{"id":"36e0eed9-0a1b-4f2a-9472-df28557a6cec","arxiv_id":"2512.22569","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"An optimized reshaping of an optical superlattice can realize a √SWAP gate between two fermionic atoms in ~21 μs with >99% simulated fidelity — over an order of magnitude faster than tunneling-based gates.","lead":"This paper proposes a way to do a two-qubit entangling gate on ultracold fermionic atoms in an optical superlattice about 50 times faster than current methods, by briefly reshaping the laser potential so the atoms collide in the middle and swap. If the simulations hold up, this removes a major speed bottleneck for neutral-atom quantum computers.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The predicted ≥99% fidelity rests on a 1D delta-interaction model whose calibration error grows with interaction strength; the proposed operating point uses γ 4–7× outside the benchmarked range, so the headline speed/fidelity claim is unvalidated extrapolation.","rationale":"The reader's conditional verdict correctly identifies the model calibration as the weak point. I agree, with one addition: the extrapolation in γ is even starker than the 0.82 rescaling alone suggests. The benchmarks use γ/h=4.36–7.35 kHz·μm; the gate uses 30.96 kHz·μm. A 4–7× extrapolation combined with known 10–20% model errors is enough to undermine a 99% claim. The paper has real independent support: single-particle tunneling frequencies agree within ~2%, spin-exchange frequencies match, and the Blackman gate fidelity matches the experimental value. But none of those tests exercise the fast-collision, high-γ regime. The fidelity in Eq. 15 is a continuum-simulation observable; without convergence tests or a 3D check, it is a prediction of an unvalidated model. I would keep the verdict CONDITIONAL (UNCHANGED) because the concern is addressable and not fatal; the proposed concrete test would settle it. I also flag that Eqs. 13–14 as printed appear singular at τ=τ_gate/2, contradicting the text's description of the envelopes; this is likely a typographical error but must be corrected for reproducibility.","tokens_in":12770,"tokens_out":10051,"duration_ms":109621,"concrete_test":"Run a full 3D two-particle time-dependent Schrödinger simulation (or a 1D model with finite-range interaction of width equal to the transverse oscillator length) using the exact envelopes and parameters of Eq. 15, 6Li mass, the superlattice geometry of Ref. [26], and a 3D scattering length mapped from γ via the quasi-1D confinement-induced resonance relation. If the optimized fidelity drops below 99% or the optimal γ shifts by more than ~20% relative to 30.96 kHz·μm, the claim is not supported. As a cheaper check, first reproduce the pair-tunneling curve of Fig. 2b with the 0.82-rescaled γ; if the frequency/phase mismatch persists, the delta model is not predictive in the collision-dominated regime.","verdict_should_be":"UNCHANGED","load_bearing_attack":"All benchmarks in Sec. III are in the tunneling/superexchange regime with γ/h = 4.36–7.35 kHz·μm. The proposed gate (Eq. 15) uses γ/h = 30.96 kHz·μm, i.e., 4–7× larger, during a fast transient release into a quasi-harmonic well where the collision energy is far above the tight-binding scale. The model is already known to be imperfect in the benchmarked regime: reproducing the strongly-interacting spin-exchange data requires an ad hoc rescaling γ→0.82γ (Sec. IIIB, Fig. 4); the pair-tunneling simulation disagrees with experiment in both frequency (4.182 vs 3.8 kHz) and phase (Sec. IIIB); and the Blackman √SWAP benchmark reaches 99.79% fidelity only after optimizing γ/h to 7.35 kHz·μm, 52% above the experimental value of 4.84 kHz·μm (Sec. IIIC). No benchmark is performed at large γ or at the high relative velocities that characterize the collisional release. The Discussion's statement that the 1D results 'strongly indicate' 3D realizability is an assertion, not a demonstrated 1D→3D correspondence. If the 1D delta model overestimates the interaction-induced phase by an amount comparable to the known 18–20% calibration error, the 99% fidelity and the 21 μs operating point are not quantitatively reliable. This is the load-bearing weak point of the paper's central claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes a fast entangling √SWAP gate for two fermionic atoms in an optical superlattice. The gate sequence uses time-dependent control of the short- and long-lattice depths (Eq. 1), transiently releasing the atoms into a quasi-harmonic central confinement; a constant contact interaction accumulates the required exchange phase. The authors solve the two-particle continuum Schrödinger equation with a regularized delta interaction (Eq. 2), benchmark it against single-particle tunneling, spin-exchange, pair-tunneling, and Blackman √SWAP data from Ref. [26], and then optimize the envelope parameters (Eqs. 13–15) to obtain a 21.2 μs √SWAP with 99% fidelity. A repeated SWAP^{3/4} composite sequence is claimed to maintain >98.7% fidelity under ±5% lattice-depth variations.","tokens_in":13182,"tokens_out":5422,"duration_ms":56008,"significance":"If the predicted fidelities survive experimental validation, this would be a substantial advance: a collision-mediated gate roughly two orders of magnitude faster than the current tunneling-based superlattice gate, using only existing lattice controls. The numerical benchmarks are genuinely valuable: single-particle tunneling is reproduced without free parameters (2–8% agreement), the spin-exchange frequency matches experiment exactly, and the Blackman gate fidelity is reproduced to 0.04%. However, the interaction model's calibration in the strongly interacting regime requires ad hoc corrections, and the proposed operating point lies outside the benchmarked parameter range; the quantitative headline result is therefore a conditional numerical prediction rather than an established fact.","major_comments":[{"comment":"The central 99%/21.2 μs claim is computed at γ/h = 30.96 kHz·μm, while every interaction benchmark in Sec. III uses γ/h between 4.36 and 7.35 kHz·μm. The model's accuracy is already questionable in the benchmarked regime: pair-tunneling frequency is 4.182 vs 3.8 kHz with a phase mismatch, spin-exchange requires an ad hoc γ→0.82γ rescaling (Fig. 4), and the Blackman gate fidelity is matched only after optimizing γ to 7.35 kHz·μm, 52% above the experimental 4.84 kHz·μm (Sec. III.C). No benchmark exercises the model at large γ or at the high relative velocities of the collisional release. Since the proposed gate relies on the interaction-induced phase during this transient, the 99% fidelity is an extrapolation. Please provide validation in the large-γ/high-momentum regime (e.g., against exact 1D solutions for delta interactions, 3D simulations, or a quantitative error analysis) or explicitl","section":"§III.B, §III.C, Eq. (15)"},{"comment":"The statement that the 1D results 'strongly indicate' realizability in realistic 3D superlattices is asserted, not demonstrated. The 1D delta model with γ extracted via Eq. (9) assumes strict transverse confinement; during the release into the quasi-harmonic well, atoms may occupy excited transverse states, and the effective 1D coupling changes with the local density and collision energy. The sensitivity of the 99% fidelity to the known calibration errors (0.82 rescale, 52% Blackman discrepancy) is not shown. I request a fidelity-vs-γ curve at the operating point, reporting the range of γ for which F>99% (or >97%), so that the reader can judge whether the 18–20% model error is tolerable.","section":"§VI, Eq. (9)"},{"comment":"The robustness analysis is incomplete for the central claim. Only global lattice-amplitude variations are tested; variations in γ are not considered, even though γ is the one parameter the model calibrates least reliably. The composite sequence reduces γ to 12.8 kHz·μm (Eq. 16), but this is still ~1.7–2.9 times larger than the benchmarked values, so the >98.7% robustness claim inherits the same extrapolation problem.","section":"§V, Eq. (16)"}],"minor_comments":[{"comment":"The abstract states the gate is 'more than an order of magnitude faster than tunneling-based implementations'; Sec. V gives a factor ~54 vs the Blackman protocol but only ~8.4 vs the fast-ramp implementation. The claim should be qualified accordingly.","section":"Abstract, §V"},{"comment":"The regularized delta interaction δreg is used without specifying the regularization prescription. Please define or cite the specific form before first use.","section":"§II, Eq. (2)"},{"comment":"Several figures lack complete axis labels and units in the current version, making quantitative comparison with the text difficult. For example, Fig. 2 shows holding time but not the occupation axis label.","section":"Figures 2, 4, 7"},{"comment":"There is a typographical mismatch in the caption: '|ψ↑↓(x1, x2, τ)' is missing a closing parenthesis. Please correct.","section":"§III.B, Fig. 3 caption"},{"comment":"The note about the closely related Ref. [42] is too vague. Please specify how the present work differs from or is related to that concurrent preprint.","section":"Note added"}],"recommendation":"major_revision","confidential_remarks":"This is a solid numerical proposal with useful benchmarks, but the central quantitative claims rest on an interaction model in a regime where it is not calibrated. If the authors can supply a convincing validation at large γ and high collision energy, or appropriately qualify the 99% fidelity claim, I would support publication. The editor may also wish to check the overlap with Ref. [42], which the authors themselves note is closely related."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. The proposal itself is new and worth taking seriously: instead of waiting for tunneling, the authors transiently reshape the superlattice into a near-harmonic well, let the atoms collide, then reverse the shape change. Optimized envelopes for the short and long lattices are given explicitly, and the single-particle transfer is very clean. Second, the paper is unusually honest about its simulator's limits—it openly reports needing a 0.82 rescaling of the contact interaction to match spin-exchange data and a clear discrepancy in pair tunneling—but the operating point of the proposed gate sits at gamma/h = 31 kHz·µm, four to seven times larger than every benchmark they ran. The 99% fidelity is a plausible extrapolation, not a validated number.\n\nWhat is done well: the benchmarking against Bojović et al. is serious. Single-particle tunneling matches to a few percent, spin-exchange frequency matches after the explicitly noted rescale, and the Blackman-pulse gate fidelity reproduces the experiment to 0.04%. They also show why a naive short-lattice shutdown fails, which is a useful negative result. The contrast with tunneling-based gates and with the authors' own tweezer work is clear.\n\nWhere it's soft: the central quantitative claim depends on a 1D delta-interaction model in a regime—large gamma, short timescale, high relative collision energy—where the model has not been checked. The pair-tunneling benchmark already misses the experimental frequency by ~10% and the phase, and the fact that two different benchmarks require different effective gamma values suggests the delta-function picture has a limited range. The paper argues 1D-to-3D correspondence by assertion. No numerical convergence data or code are given, and fidelities are reported without uncertainty. None of this makes the idea worthless, but it means the abstract's '>99%' should not be taken at face value.\n\nWho should read it: people working on neutral-atom quantum computing, especially superlattice gate schemes, and theorists who care about how much you can trust 1D benchmarks for 3D proposals. I'd send it to peer review—it deserves a serious referee—but I'd expect the referee to demand either a benchmark in the high-gamma regime or a much more careful discussion of the model's validity there, and likely a revision that tempers the quantitative claims.\n\nRecommendation: engage with it, but treat the headline numbers as hypotheses until they get tested in a real 3D setting.","headline":"A well-benchmarked proposal for a fast superlattice gate, but the headline 21 µs / >99% fidelity rests on a 4–7× extrapolation of the interaction model beyond its validated range.","tokens_in":13652,"tokens_out":4294,"would_cite":true,"duration_ms":46662,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["03.67.Lx","05.30.Fk","32.80.Qk","34.50.Cx"],"model":"deepseek-v4-flash","headline":"A 21-microsecond collisional √SWAP gate for fermionic atoms in an optical superlattice, with >99% fidelity in continuum simulation.","keywords":["collisional gate","optical superlattice","fermionic atoms","√SWAP gate","quantum computation","time-dependent Schrödinger equation","contact interaction","lattice-depth control"],"falsifier":"A 3D simulation or experiment of the proposed fast-collision gate would test the predicted fidelity. Specifically, if a full 3D simulation of the same lattice parameters yields a fidelity below 99% (e.g., due to transverse degrees of freedom or non-ideal delta-interaction), the central claim would be falsified. Alternatively, an experimental measurement of the pair-tunneling frequency during the fast ramp would reveal whether the 1D model's 0.82 rescaling holds, as a mismatch would indicate the need for a more accurate interaction model.","tokens_in":12630,"feed_emoji":"⚛️","tokens_out":1330,"duration_ms":15510,"temperature":0.7,"pith_summary":"This paper proposes a fast entangling gate for two fermionic atoms in an optical superlattice, replacing the slow tunneling-based approach with a controlled collision. The atoms are transiently released into a quasi-harmonic confinement formed by carefully timed modulation of the short- and long-lattice depths, and a contact interaction accumulates the exchange phase needed for √SWAP. The authors simulate the full continuum two-particle Schrödinger equation, benchmark it against experimental tunneling-based gates, and find a gate time of ~21 microseconds with fidelity above 99%, over an order of magnitude faster than tunneling-based implementations. They also show a composite sequence that maintains fidelity above 98.7% under ±5% lattice-amplitude variations.","feed_headline":"21-microsecond √SWAP gate for fermionic atoms in a superlattice","feed_subtitle":"A collision-mediated gate runs an order of magnitude faster than tunneling-based gates while keeping fidelity above 99%.","key_machinery":"The central mechanism is the transient release of two atoms into a quasi-harmonic confinement created by time-dependent modulation of the superlattice potential V(x,τ) = V_S(τ)cos²(πx/a_x) + V_L(τ)sin²(πx/(2a_x)). The authors keep a modest short-lattice amplitude while rapidly raising the long lattice, preserving a quasi-harmonic potential that focuses the wave packets and avoids anharmonic dispersion. The contact interaction strength γ is tuned so that the collision during the central part of the motion produces the required exchange phase. The theoretical framework is the two-particle time-dependent Schrödinger equation in the continuum (Eq. 2), which goes beyond the two-site Fermi–Hubbard","core_discovery":"The central claim is that a fast, high-fidelity √SWAP gate for fermionic atoms in an optical superlattice can be realized by dynamically controlling the short- and long-lattice depths, rather than relying on slow tunneling. The atoms are released from their initial double-well sites into a quasi-harmonic confinement centered between the sites; with a tuned contact interaction, a controlled collision during the transient release accumulates a relative π/2 phase between even and odd wave-function components, generating entanglement. The optimized protocol (with maximum long-lattice depth 140 recoil energies, initial short-lattice depth 41.35 recoil energies, and gate time 21.2 μs) achieves a s","pith_inferences":["The paper's 1D continuum model, calibrated with a 0.82 rescaling of the interaction coefficient in the strongly interacting regime, likely underestimates the complexity of the real 3D collision dynamics; the predicted 99% fidelity may degrade when applied to a 3D superlattice.","The composite SWAP^{3/2} sequence could be further optimized by exploring non-uniform lattice-depth variations or by incorporating optimal control techniques that account for spatial intensity inhomogeneities, potentially achieving even higher robustness.","The collision-mediated gate concept might extend to other species or to higher-dimensional superlattices, where the quasi-harmonic focusing mechanism could enable fast two-qubit gates in a 2D array.","A direct experimental test would measure the pair-tunneling frequency in the proposed fast-collision regime; the paper's simulation predicts a deviation from the simple tunneling picture, which could serve as a signature of the collision-mediated phase accumulation."],"forward_implications":["The proposed gate operates in ~21 μs, more than an order of magnitude faster than tunneling-based implementations (typically ~1 ms), while maintaining fidelity above 99% in simulation.","A composite sequence of two SWAP^{3/4} gates, realizing SWAP^{3/2}, suppresses sensitivity to lattice-amplitude variations, keeping fidelity above 98.7% for ±5% variations.","Increasing the maximum accessible long-lattice depth to 700 recoil energies shortens the gate to 9.2 μs with slightly higher fidelity (99.41%), suggesting a trade-off between speed and available lattice depth.","The continuum simulation, benchmarked against experimental spin-exchange and pair-tunneling data, captures dynamics beyond the tight-binding approximation, validating the protocol for realistic experimental conditions.","The collision-mediated approach offers a route to scalable neutral-atom quantum computation by decoupling gate speed from tunneling rate."],"fun_headline_variants":["21 µs √SWAP gate for fermionic atoms via collision","Collision-based √SWAP: 21 µs, >99% fidelity","Superlattice √SWAP gate: 10x faster than tunneling","Fast fermionic gate: √SWAP in 21 µs","Controlled collision yields 21 µs √SWAP gate"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The paper's central claim relies on the assumption that the 1D continuum model with a regularized delta-function contact interaction, calibrated to experiment via a 0.82 rescaling of the interaction coefficient in the strongly interacting regime, quantitatively predicts the real 3D two-atom collision dynamics during the fast transient release.","fun_headline_variants_meta":{"raw":{"variants":["21 µs √SWAP gate for fermionic atoms via collision","Collision-based √SWAP: 21 µs, >99% fidelity","Superlattice √SWAP gate: 10x faster than tunneling","Fast fermionic gate: √SWAP in 21 µs","Controlled collision yields 21 µs √SWAP gate"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00078,"raw_usage":{"total_tokens":3296,"prompt_tokens":770,"completion_tokens":2526,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":514,"completion_tokens_details":{"reasoning_tokens":2432}},"tokens_in":514,"tokens_out":2526,"duration_ms":17957,"temperature":1.0,"reasoning_tokens":2432,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T13:48:58.230740+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A 3D simulation or experiment of the proposed fast-collision gate would test the predicted fidelity. Specifically, if a full 3D simulation of the same lattice parameters yields a fidelity below 99% (e.g., due to transverse degrees of freedom or non-ideal delta-interaction), the central claim would be falsified. Alternatively, an experimental measurement of the pair-tunneling frequency during the fast ramp would reveal whether the 1D model's 0.82 rescaling holds, as a mismatch would indicate the need for a more accurate interaction model.","supporting_citations":[],"review_version":1}