{"id":"f43f9476-900b-41eb-abda-90c1a40fb5ca","arxiv_id":"1908.01240","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Drive-activated number-nonconserving Josephson terms induce a correlated qubit-cavity relaxation channel that increases the qubit decay rate approximately linearly with the readout cavity photon number.","lead":"This paper derives a mechanism for why superconducting qubits lose energy faster when the readout drive is turned on. It shows that photon-number-nonconserving terms in the Josephson potential, usually dropped in simplified models, create a correlated qubit-cavity decay channel whose strength grows with the cavity photon number.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The dominant term in Eq. (34) has a resonance denominator that cancels the anharmonicity, so the first-order-in-epsilon truncation is uncontrolled; the reported linear-in-nbar rate increase is not quantitatively established.","rationale":"The reader's weakest assumption concerned zero temperature and the absence of pure dephasing; that is a legitimate limitation, but it is explicitly disclosed in Sec. IV and affects practical relevance rather than the internal correctness of the central calculation. The concern raised here is different and more direct: the term that produces the headline effect is obtained by a perturbation expansion whose small parameter is not small for the resonant term. The paper argues the mechanism qualitatively, but the quantitative claim that the relaxation rate grows approximately linearly with nbar with the coefficient shown in Fig. 2c requires the first-order EME to be a controlled approximation. The near-cancellation of epsilon by omega_c - omega_d means the effective expansion parameter is eta_x/u_ac, which is of order unity for the plotted drive strengths. This does not prove the mechanism is wrong, but it means the reported slope is not established. The appropriate response is to condition acceptance on a concrete check of the next-order or nonperturbative result. If that check shows the first-order result is stable, the paper's central claim stands; if not, the quantitative conclusion would need to be revised.","tokens_in":40428,"tokens_out":25075,"duration_ms":260427,"concrete_test":"Compute the second-order-in-epsilon correction to the qubit collapse operator C(omega_a) using the G6(t) equation (App. C, Eq. C8), retaining the resonant channel omega_c - omega_d = chi_ac/2, and recompute delta_kappa_a/kappa_a(0) versus nbar at nbar = 0.5. Alternatively, simulate the same readout parameters with a nonperturbative Floquet-Born-Markov master equation using the full cosine Josephson potential, without the Schrieffer-Wolff truncation, and compare the extracted rate to Fig. 2c. If the second-order or full-Floquet result differs from the first-order EME by more than ~20%, the reported linear-in-nbar rate increase is not quantitatively supported.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central numerical claim (Fig. 2c, Eq. 41) rests on the EME truncated at first order in the anharmonicity epsilon. The dominant term identified by the authors is the correlated qubit-cavity operator a(eta_x* c - eta_x c-dagger) in Eq. (34), whose prefactor is proportional to epsilon * omega_d/(omega_c - omega_d). Because the drive is placed at omega_d = omega_c - chi_ac/2 (Eq. 38) and chi_ac = epsilon * omega_a * u_ac^2 * u_aa^2 / 4 (Eq. 27), this prefactor evaluates to 4(omega_d/omega_c) v_ca/u_ac, independent of epsilon. Thus the 'first-order in anharmonicity' term is not small in epsilon: the same near-resonant denominator appears in the generator G4(t) (Eq. 11), and the effective expansion parameter for the resonant channel is of order eta_x/u_ac. With u_ac ~ g/Delta ~ 0.1 and eta_x ~ 0.03-0.1 over the plotted range (Fig. 3a), this parameter is ~0.3-1, so higher-order terms, e.g. from G6(t) in App. C (Eq. C8), can contribute at a comparable level. The paper does not estimate these contributions, and the comparison to the Kerr-only model only shows that the effect is nonzero, not that the first-order result is quantitatively accurate. Since the linear-in-nbar rate enhancement is a central quantitative claim, this truncation is load-bearing.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper derives an effective master equation (EME) for a transmon qubit dispersively coupled to a readout cavity under a coherent drive, using a two-parameter perturbative expansion in the Josephson anharmonicity ε and the drive amplitude. It shows that number-nonconserving terms in the Josephson potential, when dressed by a time-dependent unitary transformation, generate drive-activated correlated qubit-cavity dissipation (dominantly âĉ) and hence a qubit relaxation rate that increases with the steady-state cavity photon number −n_c, approximately linearly at low power. The EME is compared numerically with a Kerr-only master equation, and the difference is attributed to the number-nonconserving terms. The paper also provides a one-mode version in Appendix D and discusses why the usual two-level truncation of the nonlinearity misses this effect.","tokens_in":40662,"tokens_out":10275,"duration_ms":106596,"significance":"If quantitatively reliable, this mechanism would explain the observed drive-power-dependent T1 reduction in dispersive readout without invoking dephasing noise and would identify number-nonconserving terms in the Josephson potential as the physical origin. The derivation is parameter-free in the sense that nothing is fitted to the target observation; the result follows from the circuit Hamiltonian and standard open-quantum-system methods. The explicit collapse operators and the direct comparison with a Kerr-only model make the mechanism falsifiable, and the systematic unitary-transformation technique is of broader utility. The main caveat is that the quantitative accuracy of the leading-order EME is not controlled for the near-resonant channel, so the central quantitative claim (the linear-in-−n_c rate increase) needs additional support before the numerical prediction can be taken at face value.","major_comments":[{"comment":"The dominant correlated term in Ĥ C(ω_a) is not parametrically small in ε. Its coefficient is proportional to ε ω_d/(ω_c−ω_d); with the readout detuning ω_d = ω_c−χ_ac/2 and χ_ac ∝ ε, the ratio ω_d/(ω_c−ω_d) ∝ 1/ε, so the ε-dependence cancels and the term is of order η_x times O(1) hybridization factors. More precisely, using χ_ac = ε ω_a u_ac^2 u_aa^2/2, the strength of this channel relative to the linear collapse operator is set by η_x/u_ac rather than by ε. Because the same near-resonant denominators appear in the generator equation (11) and in the second-order generator equation (C8), the effective expansion parameter for this channel is η_x/u_ac, which is of order 0.3–1 for the parameters of Fig. 3, not ε≈0.1. The paper does not estimate the next-order contributions to the collapse operators, so the quantitative prediction δκ_a/κ_a versus −n_c in Fig. 2c is not established to the claimed accuracy, even though the qualitative direction of the effect may be correct. Please provide a bound on the next-order corrections, a resummation of the resonant channel, or an explicit numerical convergence check in ε.","section":"Sec. III A, Eqs. (34), (27), (38); App. C, Eq. (C8)"},{"comment":"The numerical EME (33) uses the rotating-wave (secular) form of the master equation, whereas the derivation in App. E shows that the more general non-RWA form (E37) is required when transition frequencies become close compared with relaxation rates. The simulation parameters have κ_c ≈ 10^−2π and χ_ac ≈ 1.7×10^−3π, so the RWA is not automatically justified, and no comparison with Eq. (E37) is reported. Since the central rate extraction relies on this EME, please justify the RWA for the chosen parameters or demonstrate numerically that the non-RWA terms are negligible.","section":"Sec. III B and App. E, Eq. (E37)"}],"minor_comments":[{"comment":"There is a factor-of-two discrepancy in the definition of χ_ac: Eq. (27) gives χ_ac = ε ω_a u_ac^2 u_aa^2/4 (with ω_a replaced by ω̄_a), while the text below Eq. (38) states χ_ac = ε ω_a u_aa^2 u_ac^2/2. Please reconcile the two definitions and check the quoted numerical value 1.7×10^−3ω̄_c against the parameters in Eq. (36).","section":"Eq. (27) and Sec. III B below Eq. (38)"},{"comment":"The caption for panel b) reads 'The drive strength is adjusted such that the cavity has a mean steady state population −n_c' and appears to be grammatically incomplete; please finish the sentence (for example, 'as a function of time').","section":"Fig. 2 caption"},{"comment":"The symbol −n_c is defined in Eq. (25) as the linear-theory steady-state population, but in Sec. III B it is also used for the numerically extracted steady-state population; please state explicitly that the two agree for the chosen parameters, as asserted in the text, so the notation is unambiguous.","section":"Sec. II, Eq. (25), and Sec. III B"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the journal's scope and the mechanism is potentially important, but the quantitative claim in Fig. 2c is currently based on a first-order-in-ε calculation whose dominant resonant term is not actually small in ε. I would ask the authors to address this before acceptance, for example by estimating the G6 contributions or by providing a resummation of the resonant channel. The accept recommendation in the reader report appears to underweight this issue."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my read. The new physical idea is good: number-nonconserving terms in the Josephson potential combine with the readout drive to create a correlated qubit-cavity decay channel, and that channel can make T1 drop with power even in a pure radiative model. The derivation is careful and transparent, and the contrast with the Kerr-only master equation isolates the effect. The authors are upfront about zero temperature, no pure dephasing, and the atypical Q used in the numerics; those assumptions weaken quantitative contact with experiment but do not by themselves sink the mechanism.\n\nThe soft spot is worse than the paper lets on. The leading correction in Eq. (34) carries a factor epsilon * omega_d/(omega_c - omega_d). With the drive at omega_c - chi_ac/2 and chi_ac proportional to epsilon, that combination is independent of epsilon. So the 'first-order in anharmonicity' term is not actually small in epsilon for the resonant channel. The effective expansion parameter is on the order of eta_x/u_ac, which is 0.3-1 across the plotted range. The generator G4 then has a near-resonant denominator, and the next-order generator G6 (App. C, Eq. C8) can contribute at a comparable level. The paper does not estimate those terms, and the Kerr-only comparison demonstrates only that something nonzero happens, not that the first-order result is quantitatively right. The linear-in-nbar rate increase in Fig. 2c and Eq. (41) is therefore not established at the advertised level of rigor.\n\nThe mechanism might well survive a proper resummation, and the qualitative direction (increase with power) is plausible and consistent with the experimental puzzle. The citation pattern is fine; Part I is correctly used. Circuit QED theorists will get a clear framework, and experimentalists should read it as a caution that the readout T1 drop may have a fundamental radiative origin, but they should not take the numbers as predictive. I would send this to a careful referee, with a request to address the resonant-denominator/truncation control, preferably by computing or bounding G6 contributions or by treating the resonant channel nonperturbatively. Without that, the paper is a strong qualitative proposal rather than a quantitative prediction.","headline":"Plausible mechanism for drive-dependent T1 drop in transmon readout, but the quantitative rate increase rests on a resonant first-order term whose smallness is uncontrolled.","tokens_in":41258,"tokens_out":4314,"would_cite":true,"duration_ms":44211,"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":"A missed Josephson term explains why readout power shortens qubit T1.","keywords":["superconducting qubits","dispersive readout","drive-dependent relaxation","number-nonconserving terms","Josephson potential","effective master equation","Schrieffer-Wolff transformation","Purcell effect"],"falsifier":"Measure the qubit relaxation rate during readout at fixed steady-state cavity photon number while stepping the drive frequency away from the cavity resonance: the paper predicts the renormalization drops to near zero once the detuning exceeds roughly $10\\chi_{ac}$. Alternatively, check linearity: at small $\\bar n_c$ the normalized rate shift $\\delta\\kappa_a^{\\mathrm{EME}}/\\kappa_a^{\\mathrm{EME}}(0)$ should grow linearly with $\\bar n_c$; a rate increase that is flat in $\\bar n_c$, or that persists at large detuning, would rule out the correlated-decay mechanism.","tokens_in":40177,"feed_emoji":"⚛️","tokens_out":9483,"duration_ms":90728,"temperature":0.7,"pith_summary":"Recent experiments on transmon qubits show that the qubit relaxation time $T_1$ drops when the readout drive is made stronger, a limit on readout speed and fidelity. The paper claims the drop is not environmental but structural: number-nonconserving terms in the Josephson potential open a new decay channel once the cavity is driven. At lowest order in the weak anharmonicity $\\epsilon=\\sqrt{2E_C/E_J}$, the dominant channel is a correlated process in which one qubit photon and one cavity photon leave together, giving a relaxation rate that grows approximately linearly with the steady-state cavity photon number $\\bar n_c$. The authors derive an effective master equation exhibiting this channel and show numerically that a Kerr-only model, which keeps only number-conserving terms, predicts no drive-dependent renormalization. If the mechanism is right, the readout-power limit on $T_1$ is intrinsic to the Josephson nonlinearity rather than a symptom of extra losses.","feed_headline":"Qubit decay during readout rises with drive power","feed_subtitle":"A new calculation traces the effect to number-nonconserving Josephson terms that Kerr-only models miss.","key_machinery":"The load-bearing object is the time-dependent Schrieffer-Wolff generator $\\hat G_4(t)$, fixed by the Floquet condition $e^{-\\hat G(t)}[\\hat H_s(t)-i\\partial_t]e^{\\hat G(t)}=\\hat H_{s,\\mathrm{eff}}(t)-i\\partial_t$ at first order in the anharmonicity. It cancels every number-nonconserving monomial of the Josephson quartic potential from the driven Hamiltonian, and in doing so dresses the system-bath coupling, turning the bare bath-coupled quadrature into a sum of renormalized collapse operators. The term that does the work is the correlated operator $\\hat a(\\eta_x^*\\hat c-\\eta_x\\hat c^\\dagger)$ entering $\\hat C(\\omega_a)$ in Eq. (34), whose $1/(\\omega_c-\\omega_d)$ resonance factor explains why the effect appears only when the readout drive is near the cavity frequency. A displacement transformation that absorbs the drive fixes $\\eta_x$ self-consistently through the dissipative linear response of the cavity, Eq. (24).","core_discovery":"The paper's central claim is that, for a weakly anharmonic transmon dispersively coupled to a single-mode cavity and driven near the cavity resonance, the lowest-order correction to the qubit relaxation rate comes from number-nonconserving quartic terms of the Josephson potential. Working at zero temperature and with no intrinsic qubit decay, so that all relaxation is radiative (Purcell) through the cavity, the authors obtain an effective master equation whose qubit collapse operator is approximately $\\hat C(\\omega_a)\\approx -i[\\dots]\\hat a - i(\\epsilon/2)(\\bar\\omega_a/\\omega_c)v_{ca}u_{ac}u_{aa}^2\\,\\omega_d/(\\omega_c-\\omega_d)\\,\\hat a(\\eta_x^*\\hat c-\\eta_x\\hat c^\\dagger)$, with the second term a drive-activated correlated relaxation $\\hat a\\hat c$ (and its conjugate conversion process $\\hat a\\hat c^\\dagger$). Because the coherent displacement $\\eta_x$ grows as $\\sqrt{\\bar n_c}$, this term makes the extracted qubit relaxation rate $\\kappa_a^{\\mathrm{EME}}(\\bar n_c)$ increase roughly linearly with the steady-state cavity photon number. The same mechanism produces drive-induced qubit excitation at zero temperature, and the entire renormalization disappears when the drive is detuned far from the cavity frequency. A Kerr-only master equation with bare $\\hat a$ and $\\hat c$ dissipators shows no such effect.","pith_inferences":["If this correlated-decay channel is dominant in real devices, readout pulse optimization should target it; Kerr-only or two-level models will systematically overestimate the tolerable readout power.","The same $\\hat a\\hat c$ process should appear in driven parametric gates and cat-state stabilization protocols, which populate cavity modes; power-dependent lifetimes observed there could be the same mechanism.","The predicted slope of $1/T_1$ versus $\\bar n_c$ should scale linearly with the anharmonicity parameter $\\epsilon$, so comparing devices with different $E_C/E_J$ at fixed hybridization would isolate the effect.","Adding finite temperature and pure dephasing, which the paper explicitly leaves for future work, could either amplify or mask the predicted increase; quantitative device comparison needs those terms."],"forward_implications":["At the photon numbers typical of dispersive readout ($\\bar n_c$ of a few), the qubit relaxation rate increases roughly linearly with $\\bar n_c$, so pushing readout power buys less measurement time than Kerr-only models predict.","The correction is selective in frequency: it acts only when the drive is close to the cavity resonance and decays algebraically as the drive-cavity detuning grows beyond roughly ten times the cross-Kerr shift $\\chi_{ac}$.","Even at zero temperature the drive induces qubit excitation alongside relaxation, so the qubit's steady-state population during readout is not purely thermal.","Number-nonconserving Josephson terms affect dissipators at the same order of anharmonicity at which number-conserving terms renormalize frequencies; any calculation truncated to number-conserving (Kerr) terms misses the dominant drive dependence of $T_1$.","The double expansion in anharmonicity and drive amplitude is a general route to effective master equations for driven, weakly nonlinear dissipative bosonic circuits."],"supporting_citations":[{"why":"Supplies the unitary-transformation formalism (Part I) that this paper extends to the coherently driven case and reduces to in the zero-drive limit.","marker":"[13]"},{"why":"Reports the experimental observation of T1 decreasing by up to a factor of two at cavity photon numbers near 5 that motivates the mechanism.","marker":"[5,6]"},{"why":"Earlier dispersive-regime theory predicting that relaxation decreases with drive strength in the absence of dephasing; this paper's opposite result is the contrast target.","marker":"[3]"},{"why":"The dressed-dephasing hypothesis that predicts drive-induced increase; the paper argues it does not capture measured effective temperatures and compares against it.","marker":"[11]"},{"why":"Defines the transmon qubit and the weak-anharmonicity parameter $\\epsilon=\\sqrt{2E_C/E_J}$ used as the expansion parameter.","marker":"[4]"},{"why":"Provides the multi-mode Kerr Hamiltonian that represents the number-conserving model used as the comparison simulation.","marker":"[14]"},{"why":"The canonical Schrieffer-Wolff transformation underlying the time-dependent generator that removes number-nonconserving terms.","marker":"[44]"},{"why":"Defines the dispersive readout geometry, qubit coupled to a cavity probed by a microwave tone, that the model starts from.","marker":"[45]"}],"fun_headline_variants":["Drive-activated Josephson terms boost qubit relaxation","Readout drive power speeds qubit decay via new terms","Why stronger readout drives qubit decay","Missing Josephson terms explain readout-induced decay","Qubit decay scales with readout power, mechanism found"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The calculation assumes the qubit has no intrinsic decay or dephasing, all relaxation is radiative through a zero-temperature Markovian cavity bath, and the paper itself says quantitative comparison with experiments would require adding finite temperature and pure dephasing.","fun_headline_variants_meta":{"raw":{"variants":["Drive-activated Josephson terms boost qubit relaxation","Readout drive power speeds qubit decay via new terms","Why stronger readout drives qubit decay","Missing Josephson terms explain readout-induced decay","Qubit decay scales with readout power, mechanism found"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000261,"raw_usage":{"total_tokens":1646,"prompt_tokens":1050,"completion_tokens":596,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":666,"completion_tokens_details":{"reasoning_tokens":521}},"tokens_in":666,"tokens_out":596,"duration_ms":6043,"temperature":1.0,"reasoning_tokens":521,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:20:31.176418+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the qubit relaxation rate during readout at fixed steady-state cavity photon number while stepping the drive frequency away from the cavity resonance: the paper predicts the renormalization drops to near zero once the detuning exceeds roughly $10\\chi_{ac}$. Alternatively, check linearity: at small $\\bar n_c$ the normalized rate shift $\\delta\\kappa_a^{\\mathrm{EME}}/\\kappa_a^{\\mathrm{EME}}(0)$ should grow linearly with $\\bar n_c$; a rate increase that is flat in $\\bar n_c$, or that persists at large detuning, would rule out the correlated-decay mechanism.","supporting_citations":[{"cited_title":"Lifetime renormalization of weakly anharmonic superconducting qubits: I. Role of number non-conserving terms","cited_arxiv_id":"1809.04667","evidence_quote":"Supplies the unitary-transformation formalism (Part I) that this paper extends to the coherently driven case and reduces to in the zero-drive limit."},{"cited_title":"Purcell effect with microwave drive: Suppression of qubit relaxation rate","cited_arxiv_id":"1401.5545","evidence_quote":"The dressed-dephasing hypothesis that predicts drive-induced increase; the paper argues it does not capture measured effective temperatures and compares against it."}],"review_version":1}