F=(C+P)/2 above 0.5 while the true fidelity stays below 0.5 would refute the certification step.","tokens_in":24356,"feed_emoji":"⚛️","tokens_out":8988,"duration_ms":78697,"temperature":0.7,"pith_summary":"This paper argues that the ingredients of quantum error correction can deliver net computational gains on today's small superconducting processors when used as error-detection primitives on unencoded qubits, with only a modest overhead. It introduces a unitary entangle-disentangle protocol for long-range CNOT gates: a linear-chain GHZ state is created, reduced to a Bell pair, and used to teleport the gate, while the disentangled intermediate qubits double as error-detection flags. In experiments, this protocol sustains average gate fidelities above 85% across up to 40 lattice sites and consistently outperforms the best measurement-based alternative without extra ancillas. The same error-detection idea, applied as sparse stabilizer parity checks with at most nine flag qubits, produces a 75-qubit GHZ state whose reported fidelity exceeds the 0.5 threshold for genuine multipartite entanglement, which the paper identifies as the largest such state reported. If these claims hold, near-term users can gain from QEC building blocks before full fault-tolerant encoding becomes practical.","feed_headline":"Error flags, no encoding, lift 75-qubit entanglement","feed_subtitle":"Sparse parity checks on unencoded qubits beat the best measurement-based CNOTs and set a new GHZ record.","key_machinery":"The load-bearing object is the GHZ state as a reusable entanglement resource, together with the unitary entangle-disentangle circuit that prepares and reduces it. Step A grows a linear-chain GHZ state along the shortest path between control and target; step B applies local CNOT gates that rotate the n-qubit GHZ state into a small residual GHZ state times $|0\\rangle^{\\otimes(n-3)}$, so that the disentangled qubits are guaranteed to read all-0 in the absence of error; step C projectively measures the one remaining 'root' qubit to fix the Bell-pair parity; step D completes the teleported CNOT by local operations and classical communication. For direct GHZ generation, the machinery is the stabilizer structure of the GHZ state: ancilla 'type-0' flag qubits measure selected ZZ parity checks without any SWAP overhead on a heavy-hex topology, and any odd-parity outcome marks the shot as failed. The two uses share one principle: convert the errors a device is most prone to into classical flag outcomes, then discard the flagged shots.","core_discovery":"On its own terms, the paper establishes that error detection without logical encoding, implemented inside the circuits themselves, improves performance beyond what the same hardware achieves with the best alternative strategies. The long-range CNOT result rests on a protocol where every disentangled GHZ qubit ideally ends in |0>, so reading those qubits flags any bit-flip or amplitude-damping error that occurred during gate teleportation; with error detection enabled, effective gate infidelity falls by about half, and the protocol beats the measurement-based benchmark at every tested distance up to 40 qubits. The GHZ result rests on sparse parity checks, where a small set of ancilla flags non-destructively measures ZZ stabilizers of the GHZ state and shots with odd parity are discarded; combined with deterministic error suppression, this yields MQC fidelities above 0.78 up to 54 qubits and above 0.55 at 75 qubits, at discard fractions no higher than 78%.","pith_inferences":["The overhead accounting suggests a design rule the paper only gestures at: on near-term devices, the most cost-effective QEC primitive is the one that turns the device's dominant error channel into a detectable flag, not the one offering the most complete logical protection. A testable corollary is that the same entangle-disentangle construction should improve other controlled-unitary gates, with ","The 75-qubit record rests on identifying the MQC fidelity estimate with the true state fidelity. If systematic phase errors make the GHZ coherence complex rather than real and positive, the reported F=(C+P)/2 overestimates the true fidelity; a direct measurement of the imaginary part of the n-th Fourier mode of the parity oscillation would settle whether the entanglement claim survives.","The measured saturation of fidelity with flag count (for example, the 45-qubit case saturating around three flags) indicates residual error is dominated by Z-dephasing, which ZZ parity checks cannot see. An extension that adds refocusing or dynamical-decoupling layers and checks whether the saturation point moves would separate dephasing from other error sources in this preparation."],"forward_implications":["Long-range CNOTs with over 85% average fidelity across up to 40 qubits become practical primitives for quantum Fourier transforms, fermionic simulations, and non-local qLDPC codes on current devices.","Because the entangle-disentangle construction trades readout measurements for two-qubit gates, it is better matched to present hardware, where readout errors typically exceed two-qubit-gate errors by a factor of three to six.","A 75-qubit GHZ state prepared with 9 flag ancillas is reported to exceed the 0.5 fidelity threshold for genuine multipartite entanglement with a 78% worst-case discard fraction, and the paper identifies it as the largest such state reported to date.","The same circuit family interpolates between the linear-depth unitary limit and the constant-depth measurement-based limit, so future implementations can be tuned to a device's specific balance of gate and readout errors."],"supporting_citations":[{"why":"Supplies the measurement-based long-range CNOT protocol that the unitary entangle-disentangle protocol is benchmarked against and outperforms.","marker":"[36]"},{"why":"Introduces the unitary GHZ preparation with sparse parity checks that the paper adopts and extends with error detection.","marker":"[35]"},{"why":"Provides the error-suppression and readout error-mitigation pipeline used as the high-fidelity baseline in all experiments.","marker":"[19]"},{"why":"Supplies the deterministic error-suppression technique combined with the parity-check error detection in the GHZ experiments.","marker":"[25]"},{"why":"Gives the Monte Carlo process certification method used to estimate the long-range CNOT gate fidelity.","marker":"[50]"},{"why":"Provides the practical quantum Monte Carlo state certification procedure that the fidelity estimation follows.","marker":"[51]"},{"why":"Establishes the F>0.5 criterion used to certify genuine multipartite entanglement from MQC fidelity.","marker":"[64]"}],"fun_headline_variants":["Error detection, no encoding, boosts CNOT and GHZ records","Unencoded QEC flags beat best alternatives on 75-qubit GHZ","Sparse parity checks set GHZ record with low discard","QEC primitives without encoding give net gains on 40-site CNOTs","Error-detecting CNOTs and 75-qubit GHZ from modest overhead"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The headline 75-qubit entanglement record assumes that the measured multiple-quantum-coherence figure F=(C+P)/2 equals the true fidelity of the prepared state; that equality is guaranteed only when the GHZ coherence is real and positive, and a systematic phase error would make the reported number an upper bound rather than a certification.","fun_headline_variants_meta":{"raw":{"variants":["Error detection, no encoding, boosts CNOT and GHZ records","Unencoded QEC flags beat best alternatives on 75-qubit GHZ","Sparse parity checks set GHZ record with low discard","QEC primitives without encoding give net gains on 40-site CNOTs","Error-detecting CNOTs and 75-qubit GHZ from modest overhead"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000459,"raw_usage":{"total_tokens":2343,"prompt_tokens":1032,"completion_tokens":1311,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":648,"completion_tokens_details":{"reasoning_tokens":1214}},"tokens_in":648,"tokens_out":1311,"duration_ms":8570,"temperature":1.0,"reasoning_tokens":1214,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:04:22.511313+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the 75-qubit preparation and measure the imaginary part of the n-th Fourier mode of the MQC parity oscillation (the term that is exactly zero for a real, positive GHZ coherence). If that imaginary part is statistically nonzero, the reported fidelity is not the true fidelity and the genuine-multipartite-entanglement claim is unsupported; equivalently, a simulation with a known systematic phase error that pushes F=(C+P)/2 above 0.5 while the true fidelity stays below 0.5 would refute the certification step.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the measurement-based long-range CNOT protocol that the unitary entangle-disentangle protocol is benchmarked against and outperforms."},{"cited_title":"Eisert, K","cited_arxiv_id":null,"evidence_quote":"Provides the practical quantum Monte Carlo state certification procedure that the fidelity estimation follows."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the F>0.5 criterion used to certify genuine multipartite entanglement from MQC fidelity."}],"review_version":1}