{"id":"b6cb4a9c-aec8-4a85-bc0e-e187830e234d","arxiv_id":"2607.23463","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A transmon-plus-Purcell-filter-plus-notch hybrid coupler reaches simulated CZ F_avg=99.74%, F_min=99.62%, and max leakage 1.6e-3 by selectively damping leakage dressed states.","lead":"A superconducting CZ-gate coupler that adds a Purcell filter and notch resonator cuts simulated leakage by nearly 20× versus a plain transmon coupler. The design treats engineered dissipation as a deliberate control knob alongside coherent coupling, which matters for error-corrected quantum processors.","discovery_kind":"new_method","skeptic_critique":{"model":"moonshotai/kimi-k3","headline":"The reported engineered decay rates (Γ_leak ≈ 3.4×10⁻³ MHz) are ~3 orders of magnitude too small to act during a 327 ns gate, so the 20× leakage reduction must be coherent, not dissipative — undercutting the paper's central \"engineered dissipation\" framing.","rationale":"The reader's weakest assumption (fixed operating point, no bias transients) is a genuine external-validity concern, and I share it, but I find a sharper internal one: the paper's numbers contradict its own mechanistic story. Γ_leak ~ 3.4×10⁻³ MHz with t_CZ = 327 ns means engineered dissipation removes at most ~1% of leaked population per gate, while the headline is a 20× leakage reduction. Three internal signals converge on the coherent interpretation: (i) the rate arithmetic above; (ii) the notch improves performance while *lowering* the engineered decay rates (Sec. III.A.1 vs Table VI); (iii) the authors themselves hedge with \"combined effects of coherent state hybridization and frequency-selective coupling\" (Sec. III.B.1). This does not move the verdict off CONDITIONAL: the within-model numerical claim (hybrid beats optimized single-transmon coupler on leakage and F_min) can still be correct as a coherent multi-mode coupler result, the baselines and scans are well controlled, and the authors appropriately disclaim logical-state recovery and multi-cycle QEC benefit. But the conditionality should explicitly include the κ→0 ablation above, because the paper's stated contribution — dissipation engineering as a new design resource — is precisely what the simulations, as reported, do not isolate. If the ablation shows the gains are coherent, the paper remains a competent coupler-design study with a misattributed mechanism (novelty closer to incremental multi-resonator coupling work); if it shows the gains survive only with dissipation on, the rate numbers in Sec. III.A.1 need rechecking. Either way the concrete test is cheap, decisive, and within the authors' existing QuTiP pipeline.","tokens_in":17284,"tokens_out":4276,"duration_ms":128207,"concrete_test":"Re-run the optimized operating point (Tables I–II) in QuTiP with κ_f = κ_n = 0, all coherent parameters and the t_CZ sweep unchanged, and compare L_max, F_min, and the random-state leakage statistics (Fig. 5/Table IV) against the dissipative run. If L_max and F_min shift by less than a few percent, the dissipative mechanism is inert during the gate and the framing must change to coherent interaction engineering; if instead L_max degrades substantially, my rate arithmetic is wrong somewhere and the dissipative claim stands. As a secondary check, evolve the post-gate state for ~100 µs of idle time and extract the leakage-state lifetime, to quantify the actual dissipative benefit the architecture provides between gates.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's thesis is that \"leakage-selective dissipation\" from the filter–notch subsystem suppresses CZ leakage (Abstract; Sec. II.C; Sec. III.A.1). But the paper's own numbers are quantitatively inconsistent with that mechanism operating during the gate. Sec. III.A.1 reports average engineered decay rates of Γ_leak = 3.40×10⁻³ MHz (filter+notch) and 4.11×10⁻³ MHz (filter-only), i.e. ~3–4 kHz in ordinary-frequency units, corresponding to leakage-state lifetimes of roughly 50–300 µs. The gate lasts t_CZ = 327.3 ns. The probability of an engineered decay event during the gate is at most t_CZ × Γ_leak ≈ 10⁻³–10⁻². Yet the headline result is a reduction of L_max from ~3×10⁻² to 1.6×10⁻³ — a ~95% removal of leakage. Dissipation acting at kHz rates cannot produce this during a sub-µs gate; the reduction must come almost entirely from coherent effects (modified J_XX via J^(f)_XX, Eq. 7; dressed-state hybridization; retuned gate duration), i.e. the same physics as any multi-mode coherent coupler. The paper's own ablation supports this reading: adding the notch *reduces* both Γ_comp and Γ_leak (2.78→2.22×10⁻⁴ and 4.11→3.40×10⁻³ MHz) while *improving* leakage and fidelity (Table VI) — the opposite of what a dissipation-driven mechanism predicts. This doesn't invalidate the numerical comparison against the single-transmon baseline, but it means the claimed novelty — \"engineered dissipation as an additional design degree of freedom\" (Abstract; Sec. V) — is not what the simulations actually demonstrate. The genuine dissipative benefit (shortening leakage lifetime between gates) would be ~47–300 µs vs the 300 µs reference T1, a modest factor the paper never isolates. No code is released, so this is checkable only by the authors or by independent reimplementation.","agreement_with_reader":"partial"},"referee_report":{"model":"moonshotai/kimi-k3","summary":"The authors propose a hybrid coupler for superconducting CZ gates in which a tunable transmon coupler is supplemented by a coupled Purcell-filter/notch-resonator subsystem. The filter branch adds a coherent exchange pathway (modifying the effective J_XX) while the filter–notch pair is designed to provide frequency-selective dissipation: strong engineered decay of leakage-like dressed states with suppressed decay of the computational manifold. Using full-Hamiltonian diagonalization (J_ZZ from Eq. 4; dressed decay rates from Eqs. 11–12) and Lindblad master-equation simulations in QuTiP, the authors compare the hybrid design against an optimized single-transmon coupler and a filter-only (g_fn=0) ablation, reporting F_avg=99.74%, F_min=99.62%, and L_max=1.6×10⁻³ at t_CZ=327.3 ns, with robustness scans over filter, notch, and qubit–filter parameters and six coherence sets.","tokens_in":17807,"tokens_out":5443,"duration_ms":66928,"significance":"If the results hold, this is a useful architecture-level contribution: leakage is a leading error source for superconducting QEC, and dissipation engineering for gates (as opposed to readout protection) is comparatively unexplored — the connection to passive leakage reset (Ref. 44, Thorbeck et al.) is timely. The study has real methodological strengths: explicit and separately optimized baselines (single coupler and filter-only ablation), coherence sweeps from conservative to near-lossless limits (Table III, Fig. 7), 2D robustness scans with J_ZZ re-extraction and local time re-optimization at each point (Figs. 9–11), and generally candid caveats (e.g., the acknowledgement that filter-induced relaxation converts leakage into computational errors rather than correcting them). These strengths make the numerical comparison credible as a simulation result. However, the paper is simulation-only, the fidelity metric is incomplete (below), and the central mechanistic framing — engineered dissipation as the operative leakage-suppression mechanism during the gate — is not quantitatively supported by the paper's own numbers.","major_comments":[{"comment":"The central framing — 'leakage-selective dissipation' as the mechanism suppressing CZ leakage (Abstract; Sec. II.C; Sec. III.A.1; Conclusion) — is quantitatively inconsistent with the paper's own numbers for the single-gate dynamics reported here. The reported average engineered rates are Γ_leak = 3.40×10⁻³ MHz (filter+notch) and 4.11×10⁻³ MHz (filter-only), i.e. ~3–4 kHz in ordinary-frequency units, corresponding to leakage-state lifetimes of order 50–300 µs. During t_CZ = 327.3 ns the probability of even one engineered decay event is at most Γ_leak × t_CZ ~ 10⁻³–10⁻², yet the headline result is a ~2.8×10⁻² absolute reduction of L_max. The suppression must therefore be dominated by coherent effects — the filter-mediated exchange J^(f)_XX (Eq. 7), dressed-state hybridization, and the retuned J_ZZ/t_CZ — i.e. multi-mode coherent coupler physics, with dissipation playing at most a minor wi","section":"§III.A.1, Eqs. (11)–(12), Fig. 3, Table VI"},{"comment":"The fidelity metric is a computational-basis population fidelity after local-Z correction: F_i is the final population in the target basis state. This metric is blind to any residual unitary error that is diagonal in the computational basis — in particular a conditional-phase error (gate angle ≠ π) or residual ZZ at the final time leaves all four basis-state populations unchanged and is invisible to F_avg, F_min, and to the time-sweep optimization that selected t_CZ. For a CZ gate, coherent conditional-phase error is one of the primary error channels, so the reported F_avg = 99.74% / F_min = 99.62% are upper bounds that exclude this channel; the comparison with the single-coupler baseline (Table V) inherits the same blindness. Please report a phase-sensitive figure of merit — e.g. average gate fidelity against the ideal CZ (computable from the simulated propagator on the computational su","section":"§III.B.1, definition of F_i and F_avg/F_min"},{"comment":"All simulations use a time-independent Hamiltonian at a fixed operating point, and the idle-to-gate biasing transient is explicitly not modeled. Since the gate relies on a static J_ZZ ≈ 1/(2 t_CZ) ≈ 1.5 MHz, the architecture requires an idle configuration with strongly suppressed J_ZZ, and the ramp between the two will traverse the dressed spectrum whose selective damping is central to the mechanism. Nothing is said about the idle point: what is the residual J_ZZ there, and does the frequency trajectory cross filter/notch resonances that could either excite or beneficially reset leakage-like dressed states? At minimum, please report the idle-point parameters and residual J_ZZ, and discuss (or bound) the effect of a realistic flux ramp on the dressed-state selectivity that the fixed-point Lindblad runs rely on. This is a disclosed limitation, but for an architecture whose selling point is","section":"§III opening paragraphs; §IV"}],"minor_comments":[{"comment":"The hybrid-coupler L_max differs between Table V (1.6×10⁻³) and Table VI (3.0×10⁻³); the text explains this as a common-evaluation vs. locally-optimized-point distinction, but the caption of Table VI should state explicitly which parameter set was used for each row, since readers will otherwise read the two tables as contradictory.","section":"Tables V–VI"},{"comment":"The optimized single-transmon baseline parameters are never tabulated (only the hybrid parameters appear in Table I). Please provide them, and consider releasing the QuTiP scripts; both would substantially improve reproducibility.","section":"§III.B.2"},{"comment":"State explicitly whether the Γ values in Eqs. (11)–(12) and Fig. 3 are quoted as Γ/2π in MHz (ordinary-frequency units, as for J_ZZ) or as angular rates; the distinction matters by a factor of 2π for the lifetime estimates.","section":"§III.A.1"},{"comment":"Both panels are explicitly illustrative and 'do not correspond to the optimized device parameters'. It would be more informative to additionally show R(ω) for the actual Table I parameters, marking the dressed computational and leakage transition frequencies.","section":"Fig. 2"},{"comment":"The labels Low/Med./Good/Curr./Opt./Lossless should be mapped to Table III sets in the figure caption, not only in the text.","section":"Fig. 7"},{"comment":"Typographical: 'achievesF avg' (Abstract); 'controlled-Z(CZ)'; inconsistent spacing around equations. Ref. [17] is an arXiv-only self-citation; update if published. The relation of the present work to Ref. [44] (bath engineering for passive leakage reset in transmons) deserves a fuller comparison, as that work is the closest prior art for the dissipative claims.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The numerical work appears careful and the ablation structure (filter-only, coherence sweeps, robustness scans) is above average for this genre. The two substantive issues — the phase-insensitive fidelity metric and the dissipation-rate budget that cannot support the 'engineered dissipation' framing for within-gate suppression — are both addressable with additional simulation runs using the authors' existing pipeline, so major revision rather than rejection seems right. I flag for the editor's awareness the citation pattern: Refs. [16, 17, 38] are the first author's prior multi-resonator coupler works, which supply the coherent-pathway formulas (Eqs. 6–7); this is legitimate background but reinforces that the genuinely novel element is the dissipative claim, which is precisely the part that currently lacks quantitative support. If the κ→0 ablation shows the within-gate suppression is essentially all coherent, the paper remains a competent multi-mode coupler design study, but its positioning relative to the authors' prior double-resonator work would need to be recalibrated."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The useful result here is a concrete filter–notch + transmon coupler layout that, in Lindblad sims, beats an optimized single-transmon coupler on worst-case fidelity and bare leakage (F_avg 99.74%, F_min 99.62%, L_max 1.6e-3 vs ~3e-2). They do the comparison properly: full diagonalization, filter-only ablation, coherence sweeps, and 2D robustness maps. That package is worth having on the shelf if you care about multi-mode CZ hardware.\n\nWhat is actually new is the integrated architecture and the quantitative ablation, not the ingredients. Purcell filters, notch protection, tunable couplers, and multi-resonator pathways are all prior art; they cite the right lines. The dressed-state decay analysis and the honesty that relaxation converts leakage into computational error rather than “fixing” it are strengths.\n\nThe soft spot is load-bearing and about framing, not about fake numbers. Their own Γ_leak is a few kHz (ordinary-frequency units). Over a 327 ns gate that gives at most ~10^{-3}–10^{-2} decay probability, which cannot explain a ~20× drop in L_max. The notch also lowers both Γ_comp and Γ_leak while improving the gate—the opposite of a pure “more selective loss during the pulse” story. So the during-gate win is almost certainly coherent pathway reshaping (filter-mediated J_XX, hybridization, retuned t_CZ), i.e. multi-mode coupler physics with a dissipative environment bolted on. The genuine dissipative upside—shorter leakage lifetime between gates—is real in principle but never isolated, and they correctly leave multi-cycle QEC unclaimed. Fixed operating point with no bias ramps is a second, smaller idealization they already flag.\n\nMath and citation pattern look fine; no circular score-forcing; free-parameter burden is high because the headline percentages are optimized-model outcomes and there is no code. For coupler/QEC hardware people this is a serious methods paper that needs a referee who will force the abstract and conclusion to match the rates. I would send it to review, not desk-reject it, and I would bring it to reading group mainly to pin down coherent vs dissipative credit. I would not cite the “engineered dissipation as the CZ design knob” claim as stated until that is cleaned up; I might cite the architecture comparison if I were building a similar coupler.","headline":"Solid coupler simulation with a real framing problem: the during-gate leakage win is mostly coherent, not dissipative.","tokens_in":18720,"tokens_out":603,"would_cite":false,"duration_ms":29378,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"A hybrid coupler that pairs a transmon with a Purcell filter and notch resonator cuts superconducting CZ-gate leakage nearly twentyfold and raises worst-case fidelity above 99.6%.","keywords":["superconducting qubits","CZ gate","Purcell filter","notch resonator","leakage suppression","engineered dissipation","transmon coupler","dressed states"],"falsifier":"Build the hybrid coupler and run a fixed-point CZ with the reported parameters (or a local re-optimization): if measured basis-state fidelities and leakage fail to beat an optimized single-transmon coupler by a large margin—especially if idle-to-gate flux/frequency ramps restore L_max near a few percent—the central claim does not hold in hardware.","tokens_in":18246,"feed_emoji":"⚡","tokens_out":1027,"duration_ms":20864,"temperature":0.7,"pith_summary":"Superconducting CZ gates are limited not only by coherent control errors but by leakage into noncomputational states that can persist and spoil error correction. This paper argues that leakage can be fought with engineered dissipation, not only with pulse shaping and coupler design. It proposes a hybrid coupler: a nonlinear transmon plus a lossy Purcell-filter resonator and a weakly lossy notch resonator that together reshape both the coherent qubit–qubit interaction and the frequency-dependent environment. Dressed-eigenstate analysis and Lindblad simulations show leakage-like states decay about an order of magnitude faster than computational states, yielding F_avg = 99.74%, F_min = 99.62%, and maximum leakage of 1.6×10^{-3}—far better worst-case fidelity and roughly twentyfold less leakage than an optimized single-transmon coupler—while remaining stable across coherence sets and device-parameter scans. If the design holds in hardware, engineered dissipation becomes an extra design knob for robust two-qubit gates.","feed_headline":"Hybrid coupler cuts CZ leakage nearly twentyfold","feed_subtitle":"Purcell filter plus notch resonator lifts worst-case fidelity above 99.6% versus a single-transmon coupler","key_machinery":"The filter–notch subsystem: a lossy Purcell-filter mode coupled to a weakly dissipative notch resonator. The filter preferentially damps leakage-related dressed states (associated with 1→2 transitions); the notch reshapes the environmental response so computational 0→1 transitions stay protected. Engineered decay rates of dressed eigenstates plus Lindblad master-equation gate simulations quantify the selectivity and the resulting CZ performance.","core_discovery":"The authors claim that a Purcell-engineered notch-filter hybrid coupler—integrating a nonlinear transmon coupler with a coupled Purcell-filter and notch-resonator subsystem—combines coherent interaction engineering with leakage-selective dissipation and thereby substantially reduces residual leakage and improves worst-case computational-state fidelity for fixed-frequency CZ gates relative to an optimized single-transmon coupler. At the reported optimum they obtain F_avg = 99.74%, F_min = 99.62%, and L_max ≈ 1.6×10^{-3}, with the improvement persisting across coherence assumptions and broad filter/notch parameter ranges.","pith_inferences":["If leakage-like dressed states are preferentially emptied each gate, multi-cycle error-correction runs may see less correlated leakage propagation than single-gate L_bare alone suggests—something the paper flags as future work.","The same filter–notch spectral shaping could be tried on other leakage-prone two-qubit gates (e.g. iSWAP-family) where 1–2 transitions sit near environmental poles.","Combining this hardware environment with leakage-aware optimal control, rather than fixed-duration evolution, is a natural next optimization the baseline numbers leave open."],"forward_implications":["Engineered dissipation becomes a usable design degree of freedom alongside coherent coupler engineering for superconducting CZ gates.","Maximum leakage can drop from ~3×10^{-2} to ~1.6×10^{-3} with F_min above 99.6% under the reported conditions.","The low-leakage, high-fidelity window stays open over broad filter and notch frequency/coupling ranges, easing fabrication tolerance.","Passive linear resonators (filter and notch) can supply this improvement without new active Josephson elements."],"fun_headline_variants":["Purcell-notch hybrid coupler slashes CZ leakage to 0.16%","Hybrid coupler lifts worst-case CZ fidelity above 99.6%","Leakage-selective dissipation beats single-transmon CZ coupler","Notch-filter hybrid coupler yields 99.74% average CZ fidelity","Engineered dissipation trims CZ leakage nearly twentyfold"],"cache_read_input_tokens":128,"weakest_assumption_plain":"The gate is treated as evolution under a fixed, time-independent Hamiltonian at one operating point; the real frequency-biasing ramp into and out of the gate, and any leakage-aware pulse shaping, are not simulated.","fun_headline_variants_meta":{"raw":{"variants":["Purcell-notch hybrid coupler slashes CZ leakage to 0.16%","Hybrid coupler lifts worst-case CZ fidelity above 99.6%","Leakage-selective dissipation beats single-transmon CZ coupler","Notch-filter hybrid coupler yields 99.74% average CZ fidelity","Engineered dissipation trims CZ leakage nearly twentyfold"]},"model":"grok-4.5","effort":"low","cost_usd":0.001787,"raw_usage":{"total_tokens":917,"prompt_tokens":820,"num_sources_used":0,"completion_tokens":77,"cost_in_usd_ticks":17868000,"prompt_tokens_details":{"text_tokens":820,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":20,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":820,"tokens_out":77,"duration_ms":2180,"temperature":1.0,"reasoning_tokens":20,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-30T21:35:01.699258+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Build the hybrid coupler and run a fixed-point CZ with the reported parameters (or a local re-optimization): if measured basis-state fidelities and leakage fail to beat an optimized single-transmon coupler by a large margin—especially if idle-to-gate flux/frequency ramps restore L_max near a few percent—the central claim does not hold in hardware.","supporting_citations":[],"review_version":1}