{"id":"e63fcd81-fad0-4afc-935b-f2e6d182edf9","arxiv_id":"2509.04965","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Combining bipolar net-zero flux pulses with a tunable coupler yields a 99.48% CPhase gate with amplitude-controlled arbitrary phase angles and sub-5 kHz residual ZZ coupling.","lead":"A superconducting quantum processor achieved a 99.48% two-qubit gate by pairing bipolar net-zero flux pulses with a tunable coupler, keeping calibration simple and residual qubit coupling low. The method matters because easy, reproducible calibration of arbitrary two-qubit gates is a practical bottleneck for scaling quantum computers.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 99.9% simulation projection uses Q1's echo dephasing time T2E,gate=48.5 µs rather than its Ramsey time T2R,gate=4.5 µs; if the bipolar pulse echo does not fully refocus quasi-static flux noise during the |101>–|200> exchange, the sub-1e-3 error claim is optimistic.","rationale":"The reader's CONDITIONAL verdict is appropriate, and my read does not move it. The measured 99.48(9)% fidelity is supported by a self-consistent interleaved-XEB analysis: the reported per-cycle errors satisfy 0.81% ≈ 0.26% leakage + 0.08% coherent + 0.47% decoherence, and the p2/p1 extraction with D=4 yields the stated number when the reported cycle errors are treated as infidelities. The real soft spot is the extrapolation: Eq. (5) is solved with Q1 pure dephasing derived from T2E,gate=48.5 µs, while Table I gives T2R,gate=4.5 µs at the same operating point. The only warrant for this factor-of-10 choice is the claim that bipolar pulses echo out low-frequency flux noise. That claim is plausible and supported by prior net-zero pulse work, but it is not demonstrated for this three-body diabatic exchange, where the qubit spends part of the gate in |200> and the coupler follows a time-dependent trajectory. The proposed Monte Carlo test directly checks whether a static flux error cancels over the full bipolar sequence; if it does not, the simulated 3.4e-4 decoherence error and the 99.9% projection become unsupported, although the measured fidelity would stand. Secondary issues reinforce CONDITIONAL but are less load-bearing: the abstract mentions an 8-qubit processor validation that is absent from the body (only a 4-qubit processor is described), and the factor 1.5 in Eq. (4) is unexplained and shifts the inferred CPhase decoherence error by more than its uncertainty if varied by one.","tokens_in":16503,"tokens_out":16589,"duration_ms":170603,"concrete_test":"Run a Monte Carlo simulation of the full two-pulse CPhase gate (Eq. 1) with a per-shot static flux offset δ drawn from a Gaussian chosen to match the measured Ramsey decay at the gate point (T2R,gate=4.5 µs), applied to Q1's actual +V/-V waveforms and the coupler trajectory, and compute the average gate fidelity. If it remains near the T2E-based value of 3.4e-4, the echo assumption holds; if it approaches the T2R-limited scale (~1e-2), the 99.9% projection is unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The measured 99.48% XEB fidelity is not the weak point; the error-budget arithmetic is internally consistent. The load-bearing concern is the numerical projection that errors below 1e-3 are achievable (Section 'Leakage and decoherence contribution to gate errors', Eq. 5, Table I). The Lindblad simulation sets Q1's pure dephasing from T2E,gate=48.5 µs instead of T2R,gate=4.5 µs, invoking the built-in echo of the bipolar pulse. But the gate is not a simple free-precession echo: during each 20 ns pulse the system traverses the avoided crossing |101>↔|200> with time-dependent qubit and coupler detunings, and the +V and -V segments have different instantaneous frequencies and coupler participation. A quasi-static flux offset therefore need not cancel to first order when integrated over the two segments, particularly because the phase is accumulated while population is partially exchanged into |200>. If the effective dephasing during the exchange is closer to T2R than T2E, the simulated 1-F=3.4e-4 becomes an order of magnitude larger, and the claimed 99.9% projection is unsupported. This does not change the experimental fidelity claim, but it removes the paper's forward-looking error target.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents a flux-pulse scheme for CPhase gates on transmon qubits with a tunable coupler. A bipolar net-zero pulse pair drives a diabatic |101>–|200> exchange, and an intermediate weak bipolar pulse sets the conditional phase. The authors report a measured CPhase fidelity of 99.48(9)% from interleaved XEB on a two-qubit device, with residual ZZ below 5 kHz. They decompose the cycle error into leakage and purity components, and Lindblad simulations indicate a decoherence error of 3.4e-4 for 40-ns pulses, from which they claim 99.9% fidelity is achievable. The gate is also implemented on a 4-qubit processor with CZ fidelities 99.13–99.34%. The core experimental claim is directly benchmarked; the main concerns are the unsupported forward-looking simulation projection and several internal inconsistencies in the reported error-budget arithmetic.","tokens_in":16863,"tokens_out":12005,"duration_ms":114148,"significance":"If the measured result stands, the 99.48(9)% CPhase fidelity is competitive with the current state of the art, and the combination of tunable-coupler isolation (ZZ<5 kHz), bipolar net-zero pulsing, and a two-parameter calibration protocol is practically valuable. The paper includes a direct interleaved XEB measurement with explicit leakage and purity decomposition, and experimental validation on a 4-qubit processor. The main caveat is that the forward-looking '99.9% / error <1e-4' projection rests on an unvalidated echo-protection assumption; the measured central result is not affected by this caveat.","major_comments":[{"comment":"The Lindblad simulation sets Q1's pure dephasing from T2E,gate=48.5 us rather than T2R,gate=4.5 us, invoking the built-in echo of the bipolar pulse. This is load-bearing for the simulated 1-F=3.4e-4 and the subsequent '99.9% fidelity is achievable' statement. However, the gate is not a simple free-precession echo: during each 20-ns pulse the system traverses the |101>–|200> avoided crossing with time-dependent detunings and population exchange, and the +V and -V segments have different instantaneous frequencies and coupler participation. A quasi-static flux offset therefore need not cancel to first order. The text itself notes that the 20-ns delay is not echo protected. If the effective dephasing is closer to T2R,gate, the simulated error becomes roughly an order of magnitude larger and the 99.9% projection is unsupported. Please either rerun the simulation with a conservative T2R-based","section":"Leakage and decoherence contribution to gate errors (Eq. 5, Table I)"},{"comment":"The factor 1.5 in Eq. (4) is not derived. It is the only element that brings the extracted decoherence error (0.03(0.18)%) into agreement with the simulation (3.4e-4); with a factor of 1, the estimate is 0.18(0.18)%. Because the subsequent sentence states that the simulation 'compares well' with the experimental estimate, the choice of 1.5 appears to be made to match the simulation. Please provide an explicit timing/error-budget derivation for the factor, or replace the estimate with a direct measurement of the unprotected delay contribution.","section":"CPhase implementation (Eq. 4)"},{"comment":"The reported fidelity is inconsistent with Eq. (3). With reference error 0.29(2)% and interleaved cycle error 0.81(3)%, p=p2/p1=0.9948. Substituting into Eq. (3) with D=4 gives F=0.9961, i.e., 99.61%, not 99.48(9)%. The quoted 99.48% corresponds to p itself. Please correct Eq. (3), redefine D, or clarify the definition of 'fidelity'; the headline number must be consistent with the stated formula.","section":"CPhase implementation (Eq. 3, Fig. 2c)"},{"comment":"The abstract claims numerical simulations show an error below 1e-4 is achievable, while the body reports 1-F=3.4e-4 for the 40-ns gate and states '99.9% fidelity is achievable' (1e-3). Figure 3b shows values near 3e-4, not below 1e-4. These quantitative claims need to be harmonized; the 1e-4 statement is not supported by the data shown in the manuscript.","section":"Abstract and Fig. 3b / Conclusion"}],"minor_comments":[{"comment":"Typo: 'complete |100> <-> |200> transition' should be '|101> <-> |200>'.","section":"CPhase implementation"},{"comment":"Typo: 'Table II ummarizes' should be 'summarizes'.","section":"Scalability / Table II"},{"comment":"The phrase 'systems energy levels' should be 'system's energy levels'; the phrase 'time-scale control pulse reproducibility' is awkward and could be clarified.","section":"Abstract"},{"comment":"The statement 'available from the corresponding author upon reasonable request' is not ideal for reproducibility; consider depositing the XEB data and simulation scripts in a public repository.","section":"Data Availability"}],"recommendation":"major_revision","confidential_remarks":"The measured 99.48% CPhase fidelity is a solid experimental result and should not be dismissed. The main risk is overclaiming the simulation: the 99.9% / <1e-4 projection depends on an unvalidated echo assumption. If the authors remove or qualify that projection and fix the arithmetic inconsistencies in Eqs. (3) and (4), the paper would be acceptable. Please check that the abstract version (1e-3 vs 1e-4) is harmonized before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The measured 99.48(9)% CPhase fidelity is the real takeaway, and it holds up. The protocol sensibly combines net-zero bipolar flux pulsing with a tunable coupler, and the intermediate weak pulse for amplitude-parameterized CPhase angles is a clean practical trick: two-parameter calibration, arbitrary phases, and built-in protection against long-timescale flux distortions. The energy-level engineering that keeps coupler-state overlap below 1% while still getting 40 MHz effective coupling is well done, and the XEB analysis with leakage and purity decomposition is standard and internally consistent. On the four-qubit processor, >99.1% CZ across pairs with ZZ below 30 kHz supports the scalability argument. The soft spots are real but mostly not fatal. The abstract claims validation on an 8-qubit processor; the body only shows a 4-qubit chip. That is an overstatement and should be corrected. The 1.5 factor in Eq. (4) is unexplained and shifts the inferred decoherence error; it's a small number but the derivation should be shown. The bigger issue is the 99.9% simulation projection: it uses Q1's echo dephasing time (48.5 µs) rather than its Ramsey time (4.5 µs), relying on the bipolar pulse to refocus low-frequency flux noise. That assumption is not obviously safe during the |101>–|200> exchange, where population partially leaves the computational subspace and the two pulse segments have different instantaneous frequencies. If the effective dephasing is closer to Ramsey, the simulated error goes up by roughly an order of magnitude and the sub-1e-3 projection is not supported. This does not touch the measured fidelity, but it means the forward-looking claim should be tempered or backed by a proper noise model. No code or data is released, which is a minor annoyance. Who this is for: people working on transmon gate calibration and tunable-coupler architectures. It's a solid experimental paper with a useful practical contribution, and the measured numbers are credible. It deserves a serious referee, with requests to fix the abstract, explain the 1.5, and either justify the echo assumption or soften the 99.9% claim.","headline":"Credible 99.48% CPhase demonstration on a tunable-coupler transmon device; the abstract overclaims an 8-qubit validation and the 99.9% projection leans on an unproven echo assumption.","tokens_in":17511,"tokens_out":3413,"would_cite":true,"duration_ms":28562,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Bipolar flux pulses plus a tunable coupler yield 99.48% transmon two-qubit gates","keywords":["transmon qubits","tunable coupler","bipolar flux pulses","CPhase gate","cross-entropy benchmarking","flux noise echo","superconducting quantum processor","residual ZZ coupling"],"falsifier":"Deliberately add low-frequency flux noise on Q1 while interleaving the CPhase gate, and see whether the measured fidelity degrades as if dephasing were set by the echo time (48.5 microseconds) or the Ramsey time (4.5 microseconds); a scaling closer to Ramsey invalidates the sub-1e-3 error projection.","tokens_in":16378,"feed_emoji":"⚛️","tokens_out":6812,"duration_ms":63960,"temperature":0.7,"pith_summary":"This paper tries to show that two established techniques for superconducting qubits—tunable couplers that switch qubit-qubit coupling off, and bipolar 'net-zero' flux pulses whose average area is zero—can be combined into one two-qubit gate that is fast, quiet, and easy to calibrate. The measured payoff is a controlled-phase gate with 99.48(9)% fidelity, residual always-on qubit-qubit (ZZ) coupling below 5 kHz, and leakage to coupler states below 1e-5 for 20-ns pulses. The scheme makes the acquired phase a function of a single intermediate pulse amplitude, so any CPhase angle (not just 180 degrees) is reachable without retiming the waveform. Numerical three-qutrit simulations with measured coherence times put the decoherence contribution at 3.4e-4 and suggest the same design can reach below 1e-3 total error with better pulse shaping or longer coherence.","feed_headline":"Two-qubit gate hits 99.48% with net-zero flux pulses","feed_subtitle":"A tunable coupler plus echo-like bipolar pulses keeps ZZ crosstalk under 5 kHz and calibration at two knobs.","key_machinery":"The central object is the bipolar net-zero flux pulse: a positive Gaussian-smoothed pulse followed by an equal negative pulse on one qubit. Each strong pulse drives the non-adiabatic |101⟩↔|200⟩ population exchange while the detuned tunable coupler supplies effective coupling strength g~CZ up to 40 MHz; the opposite-polarity second pulse cancels the net flux and refocuses low-frequency noise in an echo-like manner. The machinery also includes the coupler's energy-level engineering: keeping the coupler more than 1 GHz detuned makes the leakage-state overlap grow as g1c^2/Delta^2 while the desired coupling grows as g1c^2/Delta, so strong coupling is possible with under 1% leakage overlap and s","core_discovery":"The central claim is that a CPhase gate can be executed as a three-step diabatic protocol: bring |101⟩ into resonance with |200⟩ with a positive flux pulse while the tunable coupler strengthens the effective coupling; apply a weak bipolar pulse of adjustable amplitude a_int to set the conditional phase; then apply a negative pulse of the same shape to complete the exchange and echo low-frequency flux noise. Because the two strong pulses have opposite polarity, their net flux is zero, so long-timescale flux-line distortions do not build up; because the coupler is kept strongly detuned, the computational states hybridize with coupler leakage states by under 1% while still giving an effective c","pith_inferences":["If the echo-like cancellation is as complete as assumed, the gate error should be nearly independent of the amplitude of slow flux noise on Q1; injecting low-frequency noise and measuring XEB fidelity would test this directly.","The three-step pulse template (strong +, weak phase-setting, strong −) is not tied to transmon specifics; fixed-frequency qubit platforms with a tunable coupler and comparable anharmonicity could reuse the same calibration logic.","Amplitude-only phase setting suggests variational or randomized circuits that need many different two-qubit phases could be compiled in situ without retiming the pulse clock.","The abstract claims 8-qubit validation while the body reports a 4-qubit processor; the scheme's crosstalk behaviour at 8 qubits, with each coupler parked at the ZZ-free point, remains an open check."],"forward_implications":["Arbitrary CPhase angles, including non-Clifford fractional phases, can be calibrated by a single intermediate-pulse amplitude, so a continuous family of two-qubit gates is available without waveform redesign.","Residual ZZ coupling below 5 kHz lets qubits idle and run simultaneous single-qubit gates with small frequency detunings, easing frequency crowding on multi-qubit chips.","Net-zero flux waveforms reject long-timescale flux-line distortions, so the gate does not require pre-distortion corrections and is reproducible on programmable control hardware.","Leakage to coupler-associated states is predicted below 1e-6 for 20-ns Gaussian-smoothed pulses, and coherence simulations put the floor near 3.4e-4, so the remaining gap to 99.9% is a pulse-shaping problem, not an architecture problem."],"supporting_citations":[{"why":"Supplies the tunable-coupler circuit and the three-oscillator Hamiltonian whose dispersive regime the gate design uses.","marker":"[25]"},{"why":"Introduces net-zero bipolar flux pulsing, the zero-average pulse scheme this work combines with the coupler.","marker":"[21]"},{"why":"Provides the set-net-zero variant and the continuous gate-family idea that the weak intermediate pulse replaces with amplitude-only tuning.","marker":"[23, 24]"},{"why":"Defines the cross-entropy benchmarking protocol used to measure the 99.48(9)% fidelity and the reference/interleaved depolarization rates.","marker":"[26, 27]"},{"why":"Gives the fourth-order perturbative expressions for effective XY and ZZ couplings used to engineer the ZZ-free parking point.","marker":"[28]"},{"why":"Identifies coupler-associated leakage states and non-adiabatic leakage channels that the energy-level design suppresses.","marker":"[30]"},{"why":"Supplies the purity-error method used to separate decoherence and leakage contributions in the XEB data.","marker":"[32]"},{"why":"Provides the numerical master-equation solver used for leakage and decoherence simulations.","marker":"[33]"}],"fun_headline_variants":["Bipolar flux pulses yield 99.5% two-qubit gate fidelity","Zero-net flux pulses hit 99.5% gate fidelity with tunable coupler","Bipolar pulses on tunable coupler push two-qubit gate to 99.5%","Tunable coupler plus bipolar flux echoes yield 99.5% two-qubit gate"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The simulated 99.9% projection assumes the bipolar pulse pair fully cancels low-frequency flux noise during the ~100-ns gate, so the qubit keeps its long echo dephasing time (48.5 microseconds) rather than its short Ramsey time (4.5 microseconds); if that cancellation is incomplete, the simulated 3.4e-4 decoherence error is optimistic.","fun_headline_variants_meta":{"raw":{"variants":["Bipolar flux pulses yield 99.5% two-qubit gate fidelity","Zero-net flux pulses hit 99.5% gate fidelity with tunable coupler","Bipolar pulses on tunable coupler push two-qubit gate to 99.5%","Tunable coupler plus bipolar flux echoes yield 99.5% two-qubit gate"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000555,"raw_usage":{"total_tokens":2458,"prompt_tokens":700,"completion_tokens":1758,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":444,"completion_tokens_details":{"reasoning_tokens":1665}},"tokens_in":444,"tokens_out":1758,"duration_ms":12378,"temperature":1.0,"reasoning_tokens":1665,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T05:46:22.414802+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Deliberately add low-frequency flux noise on Q1 while interleaving the CPhase gate, and see whether the measured fidelity degrades as if dephasing were set by the echo time (48.5 microseconds) or the Ramsey time (4.5 microseconds); a scaling closer to Ramsey invalidates the sub-1e-3 error projection.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the tunable-coupler circuit and the three-oscillator Hamiltonian whose dispersive regime the gate design uses."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces net-zero bipolar flux pulsing, the zero-average pulse scheme this work combines with the coupler."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the fourth-order perturbative expressions for effective XY and ZZ couplings used to engineer the ZZ-free parking point."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Identifies coupler-associated leakage states and non-adiabatic leakage channels that the energy-level design suppresses."},{"cited_title":"Johansson, P","cited_arxiv_id":null,"evidence_quote":"Provides the numerical master-equation solver used for leakage and decoherence simulations."}],"review_version":1}