{"id":"19a2e31a-91e1-4eca-96dd-d785309f5202","arxiv_id":"2506.22600","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A trapped-ion experiment realizes measurement-free fault-tolerant logical teleportation, a universal logical gate set, and a three-logical-qubit Grover search, albeit with success below the classical baseline.","lead":"Researchers demonstrated a toolbox of fault-tolerant logical operations on a trapped-ion processor that avoids mid-circuit measurements, using coherent feedback and qubit resets instead. If the approach pans out, it could simplify quantum error correction on platforms where measurements are slow or error-prone.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"FT claim rests on a local-noise Monte Carlo model that omits the dominant global dephasing; App. G shows >14% fidelity mismatch, so the experimental FT demonstration is not yet supported by the reported noise model.","rationale":"The reader's weakest assumption is also my pick. The proof of fault tolerance is not a theorem or machine-checked proof; it is a Monte Carlo scaling plot under a specific local-noise model. Appendices G and H concede that the model's dominant dephasing component is not the device's actual dominant dephasing: global magnetic-field fluctuations produce correlated phase errors, and the observed fidelities for |000>_L are more than 14% below simulation. This is not merely an engineering detail. Fault tolerance is a property of a circuit with respect to a fault model; if the fault model omits the dominant physical fault, the scaling plot cannot validate the experimental claim. I considered whether the more serious issue is that the logical Grover success probability (0.40) is below the classical baseline (0.46), but the paper states this and argues projected improvements; a demonstration can still be a valid proof-of-principle. I also considered the heavy postselection (acceptance rates 0.07-0.5), but postselection is standard in error-detection-based FT, and the authors disclose it. The non-transversal H injection is the most complex gadget, but its FT is supported by the same Monte Carlo method; the missing global dephasing is the weakest point. The proposed test is feasible with the published code and data and would settle whether the concern is quantitative only or structural.","tokens_in":24129,"tokens_out":13140,"duration_ms":155972,"concrete_test":"Augment the App. C Monte Carlo simulator with a global dephasing channel: after each gate/idle interval of duration t, apply exp(-i(θ/2)Σ_i Z_i) with θ drawn from a Gaussian of zero mean and variance calibrated to the measured magnetic-field-noise spectrum (or chosen so the simulated |000>_L fidelity at λ=1 matches the experimental value). Re-run the H_L injection circuit (App. Fig. 8) and the λ-scaling of Fig. 12 for all four input states, keeping the same postselection rules. If the |000>_L and |+00>_L fidelities and acceptance rates in App. G/Fig. 11 are reproduced and the λ-scaling remains quadratic, the concern is resolved; if a linear term appears or the scaling flattens, the experimental FT claim must be restricted to the local-noise model.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim has two load-bearing components: the circuits are fault-tolerant, and this is experimentally demonstrated. The FT evidence is Monte Carlo scaling (App. C/Fig. 12) under independent depolarizing channels plus local idling dephasing (E_idle, Eq. C2). Appendix G explicitly reports that this model misses the experiment by more than 14% for |000>_L and attributes the gap to global dephasing — a single correlated magnetic-field fluctuation acting on all physical qubits. For |000>_L, global dephasing decays coherences eight times faster than local dephasing (App. G), and App. H estimates dephasing contributes almost two-thirds of the logical error rate. Because the actual dominant error is a collective process not represented by the local-fault model, the quadratic scaling of logical infidelity vs λ in Fig. 12 does not establish that the demonstrated circuits are fault-tolerant under the device's true noise. If global dephasing produces undetected logical errors, the postselected fidelities mainly reflect rejection of bad runs, and the projected Grover gains in App. J inherit the same unvalidated noise model. The theoretical construction may still be FT for local stochastic faults; the experimental demonstration part of the claim is not supported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes and experimentally demonstrates a measurement-free toolbox for fault-tolerant logical operations on a trapped-ion processor. The central results are: (i) modular logical state teleportation between two [[4,1,2]]-code blocks using coherent feedback instead of mid-circuit measurements; (ii) a fault-tolerant universal gate set {H_L, CNOT_L, CCZ_L} on the [[8,3,2]] code, with H_L realized by state injection using an auxiliary [[4,2,2]] code; and (iii) a three-logical-qubit Grover search encoded in the [[8,3,2]] code. The experimental section reports logical fidelities for state preparation, teleportation, H_L injection, and the transversal CCZ_L gate, together with Monte Carlo simulations based on local depolarizing and dephasing noise. The fault-tolerance claims are primarily supported by simulated quadratic scaling of logical infidelity under a scaled local-noise model.","tokens_in":24398,"tokens_out":4018,"duration_ms":50373,"significance":"If the central claim survives scrutiny, this would be a notable first: a small-scale fault-tolerant universal gate set executed without mid-circuit measurements, plus the first fault-tolerant logical Grover search on three logical qubits. The paper is generally careful about circuit design, makes the explicit circuits and simulation code available, and uses independently measured error parameters rather than fitting noise rates to the target results. The theoretical construction appears plausible for local stochastic faults. However, the experimental demonstration of fault tolerance is not yet supported, because the noise model used to establish fault tolerance omits the experimentally dominant global dephasing channel, as the authors themselves document in Appendix G. The projected near-term advantage in Appendix J inherits the same limitation.","major_comments":[{"comment":"The central evidence that the implemented circuits are fault tolerant is the quadratic scaling of logical infidelity versus the scaled local-noise parameter lambda in Fig. 12, obtained from the Monte Carlo model of Appendix C. Appendix G reports that this model misses the experimental fidelity for |000>_L and |+00>_L by more than 14%, and attributes the discrepancy to global dephasing acting collectively on all physical qubits. For |000>_L, Appendix G estimates that global dephasing decays the relevant coherences eight times faster than local dephasing. This missing channel is load-bearing: a collective Z-type error can produce a logical error that the distance-2 stabilizer checks do not detect, so the quadratic scaling under local noise does not demonstrate fault tolerance under the device's actual dominant noise. Please augment the simulations with a global dephasing channel (e.g., a collective Z rotation with randomly fluctuating angle) and report the resulting logical infidelities, acceptance rates, and scaling; if the fault-tolerance claim remains valid under this augmented model, that should be shown explicitly.","section":"Appendix G and Fig. 12"},{"comment":"The error budget in Appendix H concludes that dephasing contributes almost two-thirds of the logical error rate, and the projected improvements in Appendix J (p2=0.015 or T2=100 ms) are computed with the same local-dephasing model that Appendix G shows to be incomplete. Because global dephasing accelerates logical dephasing for states such as |000>_L, the projected success probabilities of 0.52 and 0.67, and the statement that a regime outperforming the classical strategy 'is reachable today', are not yet supported. Please recompute the projections with the global-dephasing model and state whether the conclusions survive.","section":"Appendix H and Appendix J"},{"comment":"The H_L gate on the [[8,3,2]] code relies on a non-transversal inter-block CNOT gate between the [[8,3,2]] block and the [[4,2,2]] auxiliary block, and the fault-tolerance of this gadget is argued by stating that any single fault remains detectable. The provided evidence is again the Monte Carlo scaling under local noise. Given the global-dephasing issue above, please provide a direct fault-propagation analysis for the inter-block CNOT gadget under correlated Z noise, or clearly restrict the fault-tolerance claim to local stochastic faults.","section":"Section D and Fig. 8"}],"minor_comments":[{"comment":"The phrase 'fault-tolerant quantum computation' in the abstract and outlook should be qualified, since the demonstrations use a distance-2 error-detecting code with final postselection; the paper should state more explicitly that this is error-detecting fault-tolerance with postselection, not full error-correcting QEC without postselection.","section":"Abstract and Section F"},{"comment":"The notation for the global-dephasing decay prefactor is compressed and could confuse readers; a short derivation or an explicit reference to the formula in Ref. [86] would improve clarity.","section":"Appendix G, Eq. (G1)"},{"comment":"The labels in Fig. 1b and 1e are small and some overlapping text makes the two protocols difficult to distinguish; enlarging the fonts and separating the panels would help.","section":"Figure 1"},{"comment":"Several central protocol references are to the authors' own prior work (e.g., Refs. [26, 35, 37, 39, 42, 44]); this is not inappropriate, but the introduction should also cite independent measurement-free QEC proposals where available, so that the novelty framing is not overly self-referential.","section":"Reference list"}],"recommendation":"major_revision","confidential_remarks":"The experimental work is substantial and the circuits are clearly presented, but the paper's headline claim depends on a noise model that the manuscript itself shows misses the dominant dephasing. I would not reject: the theoretical constructions appear sound for local stochastic faults, and the missing global-dephasing analysis is a well-posed addition rather than a fundamental flaw. However, the current version overstates the experimental demonstration of fault tolerance, and the revision should materially rework the FT evidence and the projections in Appendix J."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is the first experimental demonstration of a measurement-free universal set of fault-tolerant logical gates, and it is a genuinely useful piece of work. New here are the H_L injection circuit on [[8,3,2]] using an auxiliary [[4,2,2]] block, modular teleportation between [[4,1,2]] blocks without mid-circuit measurement, and the logical three-qubit Grover run. Circuits, data, and simulation code are public. They also deserve credit for flagging the mismatch themselves.\n\nMain soft spot: the fault-tolerance evidence is Monte Carlo scaling under independent depolarizing channels plus local idling dephasing, and Appendix G reports the same model misses the experimental fidelity for |000>_L by over 14%, attributing this to global dephasing. Appendix H estimates dephasing accounts for almost two-thirds of the logical error rate. So the quadratic scaling in Fig. 12 shows the circuits are FT for local stochastic faults; it does not show they are FT under the device's actual dominant noise. If global dephasing creates undetected correlated logical errors, the postselected fidelities mostly reflect discarded runs. The logical Grover success probability of 0.40(4) sits below the classical 0.46, and the projected improvements inherit the same unvalidated noise model. That makes 'demonstrate measurement-free fault-tolerant universal quantum computation' stronger than the data currently support.\n\nAlso, the headline fidelities are postselected with acceptance rates as low as 10%; that is standard for error detection with a distance-2 code, but a reader should know these are conditional on passing stabilizer checks, not raw output fidelities.\n\nThe theory side holds up: the circuit constructions look sound for local faults, and the paper is careful about hook-error avoidance and flag-qubit reductions. The weak spot is the experimental demonstration claim, not the construction. The self-citations are heavy but mostly to the group's own prior protocols; they do cite the competing general constructions (refs 37–38) and do not hide them.\n\nBottom line: worth a serious referee. A referee should push for either a noise model that includes global dephasing or an explicit argument that the FT properties survive correlated dephasing, and for a clearer separation between 'fault-tolerant in the local-fault model' and 'fault-tolerant on this device.' I would not desk-reject; I would send to review, expecting revision or softening of the claims.","headline":"Impressive first measurement-free FT universal gate set on an ion trap, but the FT claim outruns the evidence because the noise model omits the global dephasing that dominates the experiment.","tokens_in":24917,"tokens_out":2583,"would_cite":true,"duration_ms":29058,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P68","81P70"],"pacs":["03.67.Pp","03.67.Lx"],"model":"deepseek-v4-flash","headline":"Universal fault-tolerant quantum computation can run without mid-circuit measurements, demonstrated on a trapped-ion processor with a universal gate set on an eight-qubit code and a three-logical-qubit Grover search.","keywords":["measurement-free quantum computation","fault-tolerant quantum computing","logical state teleportation","[[8,3,2]] color code","state injection","Grover search","trapped ions","coherent feedback"],"falsifier":"Simulate or measure the H_L injection circuit under a global dephasing channel that rotates all qubits together; if any single fault then yields a trivial stabilizer syndrome together with a logical X or Z flip on the [[8,3,2]] code, the circuits are not fault-tolerant as claimed. Experimentally, one can prepare |000>_L, apply the H_L gadget with a deliberately added global dephasing pulse, and check whether postselection removes the resulting logical error at the rate predicted by the local-dephasing model.","tokens_in":23957,"feed_emoji":"⚛️","tokens_out":4312,"duration_ms":43092,"temperature":0.7,"pith_summary":"The paper's central claim is that universal fault-tolerant quantum computation does not require mid-circuit measurements or feed-forward control. It demonstrates, on a 16-ion trapped-ion processor, a toolbox of fault-tolerant logical operations built entirely from coherent gates and qubit resets: modular logical state teleportation between two four-qubit error-detecting code blocks, and a universal fault-tolerant gate set on an eight-qubit code that hosts three logical qubits. As a capstone, it runs a fault-tolerant Grover search over eight entries on three logical qubits, finding the marked solutions with probability 0.40(4). The broader claim is that measurement-free schemes are a practical route to encoded quantum algorithms on platforms where measurements are slow or error-prone.","feed_headline":"Fault-tolerant logic runs without mid-circuit measurements","feed_subtitle":"A trapped-ion processor executes a universal gate set on an eight-qubit code and a three-logical-qubit Grover search using only coherent…","key_machinery":"The central mechanism is coherent feedback replacing measurement: logical operators are mapped onto auxiliary qubits, and conditional operations are applied directly instead of being triggered by classical measurement outcomes. Fault tolerance is bought by redundancy—using two stabilizer-equivalent representations of each logical operator with disjoint qubit support, and GHZ-stabilized auxiliary registers, so that a single fault either remains detectable or cancels. On the [[8,3,2]] code, the non-Clifford and Hadamard capabilities come from a transversal CCZ_L and state injection from a [[4,2,2]] ancilla code; the inter-block CNOT needed for injection is non-transversal but constructed so every single fault remains detectable.","core_discovery":"On its own terms, the paper establishes that the [[8,3,2]] color code—an eight-physical-qubit code encoding three logical qubits and detecting any single error—admits a complete, fault-tolerant, measurement-free universal gate set {H_L, CNOT_L, CCZ_L}: CNOT_L is done by relabeling physical qubits, CCZ_L is transversal using T- and T-dagger gates, and the missing H_L is injected coherently using an auxiliary [[4,2,2]] code prepared in |+0>_L, with the measurement and feed-forward of standard state injection replaced by CNOT-based coherent feedback. Fault tolerance is engineered by mapping two stabilizer-equivalent, disjoint-support representations of logical operators onto auxiliary GHZ states so that no single fault produces an undetected logical error. The same mechanism underlies logical teleportation between two [[4,1,2]] blocks, and the toolbox runs a two-solution Grover search whose total success probability is 0.40(4), with simulations indicating that modest error-rate reductions would push it past the classical 0.46.","pith_inferences":["Editorial inference: if the fault-tolerance construction survives non-local dephasing, then the practical bottleneck for measurement-free QEC shifts from measurement speed to idling-qubit coherence, favoring codes tailored to biased Z noise.","Editorial inference: the reported acceptance rates (as low as 10% for the H_L gate after postselection) suggest that for distance-2 codes, measurement-free fault tolerance currently trades run overhead for circuit simplicity; extending the flag-qubit ideas to distance-3 codes is the natural next test.","Editorial inference: the observed correlated logical errors across the three logical qubits imply that treating logical qubits as independent error channels underestimates the decoder complexity for small block codes."],"forward_implications":["Mid-circuit measurement is not necessary for fault-tolerant universal computation: qubit reset or replacement of auxiliary qubits suffices to remove entropy.","The teleportation protocol generalizes to higher-distance surface codes by preparing d-qubit GHZ states and mapping d disjoint operator representations, so the method is not specific to distance-2 codes.","A measurement-free logical Grover search currently achieves 0.40(4) success probability, below the 0.46 classical bound, but simulations show a two-qubit-gate error of 1.5% or a coherence time of 100 ms would exceed it.","The toolbox transfers to other platforms with all-to-all or long-range connectivity, such as neutral-atom arrays, where mid-circuit measurement is especially costly."],"supporting_citations":[{"why":"Provides the trapped-ion reset procedure and the experimental noise parameters (gate errors, coherence time) used throughout the simulations.","marker":"[7]"},{"why":"Supplies the code-switching baseline and the experimental context for fault-tolerant operations on the same trapped-ion platform.","marker":"[17]"},{"why":"Establishes the prior demonstration of fault-tolerant universal gate operations that the measurement-free toolbox extends by removing mid-circuit measurements.","marker":"[21]"},{"why":"Proposes the theoretical framework for measurement-free, scalable, fault-tolerant universal quantum computing that this paper turns into concrete circuits and experiments.","marker":"[37]"},{"why":"Defines the [[8,3,2]] color code and its transversal non-Clifford gate, which the paper uses as the logical gate set's backbone.","marker":"[48]"},{"why":"Underlies the state-injection technique that the H_L gate construction replaces with coherent feedback.","marker":"[52]"},{"why":"Supplies the Mølmer-Sørensen two-qubit interaction used as the native entangling gate in all implemented circuits.","marker":"[41]"}],"fun_headline_variants":["Measurement-free universal fault-tolerant quantum computation demonstrated","Trapped ions run universal fault-tolerant gates with zero mid-circuit measurements","Fault-tolerant quantum search without measurements, using only coherent ops","Measurement-free universal fault-tolerant computation on a trapped-ion processor","Coherent-only universal fault-tolerant quantum computation: first demonstration"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The fault-tolerance proof rests on a noise model of local depolarizing gates plus independent dephasing on idle qubits, and the experiment itself shows global magnetic-field dephasing is missing from that model, so if such global noise creates faults the circuits do not detect, the fault-tolerance claim would not hold.","fun_headline_variants_meta":{"raw":{"variants":["Measurement-free universal fault-tolerant quantum computation demonstrated","Trapped ions run universal fault-tolerant gates with zero mid-circuit measurements","Fault-tolerant quantum search without measurements, using only coherent ops","Measurement-free universal fault-tolerant computation on a trapped-ion processor","Coherent-only universal fault-tolerant quantum computation: first demonstration"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000517,"raw_usage":{"total_tokens":2516,"prompt_tokens":962,"completion_tokens":1554,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":578,"completion_tokens_details":{"reasoning_tokens":1473}},"tokens_in":578,"tokens_out":1554,"duration_ms":11802,"temperature":1.0,"reasoning_tokens":1473,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T22:02:27.670124+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate or measure the H_L injection circuit under a global dephasing channel that rotates all qubits together; if any single fault then yields a trivial stabilizer syndrome together with a logical X or Z flip on the [[8,3,2]] code, the circuits are not fault-tolerant as claimed. Experimentally, one can prepare |000>_L, apply the H_L gadget with a deliberately added global dephasing pulse, and check whether postselection removes the resulting logical error at the rate predicted by the local-dephasing model.","supporting_citations":[{"cited_title":"Pogorelov, F","cited_arxiv_id":null,"evidence_quote":"Establishes the prior demonstration of fault-tolerant universal gate operations that the measurement-free toolbox extends by removing mid-circuit measurements."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proposes the theoretical framework for measurement-free, scalable, fault-tolerant universal quantum computing that this paper turns into concrete circuits and experiments."},{"cited_title":"Guti´ errez, M","cited_arxiv_id":null,"evidence_quote":"Defines the [[8,3,2]] color code and its transversal non-Clifford gate, which the paper uses as the logical gate set's backbone."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Underlies the state-injection technique that the H_L gate construction replaces with coherent feedback."}],"review_version":1}