{"id":"3a9d55b8-720c-41af-a2cf-7cecb26c8fb1","arxiv_id":"1908.06103","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"A Rydberg-mediated controlled-Z gate entangles neutral atom pairs in a 2D array with measured Bell fidelity 0.86(2), inferred as 0.89 after error corrections.","lead":"This experiment demonstrates high-fidelity two-qubit entanglement in a 121-site 2D array of neutral atoms, using Rydberg blockade to implement a controlled-Z gate. The measured Bell-state fidelity is 0.86(2), with an inferred corrected fidelity of 0.89 after accounting for measurement and single-qubit errors.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The corrected CZ fidelity F_Bell^CZ=0.89 rests on an unexplained 'Rydberg laser dephasing' term calibrated from the same eye-diagram data, so the model-based subtraction could be circular; direct gate measurement is needed.","rationale":"The paper's raw measurement is credible, and the supplementary error model is detailed and thoughtful. The central claim, however, includes a corrected gate fidelity that depends on an error decomposition rather than on direct observation. The most vulnerable point is the a.8 Rydberg laser dephasing term: it is explicitly admitted to be unexplained by the physical model, it is calibrated by simulating the same eye-diagram experiment that it is then used to describe, and it contributes substantially to the CZ error budget. Because the corrected fidelity is obtained by subtracting modeled SPAM and single-qubit errors from the observed Bell fidelity, any misspecification of this term, or of the Markovian assumption noted in SM-IV E, changes the inferred CZ fidelity. The reader's verdict of CONDITIONAL already captures this; my concern sharpens the reason for the condition but does not move the verdict. I recommend keeping CONDITIONAL and adding a request for an independent gate-fidelity measurement or an uncertainty estimate on the corrected value. No ad hominem is intended; the authors are appropriately cautious in the supplement, but the headline number is more fragile than the raw data.","tokens_in":26058,"tokens_out":3385,"duration_ms":34828,"concrete_test":"Perform interleaved randomized benchmarking (or direct two-qubit quantum process tomography) of the Rydberg CZ gate alone on the same apparatus, independent of the Bell-state preparation sequence and of the eye-diagram calibration used to fix the a.8 dephasing term. Compare the directly measured average gate fidelity with the model-inferred F_Bell^CZ = 0.89; if the difference exceeds the ~0.02 gap between the raw 0.86 and the corrected 0.89, the model-based correction is not reliable. As a complementary analytical check, recompute the error budget with the a.8 term removed and with it re-assigned as amplitude damping; if F_Bell^CZ shifts by more than 0.01, the interpretation of that term is material to the headline claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The raw F_Bell=0.86(2) is a direct measurement and stands. The headline corrected value F_Bell^CZ=0.89, however, is not measured: it is the output of a chi-matrix error model that subtracts inferred SPAM and single-qubit errors from the observed fidelity. The largest single Rydberg-gate error in that model is the unexplained 'Rydberg laser dephasing' term (SM-I H, SM-IV B): 0.018 per atom per non-blockaded Rydberg pi pulse and 0.006 per blockaded pi pulse. The supplement states that 'our models do not explain this additional dephasing term; however, we need to include this effect to make our model consistent' and that its value was extracted by simulating the eye-diagram experiment using quantum process matrices. Because the same parameter is then used to predict the Bell-state fidelity, the model can absorb any unmodeled error, such as the non-Markovian effects explicitly excluded in SM-IV E or an unmodeled leakage channel, into a term attributed to dephasing. The corrected 0.89 therefore inherits the assumption that this calibrated term is a real dephasing channel and that all channels combine independently and Markovianly. The paper itself flags 'relatively large uncertainties' on the laser-noise and atom-position entries and gives no uncertainty on the corrected values. This is the load-bearing gap: without an independent measurement of the CZ gate fidelity, the central quantitative claim about the gate is model-dependent.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports improved Rydberg-mediated two-qubit entanglement in a 121-site two-dimensional neutral-atom array defined by blue-detuned lines of light. The authors measure a Bell-state fidelity of F_Bell = 0.86(2) from populations and parity oscillations, and then use a detailed χ-matrix error model to correct for state preparation and measurement (SPAM) errors, inferring F_Bell^{-SPAM} = 0.88, and for single-qubit errors, inferring F_Bell^{CZ} = 0.89 for the Rydberg C_Z gate. The error budget identifies finite atom temperature and laser noise as the dominant remaining error sources. The supplementary material provides the supporting calculations for atomic parameters, single-qubit errors, SPAM errors, and the process-matrix propagation, as well as explicit statements of the model's limitations.","tokens_in":26427,"tokens_out":4320,"duration_ms":41825,"significance":"If the central claims hold, the paper represents a meaningful advance for neutral-atom quantum computing: it demonstrates a two-qubit Rydberg C_Z gate in a scalable 2D geometry with a raw Bell fidelity of 0.86(2), which is a direct observable with clear methods, and it provides a detailed, state-dependent error model that goes beyond a simple multiplicative error budget. The explicit identification of finite temperature and laser noise as dominant errors is useful guidance for the field. The main strength is the direct parity-oscillation and population measurement; the main gap is that the headline corrected fidelity of 0.89 is not a direct measurement but the output of an error model containing calibrated terms with acknowledged uncertainties.","major_comments":[{"comment":"The corrected fidelity F_Bell^{CZ}=0.89 rests on the 'Rydberg laser dephasing' term of 0.018 per non-blockaded and 0.006 per blockaded Rydberg pi pulse. The supplement states explicitly that the model does not explain this additional dephasing term and that it must be included to make the eye-diagram simulation consistent, with the value extracted from the same eye-diagram experiment that the model is meant to reproduce. Since this term is the largest single Rydberg-gate error in Table I, the model-based subtraction could absorb unmodeled errors, such as non-Markovian effects or a leakage channel, into an attributed dephasing channel. The authors should either supply an independent measurement or physical mechanism for this term, or visibly weaken the claim attached to F_Bell^{CZ}=0.89 and instead emphasize the directly measured F_Bell=0.86(2).","section":"SM-I H and SM-IV B"},{"comment":"The atom-position error sigma=0.16 micrometers used for entry a.5 is not measured directly; it is inferred from observed ground-Rydberg Rabi errors, and the supplement states that confidence in this estimate is lower than for other entries. The laser noise error of 0.0025 per pi pulse is similarly described as difficult to quantify, with 'relatively large uncertainties.' Because a.5 and a.6 are jointly extracted from 2-pi Rabi data and have similar physical effects (leaving population in the Rydberg state), the partition between them is not unique. The corrected fidelities should be reported with uncertainties propagated from these inputs, rather than as point values.","section":"SM-I E and SM-I F"},{"comment":"The error propagation uses chi-matrices, which cannot represent non-Markovian processes; the supplement assumes that such effects are small. Given that the a.8 dephasing term is calibrated from the same eye-diagram data used to check the model, the assumption that the residual is a Markovian dephasing channel is precisely the load-bearing assumption for F_Bell^{CZ}=0.89. A master-equation treatment or, preferably, a direct characterization of the CZ gate (e.g., randomized benchmarking or process tomography of the two-qubit gate) would be needed to substantiate the corrected value.","section":"SM-IV E"}],"minor_comments":[{"comment":"The title contains a typo: 'neutra l' should be 'neutral'.","section":"Title"},{"comment":"The corrected fidelities F_Bell^{-SPAM} and F_Bell^{CZ} are quoted without uncertainties; please add a statement clarifying that these are model point estimates, and avoid implying they have the same statistical status as the measured F_Bell=0.86(2).","section":"Abstract and Table I"},{"comment":"In the discussion of ground-Rydberg Doppler dephasing, the relation between the measured Ramsey coherence time T2,gR=4 microseconds and the calculated T2,D=6 microseconds is stated but not derived; a short explanation of why the calculation is preferred for Table I would aid the reader.","section":"SM-I A"},{"comment":"The entry 'global per atom per micro wave pi/2 pulses' should read 'microwave,' and the Stark-shift pulse entry should specify whether the 0.006 error is per pulse or per gate, for consistency with the other entries.","section":"Table I and SM-II"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the journal's scope, and the raw data are valuable. The main risk is that the abstract presents F_Bell^{CZ}=0.89 as the central quantitative claim without carrying the caveats that the supplementary material itself states. I would urge the editor to require a revised manuscript that either directly characterizes the CZ gate or visibly re-weights the claims toward the measured fidelity."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nKey takeaway: this is a solid experimental advance with a headline number that deserves scrutiny. The measured Bell fidelity 0.86(2) in a 121-site 2D array is direct and credible. The corrected F^CZ=0.89 is an inference from a detailed chi-matrix error model, and the model has a load-bearing soft spot.\n\nWhat's new: they improve their own prior 0.79 result in a 2D geometry to 0.86(2), using a blue-detuned line array, lower Rydberg n, AOD scanning, and better laser stabilization. The supplementary is unusually thorough: full error budget, process matrices, calibration details. The raw fidelity is an external observable; that part is not circular.\n\nWhere I'd push back: the corrected fidelity rests on several calibrated inputs with no quoted uncertainty. The biggest is the \"Rydberg laser dephasing\" term—0.018 per non-blockaded pi pulse, 0.006 blockaded—which the supplement admits is not explained by their models and is extracted by simulating the eye diagram. Using that same term to predict the Bell-state fidelity injects a real circularity risk: any unmodeled error can be absorbed into \"dephasing.\" Same for the atom localization sigma=0.16 um, inferred from Rabi errors rather than measured. The Markovian assumption (SM-IV E) is also stated, not tested. So the corrected 0.89 should be read as a model-dependent estimate, not a measured gate fidelity.\n\nThat said, the stress-test concern is legitimate but not fatal. The paper's core claim—that Rydberg entanglement in a scalable 2D array reaches ~0.86 measured fidelity—stands on its own. The raw value is a milestone for hyperfine-qubit neutral atoms in 2D. I'd trust the error budget directionally; temperature and laser noise as dominant errors is plausible.\n\nFor a referee: the paper deserves serious review, not desk rejection. The main revision should be to separate the measured result from the inferred one much more sharply, quote uncertainties on the corrected values, and ideally report a direct CZ gate measurement (e.g., randomized benchmarking) as future work. The manuscript is honest about its limitations, which helps.\n\nI'd cite the raw fidelity result. It's useful for anyone tracking neutral atom gate progress, and the supplementary is a good template for error budgeting.\n\nRecommendation: send to peer review. Conditional acceptance with request for clarity on model dependence.","headline":"Raw Bell fidelity 0.86(2) is a real step forward in 2D neutral atom arrays; the corrected 0.89 is model-dependent and should be treated as such.","tokens_in":27022,"tokens_out":1685,"would_cite":true,"duration_ms":16602,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A Rydberg-mediated CZ gate entangles two qubits in a 121-site 2D neutral-atom array with measured Bell fidelity 0.86(2) and inferred gate fidelity 0.89.","keywords":["Rydberg blockade","neutral atom qubit array","CZ gate","Bell state fidelity","quantum process matrix","optical lattice","two-qubit entanglement"],"falsifier":"Run interleaved randomized benchmarking on the CZ gate in the same array and compare the extracted fidelity with the corrected $F_{\\rm Bell}^{C_Z}=0.89$; if benchmarking returns a systematically lower fidelity, one or more calibrated terms in the error model are over-attributing error to SPAM, single-qubit gates, or assumed dephasing.","tokens_in":25861,"feed_emoji":"⚛️","tokens_out":5477,"duration_ms":49895,"temperature":0.7,"pith_summary":"This paper reports a two-qubit entangling gate in a 121-site two-dimensional array of neutral cesium atoms, with a measured Bell-state fidelity of $F_{\\rm Bell}=0.86(2)$. After correcting for state-preparation-and-measurement errors with a process-matrix error model, the authors infer $F_{\\rm Bell}^{\\rm -SPAM}=0.88$, and after removing single-qubit gate errors they infer an intrinsic CZ gate fidelity of $F_{\\rm Bell}^{C_Z}=0.89$. The result matters because it moves neutral-atom arrays toward a scalable architecture: the same hardware supports site-addressed gates throughout the array, and the error model identifies finite atom temperature and laser noise, not Rydberg physics, as the dominant remaining infidelity.","feed_headline":"Neutral-atom 2D array entangles qubits at 0.89 fidelity","feed_subtitle":"Rydberg blockade plus a 121-site optical lattice brings scalable neutral-atom computing closer.","key_machinery":"The load-bearing object is the Rydberg CZ gate: two laser pulses on a control atom (a $\\pi$ pulse, then another $\\pi$ pulse after a gap) and one $2\\pi$ pulse on the target atom, using two-photon excitation to the $66s_{1/2}$ Rydberg state. The Rydberg blockade, in which one atom's Rydberg excitation suppresses the other's, imprints the controlled phase that creates entanglement. The supporting machinery is the $\\chi$-matrix process model, which propagates every measured or calculated error source through the full pulse sequence as independent quantum channels and yields the corrected fidelity values.","core_discovery":"The paper's central claim is that the Rydberg blockade mechanism, implemented as the standard $\\pi$--gap--$2\\pi$--gap--$\\pi$ pulse sequence between two hyperfine clock states and a $66s_{1/2}$ Rydberg state, can entangle two qubits in a two-dimensional optical-lattice array at $F_{\\rm Bell}=0.86(2)$ raw, rising to $F_{\\rm Bell}^{C_Z}=0.89$ once SPAM and single-qubit errors are accounted for. The evidence is population data and a parity oscillation amplitude $C=0.391(6)$, together with a $\\chi$-matrix error budget whose predicted output $F_{\\rm Bell}=0.853$ matches the measurement. The corrected value is the paper's headline claim: it separates the CZ gate's intrinsic quality from the measurement overhead.","pith_inferences":["My inference: if the same error model holds after cooling to a few microkelvin and resonator-filtering the Rydberg lasers, the gate fidelity should climb well above 0.95; this is a testable prediction the paper does not make.","My inference: the unexplained Rydberg laser dephasing term (0.018 unblockaded versus 0.006 blockaded per $\\pi$ pulse) is the least physically motivated entry in the error budget; a side-by-side measurement of ground-Rydberg Ramsey decay with and without laser illumination would separate it from genuine laser noise.","My inference: a direct measurement of the atom localization width $\\sigma$, rather than the inferred $\\sigma=0.16\\,\\mu\\mathrm{m}$, would tighten the error budget; trap-frequency spectroscopy can provide that measurement.","My inference: interleaved randomized benchmarking of the CZ gate in the same array would test whether the Markovian error model's corrected fidelity is trustworthy, since benchmarking does not rely on the same SPAM corrections."],"forward_implications":["A neutral-atom architecture with site-addressable gates throughout a large 2D array can sustain two-qubit entangling operations near the 0.89 level, not just in one-dimensional or few-qubit settings.","Because the dominant errors are finite atom temperature and laser phase and intensity noise, straightforward technical improvements such as colder atoms and quieter lasers should directly raise the gate fidelity.","Rydberg lifetime and blockade-strength errors are each below 1%, so the choice of Rydberg state and lattice spacing is not the bottleneck in this geometry.","Combined with demonstrated single-qubit fidelities above 0.99 and atom rearrangement capabilities, the setup provides the ingredients for multi-qubit algorithms on a 2D array."],"supporting_citations":[{"why":"Proposes the $\\pi$--$2\\pi$--$\\pi$ Rydberg CZ gate sequence that the experiment implements.","marker":"[6]"},{"why":"Earlier demonstration of a Rydberg-blockade CNOT and entanglement in a 2D array, the baseline this work upgrades.","marker":"[10]"},{"why":"Shows how resonator-filtered Rydberg lasers improve Rydberg control, the comparison for the no-filtering performance demonstrated here.","marker":"[14]"},{"why":"Detailed atomic-structure analysis showing the heavy-alkali pulse sequence can in principle reach fidelity above 0.998.","marker":"[24]"},{"why":"Provides the Rydberg quantum-information formalism, including the blockade-leakage error estimate used in the model.","marker":"[5]"},{"why":"First entanglement of two atoms via Rydberg blockade and the source of the Doppler-dephasing expression used for the finite-temperature error.","marker":"[28]"},{"why":"Analyzes how laser phase noise degrades coherent Rydberg excitation, the basis for the laser-noise error estimate.","marker":"[22]"},{"why":"Randomized benchmarking of single-qubit gates in a 2D neutral-atom array, supporting the single-qubit fidelity assumptions.","marker":"[7]"},{"why":"The supplementary material presenting the $\\chi$-matrix process model that produces the corrected fidelity values.","marker":"[19]"}],"fun_headline_variants":["Rydberg 2D array hits 0.89 Bell fidelity","121-site lattice entangles qubits at 0.89","Corrected Bell fidelity 0.89 via Rydberg blockade","Two-qubit gate at 0.89 fidelity in 2D array"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The corrected 0.89 fidelity stands only if the error model includes every real error source and those errors combine as independent memoryless processes; the paper assumes memory effects are small and calibrates its Rydberg laser-dephasing term and atom position spread to match the data rather than measuring them outright.","fun_headline_variants_meta":{"raw":{"variants":["Rydberg 2D array hits 0.89 Bell fidelity","121-site lattice entangles qubits at 0.89","Corrected Bell fidelity 0.89 via Rydberg blockade","Two-qubit gate at 0.89 fidelity in 2D array"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000731,"raw_usage":{"total_tokens":3236,"prompt_tokens":874,"completion_tokens":2362,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":490,"completion_tokens_details":{"reasoning_tokens":2285}},"tokens_in":490,"tokens_out":2362,"duration_ms":15573,"temperature":1.0,"reasoning_tokens":2285,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:56:01.360363+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run interleaved randomized benchmarking on the CZ gate in the same array and compare the extracted fidelity with the corrected $F_{\\rm Bell}^{C_Z}=0.89$; if benchmarking returns a systematically lower fidelity, one or more calibrated terms in the error model are over-attributing error to SPAM, single-qubit gates, or assumed dephasing.","supporting_citations":[{"cited_title":"Single- qubit gates based on targeted phase shifts in a 3D neutral atom array,","cited_arxiv_id":null,"evidence_quote":"Earlier demonstration of a Rydberg-blockade CNOT and entanglement in a 2D array, the baseline this work upgrades."},{"cited_title":"Entangling two individual atoms of different isotopes via rydberg blockade,","cited_arxiv_id":null,"evidence_quote":"Shows how resonator-filtered Rydberg lasers improve Rydberg control, the comparison for the no-filtering performance demonstrated here."},{"cited_title":"Analysis of imperfections in the coher- ent optical excitation of single atoms to Rydberg states,","cited_arxiv_id":null,"evidence_quote":"Detailed atomic-structure analysis showing the heavy-alkali pulse sequence can in principle reach fidelity above 0.998."},{"cited_title":"Superconducting quantum circuits at the surface code threshold for fault tolerance,","cited_arxiv_id":null,"evidence_quote":"Provides the Rydberg quantum-information formalism, including the blockade-leakage error estimate used in the model."},{"cited_title":"A high fidelity Rydberg blockade entangling gate us- ing shaped, analytic pulses,","cited_arxiv_id":null,"evidence_quote":"First entanglement of two atoms via Rydberg blockade and the source of the Doppler-dephasing expression used for the finite-temperature error."},{"cited_title":"Magic- wavelength optical traps for Rydberg atoms,","cited_arxiv_id":null,"evidence_quote":"Analyzes how laser phase noise degrades coherent Rydberg excitation, the basis for the laser-noise error estimate."},{"cited_title":"Quantum information with Rydberg atoms,","cited_arxiv_id":null,"evidence_quote":"Randomized benchmarking of single-qubit gates in a 2D neutral-atom array, supporting the single-qubit fidelity assumptions."},{"cited_title":"Coherent operations, entanglement, and progress towards quantum search in a large 2D ar- ray of neutral atom qubits,","cited_arxiv_id":null,"evidence_quote":"The supplementary material presenting the $\\chi$-matrix process model that produces the corrected fidelity values."}],"review_version":1}