{"id":"e4959eed-a24d-4cd6-afef-815f2cc6128b","arxiv_id":"2509.04776","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"Proposed transmon-coupler schemes for integer fluxoniums achieve simulated controlled-Z coherent errors near 1e-6 in tens of nanoseconds, plus static ZZ cancellation.","lead":"Two new simulated gate designs let a recent type of superconducting qubit, the integer fluxonium, perform two-qubit controlled-Z operations with very low predicted error. The designs route the interaction through a tunable transmon coupler, which also cancels the always-on ZZ crosstalk between idle qubits.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Adiabatic gate's 1e-6 error relies on exact Hamiltonian knowledge and waveform fidelity; no sensitivity analysis is provided despite the authors' own caveat that parameter variations and waveform distortion are unavoidable.","rationale":"The paper is a carefully executed theory proposal with shared code, high truncation levels (150 fluxonium levels, 50 charge states), and transparent discussion of limitations. The coherent-error numbers in Fig. 3(a) and Fig. 5(a) appear plausible given the simulation methodology and the use of standard tools (scqubits, QuTiP). I do not see a clear internal inconsistency in the simulations themselves. The most defensible load-bearing concern is the robustness of the adiabatic gate's leakage suppression to exact Hamiltonian knowledge and pulse fidelity. The reader's weakest_assumption identifies precisely this issue, and I agree. The paper explicitly says that calibration is future work, which is honest but leaves the headline claim conditionally dependent on the simulation model. The microwave gate also relies on parameter values, but it includes a robustness study against spectator coupling and a broad optimal flux range, whereas the adiabatic gate has no sensitivity analysis at all. The |3>-state coherence issue is real but affects total error, not the coherent-error claim, so it is secondary for the central claim. The proposed concrete test would settle whether the concern actually lands: if the error remains low under realistic perturbations, the abstract survives; if not, the numbers should be described as ideal-model predictions rather than device-level targets.","tokens_in":28560,"tokens_out":14088,"duration_ms":127650,"concrete_test":"Recompute the adiabatic CZ gate error for pulse P3 (32-ns edge, 65.5-ns total) from Fig. 3(a) using the same pulse waveform but with perturbed Hamiltonian parameters: vary each of E_C, E_J, E_L of both fluxoniums and the coupler by ±1%, and J_12, J_c1, J_c2 by ±5%, in a Monte Carlo sweep of at least 100 samples. Additionally, pass the flux pulse through a low-pass filter with a 1-GHz cutoff to model waveform distortion. If the 95th-percentile CZ error remains below 1e-5, the 1e-6 claim is robust; if it exceeds 1e-4, the concern lands and the headline should be framed as an ideal-model result.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central 1e-6 coherent-error numbers for the adiabatic CZ gate (Sec. III B) come from a pulse whose edges are designed using the D_ij factors of Eq. (5) to maintain a constant leakage rate, and whose flat duration is chosen so that residual leakages from the two edges interfere destructively. Both design steps require quantitative knowledge of every matrix element in the full system Hamiltonian, including coupling strengths J_12, J_c and transition matrix elements n_i,01, n_c,01, n_c,12. The authors explicitly state (Sec. III C) that 'parameter variations are almost inevitable' and 'waveform distortion is also unavoidable,' and that calibrating the edge shape experimentally is outside the scope of the work. No sensitivity analysis is presented: the reported errors are for the exact model only. Because the flat-duration optimization works via interference of leakage amplitudes (Fig. 9(a)), even a small parameter shift can detune the interference condition and raise the leakage far above 1e-6. This is the single most load-bearing assumption behind the headline coherent-error claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper proposes two controlled-Z gate schemes for a fluxonium-transmon-fluxonium (FTF) architecture in which both fluxoniums are integer fluxoniums operated at zero flux bias. The authors first show that the static ZZ interaction can be cancelled by a suitable choice of coupler parameters, and then design two gates: a flux-activated adiabatic CZ gate whose pulse edges maintain a constant leakage rate and whose flat duration is optimized for destructive interference of residual leakage, and a microwave-activated CZ gate driven through the coupler charge line that exploits level repulsion between the coupler and the fluxonium third-excited states. For the adiabatic gate, coherent errors below 1e-4, 1e-5, and 1e-6 are reported for gate durations below about 40, 50, and 70 ns, respectively; for the microwave gate, coherent errors of 1e-5, 1e-6, and 1e-7 are reported at 50, 70, and 100 ns. The authors include T1 dissipation and quasistatic 1/f flux noise in their error budgets and argue that CZ fidelities above 99.99% are feasible if the qubit T1 is around 500 microseconds. A hybrid integer-fluxonium/conventional-fluxonium system is also discussed qualitatively. The numerical work uses scqubits and QuTiP, and the simulation code is shared on GitHub.","tokens_in":28799,"tokens_out":17350,"duration_ms":160585,"significance":"If the reported error levels are robust, this is a useful proposal for entangling gates in a relatively new qubit platform: it extends the transmon-coupler toolbox to integer fluxoniums, demonstrates static-ZZ cancellation with small delocalization, and reduces crosstalk by using a dedicated coupler line for the microwave gate. The paper also provides a detailed capacitance-matrix analysis for scalable 1D and 2D architectures, and the availability of the simulation code is a strength. The central results are numerical predictions for specified circuit Hamiltonians rather than fits to external data, so they are not circular by construction. However, the headline coherent-error numbers are obtained for the nominal Hamiltonian and optimized pulse parameters, and the robustness of these numbers to parameter uncertainty and waveform distortion is not quantified; this is the main risk to the central claims.","major_comments":[{"comment":"The coherent-error numbers for the adiabatic CZ gate (below 1e-4, 1e-5, and 1e-6 for gate durations below about 40, 50, and 70 ns in Fig. 3(a)) are computed for the nominal Hamiltonian only. The constant-leakage-rate pulse edge is constructed from the D_ij functions in Eqs. (5)-(6), which require quantitative knowledge of all matrix elements and level spacings, and the flat-duration optimization suppresses leakage by destructive interference of the residual leakage amplitudes from the two edges (Fig. 9(a) and Appendix C 1). The authors state in Sec. III C that parameter variations are almost inevitable and waveform distortion is unavoidable, and that experimental calibration of the edge shape is outside the scope. No sensitivity analysis is provided for typical fabrication spreads in E_J, E_C, E_L, J_c, or J_12, nor for finite-bandwidth distortion of the flux pulse. Since the reported 1e-5 and 1e-6 levels are below the scale at which such imperfections will dominate, the headline claim is not yet supported for a physical device. I request a robustness scan that quantifies the tolerable parameter deviations, or a clear rephrasing of these numbers as ideal-model results.","section":"Sec. III B and III C"},{"comment":"The microwave-activated gate has the same model-exactness limitation. The coherent errors of 1e-5 at 50 ns, 1e-6 at 70 ns, and 1e-7 at 100 ns (Fig. 5(a)) are obtained after optimizing both the drive frequency and the drive amplitude for a specific set of energy levels and matrix elements. The only sensitivity study reported (Appendix D, Fig. 11(d)) concerns a spectator-induced 0.15 MHz shift of the coupler frequency; it does not cover fabrication uncertainty in E_J,c, E_C,c, or in the fluxonium |3>-state frequencies that produce the level repulsion. Because the authors propose that a fixed-frequency transmon is feasible, the required frequency and parameter precision must be quantified. I recommend adding an offset-sensitivity analysis, for example gate error versus target-transition-frequency offset, or explicitly stating the precision requirement.","section":"Sec. IV C and Appendix D"},{"comment":"The quasistatic flux-noise error estimate used to obtain the total-error level near 1e-4 is computed by assigning a static flux offset with variance sigma_phi^2 = integral of A_phi^2/f over f_IR = 1e-6 Hz to f_UV = 1 GHz. This is internally inconsistent with the quasistatic assumption: noise components with frequencies comparable to or larger than the gate rate are not static on the gate timescale, and extending the integral to 1 GHz is not justified. The resulting estimate should be presented as a rough order-of-magnitude model, with a cut-off choice limited to frequencies below roughly 1/T_g or with an explicit justification of the chosen band. The authors' caveat in Sec. VI that the noise spectrum may differ in experiment partially mitigates this concern, but as written the 1e-4 bound is not derived from a well-defined noise model.","section":"Appendix C 2 and Sec. III C"}],"minor_comments":[{"comment":"The heading 'Comparision' should be 'Comparison'.","section":"Sec. IV D"},{"comment":"The sentence 'the quasi-static flux-noise-induced error can reach below x10^{-4}' appears to be missing a leading factor; it should read 'below 1x10^{-4}'.","section":"Sec. III C"},{"comment":"The statement that the pulse amplitude is set to 1/(A N_T) to achieve a 2pi rotation is dimensionally incomplete; a 2pi pulse requires integral of epsilon(t) N_T dt = 2pi, so either the definition of A or the formula should include the 2pi factor, or the normalization should be clarified.","section":"Sec. IV B"},{"comment":"The 1e-6 coherent error for the microwave gate is achieved at 70 ns duration, not at 50 ns; the phrase 'several tens of nanoseconds' is defensible but should be tied to the simulated durations to avoid overstatement.","section":"Abstract and Sec. IV C"},{"comment":"The definition sigma_phi = sqrt(<delta_phi_ext,c>) should be sigma_phi^2 = <delta_phi_ext,c^2>, i.e., sigma_phi = sqrt(<delta_phi_ext,c^2>).","section":"Appendix C 2"},{"comment":"The conclusion cites transmon T1 values above 100 microseconds (Ref. [52]) to support the feasibility of a 99.99% CZ gate, but the simulations require fluxonium T1 around 500 microseconds; this argument should be rephrased or supported by fluxonium coherence data.","section":"Sec. VI"},{"comment":"Flux bias is written in inconsistent forms, for example 'phi_ext,c/2pi = 0.21' and 'phi_ext,c = 0.22 x 2pi'; please use a single convention consistently.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of the journal and the numerical study is careful in many respects. The key revision I would require is a sensitivity analysis for both gate schemes with respect to parameter deviations and control distortions; if such an analysis is not possible, the abstract and conclusions should be adjusted to present the 1e-6 and 1e-7 coherent errors as ideal-model results rather than device-level predictions. I do not see a circularity problem in the central results. I would also ask the editor to verify that the GitHub repository linked in Ref. [53] is accessible, since it is central to reproducibility."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new bit here is that two-qubit gates for integer fluxoniums have been missing from the literature, and this paper fills the gap with two concrete schemes: a flux-activated adiabatic CZ gate and a microwave-activated non-adiabatic CZ gate, both using a transmon coupler. The ZZ-cancellation analysis is clean, the use of the constant-leakage-rate pulse edge and flat-duration interference is well motivated, and the error budget includes T1 and quasistatic flux noise. The simulations look careful, and they shared the code. That is real evidence of a solid working proposal, not a sketch.\n\nWhere the paper is soft is exactly where the reader and stress-test put their fingers. The adiabatic gate's 1e-6 coherent error depends on knowing every matrix element in the Hamiltonian to design the pulse edge, and the flat-duration suppression works by interfering leakage amplitudes. The authors themselves say parameter variations and waveform distortion are unavoidable and calibration is out of scope, but they do not quantify how the error degrades under fabrication mistargeting or pulse distortion. That is a load-bearing premise, and the lack of sensitivity analysis means the headline number should be read as a best-case simulation, not a robustness claim. The microwave scheme also leans on the |3> state coherence of the fluxonium, which they correctly flag as experimentally unexplored. The more realistic total error of around 1e-4, once decoherence is folded in, is stated honestly, so this is not a case of overclaiming: the abstract's 1e-6 is clearly labeled as coherent error in an ideal model.\n\nThe spectator-qubit analysis in Appendix D is a nice attempt to probe robustness, and the reduced-coupling benchmark in Fig. 9 goes in the right direction, but neither replaces a proper parameter-variation study. That said, the central physics holds up: the FTF architecture can suppress static ZZ, the two gate schemes are distinct and plausible, and the limitations are stated rather than hidden. A serious referee could push for a sensitivity analysis and a more careful statement of what the quoted fidelities assume.\n\nWho is this for? Mostly superconducting-circuit theorists and experimentalists working on fluxonium processors, especially anyone wondering whether integer fluxoniums can be used for entangling gates. I think it deserves a real peer review, not a desk rejection. If I were refereeing, I would ask for the robustness analysis before accepting the headline numbers as device targets, but the work itself is honest and useful.","headline":"A genuinely new two-qubit gate proposal for integer fluxoniums with careful numerics, but the headline 1e-6 coherent error is not yet a device target without a sensitivity analysis.","tokens_in":29362,"tokens_out":1786,"would_cite":true,"duration_ms":19703,"reading_group":"yes","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":"Using a transmon coupler between two integer fluxonium qubits, this paper proposes two controlled-Z gate schemes—one flux-activated and one microwave-activated—that reach coherent errors around 1e-6 within tens of nanoseconds.","keywords":["integer fluxonium qubit","transmon coupler","controlled-Z gate","ZZ interaction suppression","adiabatic flux pulse","microwave-activated gate","flux noise","superconducting qubit architecture"],"falsifier":"Fabricate an FTF pair with the paper's quoted parameters, measure the static ZZ at the predicted cancellation flux, and run the designed 70 ns adiabatic pulse; if the residual ZZ is not near zero or the leakage error exceeds $10^{-6}$ by more than a small factor under calibrated control, the central claim fails.","tokens_in":28358,"feed_emoji":"⚛️","tokens_out":8199,"duration_ms":71678,"temperature":0.7,"pith_summary":"This paper proposes a route to high-fidelity two-qubit gates for integer fluxoniums, a protected superconducting qubit that has so far demonstrated single-qubit control but no entangling gates. The authors argue that coupling two integer fluxoniums through a tunable transmon coupler can completely cancel the always-on ZZ coupling between the qubits at the idling point, and that the same coupler can implement two different controlled-Z gates: a flux-pulse adiabatic gate and a microwave-activated non-adiabatic gate. In simulation, both schemes reach coherent errors around $10^{-6}$ within roughly 50 to 70 ns, and after including relaxation and $1/f$ flux noise the estimated total error stays near $10^{-4}$, suggesting 99.99% CZ fidelities are within reach for current coherence times. The same architecture also extends to a hybrid integer-fluxonium/conventional-fluxonium system, which broadens the frequency range available for scalable processors.","feed_headline":"Integer fluxonium CZ gates hit 1e-6 error in tens of nanoseconds","feed_subtitle":"A transmon coupler cancels idle ZZ crosstalk and enables both flux-pulse and microwave entangling gates.","key_machinery":"The load-bearing objects are the leakage-rate factors $D_{ij}$ defined in Eq. (5), which measure how strongly each computational state couples to all other eigenstates during a flux ramp; the pulse edge is shaped so that the flux velocity is inversely proportional to the summed factor, giving a constant leakage rate, and the flat duration is then tuned so residual leaked populations interfere destructively at the end. For the static coupling, the ZZ strength is decomposed into second-, third-, and fourth-order perturbative terms, and the coupler parameters are chosen so these terms cancel. For the microwave gate, the machinery is the repulsion of the coupler's first excited state by the fluxoniums' third excited states, which creates a selective drive target, plus a drive detuning that compensates the ac-Stark phase deviations.","core_discovery":"Working at zero external flux, two integer fluxoniums coupled through a transmon coupler can have their static ZZ coupling eliminated by destructive interference among second-, third-, and fourth-order perturbative contributions, while the computational states remain only weakly hybridized. The paper then claims two coupler-control CZ schemes: an adiabatic scheme in which a flux pulse with a constant-leakage-rate edge and an optimized flat plateau accumulates the conditional $\\pi$ phase, reaching coherent errors below $10^{-4}$, $10^{-5}$, and $10^{-6}$ for gate durations under 40, 50, and 70 ns respectively; and a microwave-activated scheme in which driving the coupler's 0-1 transition through a full Rabi cycle acquires a geometric phase, with the coupler level repelled by the fluxoniums' third excited states, reaching $10^{-5}$ error at 50 ns and lower at longer durations. Both schemes keep the coupler as the only driven element, and with relaxation and flux noise taken into account the paper estimates total CZ errors at the $10^{-4}$ level, projecting fidelities above 99.99% for qubit $T_1$ above 500 microseconds.","pith_inferences":["The constant-leakage-rate edge construction depends only on the $D_{ij}$ factors and the ZZ curve, so it should transfer to other tunable-coupler architectures whose level structure is not transmon-like, provided those factors are measured or modelled.","The perturbative cancellation picture suggests a design rule: fix the qubit-coupler and direct qubit-qubit capacitances so the second-, third-, and fourth-order ZZ contributions interfere destructively, which could be used to engineer ZZ-free idling in larger lattices.","The microwave scheme's 99.99% projection relies on the fluxonium third-excited states having useful coherence, which the paper notes is experimentally uncharacterized; a short |3>-state lifetime would tighten the achievable fidelity bound.","A natural next step is to calibrate the pulse edge derived from $D_{ij}$ experimentally on a single FTF pair and compare the measured leakage against the predicted curve; agreement would validate using the same method for spectator-heavy multi-qubit circuits."],"forward_implications":["If the simulated errors hold, integer fluxonium qubits can be entangled with CZ fidelities above 99.99%, matching or exceeding current transmon two-qubit gates.","The static ZZ cancellation works across a range of coupling strengths, so idling qubits in a multi-qubit array can be left coupled without a residual always-on error.","The adiabatic scheme's flat-duration optimization cuts leakage by up to about three orders of magnitude over a fixed-edge pulse, so pulse shaping alone can close most of the coherent-error budget.","The microwave scheme's single charge line on the coupler removes the need for simultaneous drives on both data qubits, reducing crosstalk and calibration overhead in scaled-up circuits.","In the hybrid IF-T-F system, both gate families remain viable with redesigned spectra, allowing integer and conventional fluxoniums to coexist in one processor without frequency crowding."],"supporting_citations":[{"why":"Defines the integer fluxonium qubit and shows its single-qubit fidelity above 99.9%, the platform this work builds on.","marker":"[1]"},{"why":"Demonstrated high-fidelity fluxonium gates with a transmon coupler and the ZZ-suppression mechanism this work extends to integer fluxoniums.","marker":"[28]"},{"why":"Supplies the constant-leakage-rate edge construction via $D_{ij}$ factors used to shape the adiabatic flux pulse.","marker":"[45]"},{"why":"The all-transmon tunable-coupling CZ gate whose architecture and adiabatic method are adapted and compared against.","marker":"[13]"},{"why":"Earlier microwave-activated controlled-Z gate on low-frequency fluxoniums, the approach the microwave scheme refines with single-line coupler drive.","marker":"[32]"},{"why":"Experimental coupler-based microwave-activated controlled-phase gate on fluxoniums that motivates the second scheme.","marker":"[14]"},{"why":"Defines the delocalization parameter used to quantify residual qubit hybridization and crosstalk.","marker":"[49]"}],"fun_headline_variants":["Transmon coupler kills ZZ, enables 1e-6 CZ for integer fluxoniums","Fluxonium-transmon CZ: 1e-6 error in tens of ns","Two CZ schemes for integer fluxoniums: 1e-6 error, tens of ns","Zero-flux fluxonium CZ: 1e-6 error via transmon coupler","Adiabatic and microwave CZ for integer fluxoniums at 1e-6"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The headline coherent-error numbers assume the real circuit matches the model Hamiltonian, so the precomputed constant-leakage-rate pulse stays optimal; the authors state that fabrication parameter spread and flux-line waveform distortion will inevitably disturb this pulse.","fun_headline_variants_meta":{"raw":{"variants":["Transmon coupler kills ZZ, enables 1e-6 CZ for integer fluxoniums","Fluxonium-transmon CZ: 1e-6 error in tens of ns","Two CZ schemes for integer fluxoniums: 1e-6 error, tens of ns","Zero-flux fluxonium CZ: 1e-6 error via transmon coupler","Adiabatic and microwave CZ for integer fluxoniums at 1e-6"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00172,"raw_usage":{"total_tokens":6835,"prompt_tokens":1011,"completion_tokens":5824,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":627,"completion_tokens_details":{"reasoning_tokens":5703}},"tokens_in":627,"tokens_out":5824,"duration_ms":35968,"temperature":1.0,"reasoning_tokens":5703,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T16:27:17.903253+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Fabricate an FTF pair with the paper's quoted parameters, measure the static ZZ at the predicted cancellation flux, and run the designed 70 ns adiabatic pulse; if the residual ZZ is not near zero or the leakage error exceeds $10^{-6}$ by more than a small factor under calibrated control, the central claim fails.","supporting_citations":[{"cited_title":"The circuit diagram is shown in Fig","cited_arxiv_id":null,"evidence_quote":"Defines the integer fluxonium qubit and shows its single-qubit fidelity above 99.9%, the platform this work builds on."},{"cited_title":"DiCarlo, J","cited_arxiv_id":null,"evidence_quote":"Supplies the constant-leakage-rate edge construction via $D_{ij}$ factors used to shape the adiabatic flux pulse."},{"cited_title":"Wang, T.-Q","cited_arxiv_id":null,"evidence_quote":"Experimental coupler-based microwave-activated controlled-phase gate on fluxoniums that motivates the second scheme."}],"review_version":2}