{"id":"92125d9f-b37d-4c41-96eb-eb15f7ece992","arxiv_id":"2412.11821","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Fast non-perturbative single-qubit gates for continuously driven (CDD) qubits are demonstrated on a transmon, with average Clifford fidelity 0.9947(1) and over tenfold coherence improvement.","lead":"A team shows that qubits protected by a continuous microwave 'spin-locking' drive can be operated with fast, high-fidelity single-qubit gates. They report 99.47% average Clifford fidelity on a noisy flux-tunable superconducting transmon, more than ten times better coherence than without the drive.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Three-level leakage model and stated node condition are quantitatively inconsistent, so the claimed generality to any CDD qubit architecture rests on an unvalidated approximation.","rationale":"The reader's verdict CONDITIONAL is appropriate. The paper provides a credible experimental demonstration of fast single-qubit gates on a CDD transmon, with a clear coherence enhancement and a randomized benchmarking fidelity of 0.9947(1). The main weakness is the theoretical basis for the broad generality claim: the three-level model uses a non-unitary approximate transformation and a leakage-suppression condition that does not match the stated parameters. This does not invalidate the transmon demonstration, but it means the claim of applicability to any CDD qubit architecture is not rigorously supported. My proposed test directly checks whether the three-level truncation and the node condition are quantitatively predictive, which is the load-bearing assumption for the generality claim. I agree with the reader that the verdict should remain CONDITIONAL rather than REJECT, because the central experimental results are plausible and the identified gaps are addressable with additional simulations or a second device test.","tokens_in":12164,"tokens_out":30491,"duration_ms":259718,"concrete_test":"Recompute the leakage error using a unitarized version of T (normalize each column of Eq. 7) and include the fourth transmon level |3> in a numerical simulation of Eq. 1 with the experimental parameters (ACDD/2pi=23 MHz, Ag/2pi=29.12 MHz, tg=40 ns). If the predicted leakage to |2> changes by more than a factor of 2, or if leakage to |3> is non-negligible (>1e-4), the three-level truncation is insufficient. Then, on the fixed-frequency transmon or a device with EC/h ≈ 200 MHz, measure RB fidelity and leakage using the same protocol; if the optimal tg deviates from the node-condition prediction by more than 10%, the claimed generality is not quantitatively supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that the gating scheme is applicable to any CDD qubit architecture relies on the three-level analysis in Sec. III B-D (Eqs. 6-9). The approximate dressed-basis transform T in Eq. 7 is not unitary; normalization corrections are neglected, so the leakage couplings in Eq. 9 are only leading-order in beta = hbar*ACDD/EC. The paper states that suppressing leakage is 'roughly equivalent to making EC/hbar*ACDD = 7 an integer ratio,' but for the experimental parameters EC/h = 137 MHz and ACDD/2pi = 23 MHz, the ratio EC/(hbar*ACDD) = 137/23 ≈ 5.96, not 7. This discrepancy is unexplained. The pulse tuning procedure in Sec. III D uses this approximate model to select tg before experimental calibration. If the model is quantitatively wrong, the optimal pulse parameters for a different CDD platform (different anharmonicity, level structure, or drive coupling) cannot be predicted, so the claimed universality to any CDD qubit architecture is not established. The empirical demonstration on the tunable transmon stands, but the broad applicability claim is not supported by the presented theory.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper develops a universal single-qubit gating scheme for qubits encoded in continuously driven, dynamically decoupled (CDD) dressed states. The scheme uses two fixed-amplitude, fixed-duration pulses separated by wait times, with the wait times implementing R_z rotations and the pulses providing Rxz(π/2) rotations. The authors give a two-level analysis, a three-level transmon analysis with a leakage-suppression criterion, a pulse-tuning procedure, and randomized benchmarking on a flux-tunable transmon, reporting Clifford fidelity F = 0.9947(1) at a flux-sensitive point while increasing Ramsey and Hahn-echo coherence times by more than a factor of ten. A second, fixed-frequency transmon gives F = 0.9985. The paper concludes that the method should be applicable to any CDD qubit architecture.","tokens_in":12338,"tokens_out":13426,"duration_ms":122276,"significance":"If the stated scope is confirmed, this is a useful practical advance: gating dressed CDD qubits via perturbative Rabi pulses is slow, and the proposed wait-time-plus-fixed-pulse construction is simple and minimally parameterized once ACDD is calibrated. The experimental data are a strength: randomized benchmarking over 354 sequences, explicit error bars on the tunable device, coherence measurements on two devices, and an honest discussion of the fixed-frequency transmon's CDD T2E. The two-level gate construction is internally consistent and the central empirical claim is credible. However, the analytic multi-level model used to motivate the pulse parameters is approximate and contains a quantitative inconsistency in the leakage-matching condition, so the broad 'any CDD qubit architecture' claim is not yet supported.","major_comments":[{"comment":"The multi-level analysis that supports the claimed generality is explicitly approximate. The dressed-basis transformation T in Eq. (7) neglects normalization corrections, and the text states that the treatment is valid for β = ℏACDD/EC < 0.2. Consequently the leakage couplings in Eq. (9) are only leading order in β. Since the pulse-tuning procedure in Sec. III D uses this model to select ACDD and tg, the abstract's claim that the gates are 'applicable to any CDD qubit architecture' is not established for platforms with different anharmonicity or level structure. Please either restrict the scope to weakly anharmonic, transmon-like systems, or supply a unitarized higher-order dressed-basis analysis that quantifies leakage outside the β < 0.2 regime. The experimental demonstration itself is not invalidated by this issue.","section":"Sec. III B–D, Eqs. (6)–(9)"},{"comment":"The stated leakage-matching criterion is quantitatively inconsistent with the reported experimental parameters. The text says that suppressing leakage is 'roughly equivalent to making EC/ℏACDD = 7 an integer ratio' and later refers to 'our choice of β = 1/7.' With EC/h = 137 MHz and ACDD/2π = 23 MHz, one obtains β = ℏACDD/EC = 23/137 ≈ 0.168, i.e., EC/ℏACDD ≈ 5.96, not 7. This is not a small rounding discrepancy: it changes the node-matching condition that motivates the choice of ACDD and tg in Sec. III D. Please reconcile the analytic criterion with the data, or state clearly that the final parameters were chosen from the full numerical simulation (Fig. 5) rather than from the integer-ratio rule.","section":"Sec. III C–D and Sec. II B"},{"comment":"The error budget assigns the largest contribution, 4.1 × 10^-3 out of 5.6 × 10^-3, to 'slow changes to the static value of Δ from flux drift,' but no direct measurement of this drift or a control experiment (e.g., repeated recalibration of Δ, or interleaved randomized benchmarking with active flux correction) is presented. Because this attribution is the difference between the claimed gate performance and a closed error budget, it should be supported by data or explicitly labeled as a hypothesis. The fixed-frequency transmon result strengthens the plausibility, but the error budget on the tunable device remains open.","section":"Sec. IV"}],"minor_comments":[{"comment":"The definition of the 'speed limit' is vague: the text refers to the 'lowest tg along any 50:50 superposition contour in Fig. 5c-d,' but the speed limit that sets the comparison with perturbative Rabi gates should be stated quantitatively (e.g., in terms of the pulse area or the Margolus–Levitin bound).","section":"Sec. III A"},{"comment":"The factor 1.3 in the pulse envelope is not explained; please state whether it normalizes the peak amplitude and how it was chosen.","section":"Eq. (5)"},{"comment":"The formula for the average Clifford time, 24(tg + tc) + 9(2π/ACDD))/24, needs a brief derivation, and the placement of the opening parenthesis should be corrected for readability.","section":"Table I / Appendix A"},{"comment":"The fixed-frequency transmon result F = 0.9985 is quoted without an uncertainty or a description of the randomized-benchmarking protocol; please include the same analysis as for the tunable device.","section":"Sec. IV"},{"comment":"The statement that no |f⟩ population was observed in single-shot measurements after randomized benchmarking is weak evidence against leakage, because leaked population can decay back into the qubit subspace; an interleaved leakage measurement would be more convincing.","section":"Sec. III C"},{"comment":"The universal decomposition in Eq. (4) is taken from Ref. [23] without derivation in this paper; please state this provenance explicitly in the main text so the incremental contribution of the present work is clear.","section":"Sec. III A, Eq. (4)"},{"comment":"Reference [13] and reference [35] are the same paper (Yan et al., Nature Communications 7, 12964 (2016)); please merge or cite them distinctly.","section":"References"},{"comment":"The figure caption contains LaTeX artifacts ('/uni03BCs') that should be corrected; also specify whether the CDPQ Ramsey sequence included the initialization and readout pulses.","section":"Fig. 2"}],"recommendation":"major_revision","confidential_remarks":"The methodological overlap with Ref. [23], which shares an author and supplies both the universal gate decomposition and the randomized-benchmarking protocol, deserves explicit discussion. The present manuscript should state what is new relative to that work: the CDD-specific pulse construction, the three-level leakage analysis, the adiabatic initialization, and the transmon experiments. This is mainly a framing and novelty-disclosure concern rather than a correctness concern."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the meaningful new thing here is empirical — a CDD-protected transmon at a flux-sensitive point gets single-qubit Clifford gates at 93 ns with fidelity 0.9947(1), while retaining the expected coherence protection. That is a concrete step beyond prior CDD coherence demonstrations and beyond the small-gap universal-control work in Refs. [23,24]. The paper credits that lineage clearly.\n\nThe gate construction (Eq. 4, the two-pulse control) is inherited from Campbell et al., but the adaptation to a low-anharmonicity CDD system is genuinely new: the adiabatic initialization to handle the dressed |f> contamination, the three-level leakage analysis, and the simulated pulse tuning that suppresses leakage. The coherence data (10x improvement, Hahn echo 4.4 to 51 microseconds, Ramsey 1.29 to 31 microseconds) are clean and internally consistent. The RB measurement looks standard, and the error budget is plausible, though not error-barred.\n\nSoft spots, in rough order of importance. First, the paper claims the scheme is 'applicable to any CDD qubit architecture,' but the three-level model it rests on is visibly approximate. Eq. 7 neglects normalization in order beta, so the leakage couplings in Eq. 9 are leading-order only. More concretely, the text says the leakage-suppression condition is 'roughly equivalent to making EC/hbar*ACDD = 7 an integer ratio,' but for the stated experimental parameters (EC/h = 137 MHz, ACDD/2pi = 23 MHz) that ratio is about 5.96, not 7. The paper never explains that discrepancy. That does not invalidate the experiment — the leakage was checked via simulation and in-situ gate tuning — but it does undermine the generality claim until the model is made quantitative.\n\nSecond, the fidelity is demonstrated for the Clifford set only, not for arbitrary-angle rotations. That is fine for RB, but the 'universal' wording overstates what is directly verified.\n\nThird, no data or code are provided. The measurements are standard enough to reproduce, but a reader cannot check the tuning procedure or the leakage simulations without more.\n\nNone of these are fatal. The central demonstration — fast, high-fidelity gates on a CDD-protected transmon at a flux-sensitive point — is solid and useful to the quantum-control community. The theory is sloppier than the data deserve, but it is fixable.\n\nRecommendation: send it to review. The authors should be asked to reconcile the node-condition ratio, report the leakage calculation with proper normalization, and ideally address arbitrary-angle gates. With those changes, it is a solid contribution to a good journal.","headline":"Strong experimental result on fast CDD-transmon gates; the three-level leakage theory is sloppier than the data, but the core demonstration is real and worth refereeing.","tokens_in":12913,"tokens_out":3133,"would_cite":true,"duration_ms":28252,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["03.67.Lx","85.25.Cp"],"model":"deepseek-v4-flash","headline":"A continuously decoupled qubit can be gated fast and faithfully using just two fixed-amplitude pulses plus waits, demonstrated at fidelity 0.9947(1) on a flux-tunable transmon.","keywords":["continuous dynamical decoupling","spin locking","single-qubit gates","transmon","superconducting circuits","randomized benchmarking","leakage suppression","dressed qubit"],"falsifier":"Run the two-pulse gate sequence on a CDD-protected qubit with $\\beta \\geq 0.2$ while sweeping $A_{\\mathrm{CDD}}$ and measuring leakage into the second excited state; if the Clifford fidelity at the designed Fourier nodes does not stay near the reported level, the dressed-basis approximation and its leakage-suppression criterion break down. A cleaner check is to port the same recipe to a non-transmon CDD platform and verify that the predicted node structure in the leakage versus $A_{\\mathrm{CDD}}$ curve appears where the paper's tuning rule says it should.","tokens_in":11930,"feed_emoji":"⚛️","tokens_out":9838,"duration_ms":81332,"temperature":0.7,"pith_summary":"This paper establishes that qubits protected by continuous dynamical decoupling—where a steady drive hybridizes the two logical states into a noise-insensitive dressed pair—can be controlled by universal single-qubit gates that are fast rather than perturbatively slow. The recipe is to apply two fixed-amplitude pulses, each roughly one drive-period long, separated by a wait time that implements the needed phase rotation; any single-qubit unitary follows by composing these blocks. The authors demonstrate the scheme on a flux-tunable transmon deliberately biased where it is most sensitive to flux noise, obtaining a randomized-benchmarking Clifford fidelity of $\\mathcal{F}=0.9947(1)$ and coherence-time improvements of more than a factor of ten over the undriven device. The result matters because it turns continuous dynamical decoupling's noise protection from a measurement trick into a practical basis for quantum information processing in noisy environments.","feed_headline":"Two-pulse recipe gates CDD qubits at 0.9947 fidelity","feed_subtitle":"Universal control now works near the speed limit on qubits continuously shielded from dephasing noise.","key_machinery":"The load-bearing object is the dressed-qubit decomposition of the driven Hamiltonian: with the CDD drive along $\\sigma_z$, the gate pulse $A(t)$ and detuning $\\Delta(t)$ act as orthogonal transverse fields, $H/\\hbar = A_{\\mathrm{CDD}} \\sigma_z/2 + A(t) \\sigma_x/2 + \\Delta(t) \\sigma_y/2$. Universal control is assembled from a single two-pulse primitive, $U = R_{xz}(\\pi/2,\\gamma) R_z(\\gamma') R_{xz}(\\pi/2,\\gamma) R_z(\\gamma'')$, where the $R_z$ rotations are implemented for free by waiting (phase advances at rate $A_{\\mathrm{CDD}}$) and negative phases use the complement time $t_c = 2\\pi/A_{\\mathrm{CDD}} - \\gamma/A_{\\mathrm{CDD}}$. On the three-level transmon, an approximate dressed-basis transformation (valid for $\\beta = \\hbar A_{\\mathrm{CDD}}/E_C < 0.2$) exposes the leakage couplings. The pulse envelope of Eq. (5) is then chosen so that the Fourier transform has nodes at the $|\\pm\\rangle\\leftrightarrow|f\\rangle$ transition frequencies, which suppresses leakage to the $\\sim 3\\times 10^{-4}$ level.","core_discovery":"The central claim is that continuous dynamical decoupling need not be a spectator: the same drive that creates the protected dressed basis $|\\pm\\rangle = (|0\\rangle \\pm |1\\rangle)/\\sqrt{2}$ can be supplemented by gate pulses at the same frequency, 90 degrees out of phase, to perform universal single-qubit rotations. In an ideal two-level system, two such pulses, each with duration $t_g \\approx 2\\pi/A_{\\mathrm{CDD}}$ and amplitude $A_g \\approx A_{\\mathrm{CDD}}/2$, plus a wait time $\\delta t$ that advances the phase at rate $A_{\\mathrm{CDD}}$, generate any unitary of the form $U(\\gamma',\\gamma'') = R_{xz}(\\pi/2,\\gamma) R_z(\\gamma') R_{xz}(\\pi/2,\\gamma) R_z(\\gamma'')$. On the transmon, the non-computational $|f\\rangle$ state is kept quiet by choosing $A_{\\mathrm{CDD}}$ and the pulse envelope so that the $|+\\rangle\\leftrightarrow|f\\rangle$ and $|-\\rangle\\leftrightarrow|f\\rangle$ transitions sit at nodes of the pulse's Fourier spectrum. With this tuning the authors measure an average Clifford fidelity of $0.9947(1)$ at gate durations near 93 ns, with the residual error dominated by slow flux drift rather than leakage or gate speed.","pith_inferences":["Applied to higher-anharmonicity superconducting qubits, the same protocol could use larger $A_{\\mathrm{CDD}}$ and therefore yield even faster gates and stronger noise protection than the transmon's $\\beta<0.2$ tuning limit permits.","Since slow flux drift rather than leakage or gate speed dominates the reported error, periodically measuring and correcting the flux offset—or using a fixed-frequency device, where the paper reports a Clifford fidelity of $0.9985$—could push CDPQ fidelities above $0.999$.","If coupled CDPQs inherit the frequency insensitivity demonstrated here, two-qubit gates between CDPQs at different frequencies should exhibit suppressed $\\sigma_z\\sigma_z$ crosstalk, potentially eliminating tunable couplers from superconducting processor designs."],"forward_implications":["Any CDD-protected qubit whose two-state transition can be isolated from other levels can be gated by the same two-pulse-plus-wait recipe, without requiring slow perturbative Rabi pulses.","On the demonstrated transmon, Clifford gate durations of roughly 93 ns at fidelity $0.9947(1)$ leave room for hundreds of gate operations within the CDPQ Ramsey coherence time of $31(2)\\,\\mu$s.","At the flux-sensitive bias point $\\varphi = 0.367$, the CDD protection raises Ramsey and Hahn echo coherences from $1.29(4)\\,\\mu$s and $4.4(2)\\,\\mu$s to $31(2)\\,\\mu$s and $51(7)\\,\\mu$s, respectively.","Because the gate speed is set by the drive amplitude $A_{\\mathrm{CDD}}$ rather than by a weak Rabi rate, larger drive amplitudes offer both stronger noise protection and faster gates, with leakage managed by the Fourier-node tuning condition."],"supporting_citations":[{"why":"supplies the two-pulse universal-control decomposition and the randomized-benchmarking sequence used and adapted here to the CDD-protected qubit.","marker":"[23]"},{"why":"introduces the notion of operating a continuously driven pair of states as a qubit, the foundation of the CDPQ concept.","marker":"[8]"},{"why":"demonstrates pulsed spin-locking preparation and the coherence limits for continuously driven superconducting qubits that this work builds on.","marker":"[12]"},{"why":"provides an independent demonstration of universal control for small-gapped systems, supporting the generality of the approach.","marker":"[24]"},{"why":"defines the quantum speed limit used to quantify how fast the two-pulse gates are relative to perturbative Rabi gates.","marker":"[25]"},{"why":"supplies the transmon Hamiltonian and anharmonic level structure underlying the three-level leakage analysis.","marker":"[27]"},{"why":"introduces randomized benchmarking, the method used to extract the reported Clifford fidelity.","marker":"[30]"}],"fun_headline_variants":["Fast universal gates for CDD qubits at 0.9947","Two-pulse gates on CDD transmon hit 0.9947","Universal control for CDD qubits at 0.9947 fidelity","Fast single-qubit gates for CDD systems: 0.9947","CDD qubit gates at 0.9947: two-pulse recipe"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The scheme depends on being able to choose the continuous-drive amplitude and the gate-pulse envelope so that transitions to the second excited state land exactly at nodes of the pulse's Fourier spectrum, and on the small-drive approximation ($\\beta = \\hbar A_{\\mathrm{CDD}}/E_C < 0.2$) used to design that tuning; if another CDD platform cannot satisfy this condition, the claimed universality to any CDD qubit architecture fails, although the transmon demonstration could still stand.","fun_headline_variants_meta":{"raw":{"variants":["Fast universal gates for CDD qubits at 0.9947","Two-pulse gates on CDD transmon hit 0.9947","Universal control for CDD qubits at 0.9947 fidelity","Fast single-qubit gates for CDD systems: 0.9947","CDD qubit gates at 0.9947: two-pulse recipe"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000526,"raw_usage":{"total_tokens":2581,"prompt_tokens":1027,"completion_tokens":1554,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":643,"completion_tokens_details":{"reasoning_tokens":1454}},"tokens_in":643,"tokens_out":1554,"duration_ms":11145,"temperature":1.0,"reasoning_tokens":1454,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T14:32:51.804287+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the two-pulse gate sequence on a CDD-protected qubit with $\\beta \\geq 0.2$ while sweeping $A_{\\mathrm{CDD}}$ and measuring leakage into the second excited state; if the Clifford fidelity at the designed Fourier nodes does not stay near the reported level, the dressed-basis approximation and its leakage-suppression criterion break down. A cleaner check is to port the same recipe to a non-transmon CDD platform and verify that the predicted node structure in the leakage versus $A_{\\mathrm{CDD}}$ curve appears where the paper's tuning rule says it should.","supporting_citations":[{"cited_title":"Huang, P","cited_arxiv_id":null,"evidence_quote":"supplies the two-pulse universal-control decomposition and the randomized-benchmarking sequence used and adapted here to the CDD-protected qubit."},{"cited_title":"Yoshihara, Y","cited_arxiv_id":null,"evidence_quote":"introduces the notion of operating a continuously driven pair of states as a qubit, the foundation of the CDPQ concept."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides an independent demonstration of universal control for small-gapped systems, supporting the generality of the approach."},{"cited_title":"Zhang, S","cited_arxiv_id":null,"evidence_quote":"defines the quantum speed limit used to quantify how fast the two-pulse gates are relative to perturbative Rabi gates."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the transmon Hamiltonian and anharmonic level structure underlying the three-level leakage analysis."}],"review_version":1}