{"id":"6d2e68c5-af34-47f8-970f-f9140f64cf0d","arxiv_id":"2608.04684","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"In a silicon nitride microresonator, the linewidth of photon pairs generated by four-wave mixing narrows continuously from the cavity-limited scale (about 85 MHz) toward the pump scale (193 kHz) near the OPO threshold, and a linearized Lindblad model reproduces the trend.","lead":"A silicon nitride microresonator's photon-pair emission narrows in frequency from about 100 MHz to below 200 kHz as the pump approaches the oscillation threshold. The work maps this spectral evolution with four complementary techniques, providing a practical guide for designing narrowband integrated quantum light sources.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Interferometric and heterodyne measurements at the same detuning disagree by ~50–80×; the claimed near-threshold linewidth of 2π×193 kHz rests on one method while the other reports ~15 MHz.","rationale":"The paper has genuine strengths: four complementary measurement techniques, an analytic spectrum (Eq. 12) with a clear physical interpretation, and numerical cumulant simulations using independently estimated parameters. The qualitative trend of linewidth narrowing below threshold is supported by coincidence and autocorrelation data. However, the central quantitative claim—the linewidth 'reaching' 2π×193(5) kHz near threshold—is undercut by an unresolved factor-of-50–80 discrepancy between the heterodyne and interferometric measurements at the same nominal detuning. The reader's weakest_assumption focused on the single-Lorentzian fitting model, which is a legitimate systematic affecting all four extractions. I regard the inter-method contradiction as more load-bearing because it directly challenges the endpoint of the claimed narrowing and because the paper presents the interferometer as an independent confirmation while the example data contradict the heterodyne result. Near threshold in the real-ρbar regime, the theory itself predicts that the narrow γ− component dominates the spectrum, so the single-Lorentzian approximation for the heterodyne extraction is far less dangerous than an uncalibrated interferometer that cannot distinguish a 31 ns coherence time from a 1.6 μs one. The paper's asserted mapping between δeff and ℑρbar0 is a separate weakness for the theory comparison, but it does not affect the raw experimental narrowing. Since the contradiction is concrete and testable, and the paper could be revised by removing or re-interpreting the interferometer points and correcting the 'more than three orders of magnitude' claim (the reported data span a factor of ~500 at best), the appropriate verdict remains CONDITIONAL, matching the reader's verdict.","tokens_in":28912,"tokens_out":9131,"duration_ms":110021,"concrete_test":"Perform simultaneous or back-to-back interferometric and heterodyne measurements of the idler mode at δeff ≈ 1.27γ. For the interferometer, scan the delay over at least 500 m of fiber (≈2.5 μs of delay) so that a 1.6 μs coherence time would produce a measurable visibility decay; record an Allan-type stability trace of the fringe visibility at fixed delay to quantify slow-drift noise. Independently calibrate the interferometer by injecting the pump laser (γp = 2π×43.2 kHz, coherence length ~4.7 km) and verifying the measured |g(1)| matches this known linewidth. If the calibrated interferometer still reports ΔL ≈ 6 m at δeff = 139 MHz while heterodyne reports 193 kHz, the interferometer's slow-drift floor must be the cause and its near-threshold points should be excluded; if it instead reports long ΔL, the heterodyne value needs re-examination.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's headline quantitative result—that the idler linewidth narrows to 2π×193(5) kHz near the OPO threshold—is supported by only one of the two phase-sensitive methods. In Fig. 2(f), the interferometric measurement at δeff = 2π×139(1) MHz (≈1.27γ) yields ΔL = 6.2(2) m, i.e., τ_i = 31(1) ns; even correcting the caption's arithmetic (31 ns corresponds to 2π×10.3 MHz, not 2π×15 MHz), this implies a linewidth of order 10 MHz. In Fig. 2(h), heterodyne at δeff = 2π×139.2(3) MHz—the same detuning—yields τ_i = 1.6(1) μs, i.e., 2π×193(5) kHz. The two methods disagree by a factor of ~50–80 at the same operating point. The paper does not resolve this contradiction; it merely asserts that 'interferometric points agree well with the overall trend.' This matters because the central claim's endpoint (pump-linewidth scale) relies on heterodyne alone. If the interferometer is correct, the narrowing stops at ~10 MHz and the abstract's 'toward the pump-linewidth scale' is unsupported. If heterodyne is correct, the interferometer is not a valid cross-check and its use as 'final verification' is misleading. The most plausible explanation—long-term phase drift in the manually scanned Mach–Zehnder—should be demonstrated, not assumed. Separately, the single-Lorentzian fitting assumption identified by the reader also affects quantitative extraction, but near threshold (real ρbar regime) the theory itself says γ− dominates, so the heterodyne value is less sensitive to that issue; the unresolved inter-method discrepancy is the more direct threat.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports an experimental and theoretical study of the spectral linewidth of signal and idler photons generated by four-wave mixing in a high-Q Si3N4 microresonator as the pump is tuned toward the OPO threshold. Four techniques—photon-pair coincidence, second-order autocorrelation, variable-delay Mach–Zehnder interferometry, and heterodyne detection—are combined to track the linewidth from about the cavity-lifetime scale (≈2π×85 MHz) down to 2π×193(5) kHz near the OPO threshold. The theory section derives an analytic emission spectrum from a linearized Lindblad model, Eq. (12), in which the spectrum is a product of two Lorentzians with widths γ± = γ ± 2ρ̄, and numerical cumulant-expansion simulations are used to produce a predicted linewidth-vs-detuning curve. The paper concludes that the emission continuously narrows toward the pump-linewidth scale and that the single-Lorentzian fitting approximation is adequate near threshold.","tokens_in":29192,"tokens_out":14715,"duration_ms":198324,"significance":"If the central narrowing trend is established, the work is valuable: it connects the spontaneous quantum regime and the OPO regime in a single device, provides an analytic spectrum that goes beyond the common single-Lorentzian approximation, and combines four independent measurement modalities with explicit experimental parameters. Strengths include the absence of a fitted constant for the linewidth trend, the use of independently measured parameters (γ, g, ζ2/γ, P_in), the analytic derivation of Eq. (12), and numerical cross-checks with the QuantumCumulants.jl framework. However, the current manuscript contains an unresolved factor-of-50–80 discrepancy between the interferometric and heterodyne measurements at the same operating point, and this discrepancy sits exactly on the endpoint of the main quantitative claim, so the significance is contingent on a convincing resolution.","major_comments":[{"comment":"The interferometric and heterodyne measurements at the same effective detuning δeff ≈ 1.27γ are irreconcilable as presented. Figure 2(f) reports ΔL = 6.2(2) m, i.e., τ_i = 31(1) ns; using Eq. (1), this corresponds to γ_i/2π ≈ 10.3 MHz, not the stated 2π×15(1) MHz (the latter would require τ_i ≈ 21 ns). Figure 2(h), at δeff = 2π×139.2(3) MHz ≈ 1.27γ, reports τ_i = 1.6(1) μs, i.e., γ_i/2π = 193(5) kHz. The two methods therefore disagree by a factor of about 50–80 at the same operating point. If the true coherence time were 1.6 μs, the visibility over the measured path delays up to 6.2 m would decay by only a few percent, whereas the observed ΔL = 6.2 m implies a complete exponential decay on that scale. The sentence in §4 that 'the interferometric points agree well with the overall trend' is not supported by this data pair. Because the claimed near-threshold linewidth of 193 kHz rests on heterodyne alone, the paper must either provide a demonstrated cause for the interferometric discrepancy (for example, slow phase drift during the manual Mach–Zehnder scan, with a quantitative stability test) or withdraw the interferometric point as a cross-check.","section":"§2.2.2 and Fig. 2(f)–(h)"},{"comment":"The theory–experiment comparison is not quantitatively testable as presented because the two curves are plotted on different abscissae: the experimental data are shown versus δeff/γ, while the simulation is shown versus Im(ρ̄0)/γ. The text at the end of §3 states that 'the experimental measurement of δeff corresponds to Im(ρ̄0)', but δeff includes both nonlinear and thermal shifts, whereas Im(ρ̄0) is a purely Kerr-derived quantity; the two are not equal even up to the stated 'negligible second-order dispersion.' The paper later acknowledges that thermal effects 'primarily reparameterize the detuning axis,' but it provides no mapping, no thermal model, and no way to place the data and the simulation on the same axis. Since the central claim of theory–experiment agreement rests on Fig. 4, the authors should either derive and state the mapping between δeff and Im(ρ̄0), or overplot the data and simulation on a common axis using independently inferred parameters.","section":"§4 and Fig. 4"},{"comment":"All quantitative linewidth values in Fig. 4 are extracted with the single-Lorentzian assumption of Eq. (1), but the paper's own analytic spectrum, Eq. (12), is a product of two Lorentzians, and the paper shows that the single-exponential fit overestimates the linewidth in the imaginary-ρ̄ regime (inset of Fig. 4 and Fig. S7(b), where the fitted width changes from 117 MHz to 72 MHz). The far-detuned anchor values (e.g., γ_i = 2π×85(4) MHz from coincidences) are obtained in exactly this regime, so the starting point of the claimed narrowing is model-dependent. The authors do state that the approximation becomes accurate near threshold, but they should provide a quantitative estimate of the systematic bias on the far-detuned points and on the resulting compression ratio, rather than reporting the single-Lorentzian values as the sole experimental linewidths.","section":"§2.2, §4, Eq. (12), Eq. (13)"},{"comment":"The conclusion states that the measurements demonstrate 'a continuous linewidth narrowing of more than three orders of magnitude.' The data shown in Fig. 4 span from about 2π×85 MHz to 2π×193 kHz, which is a factor of about 440, i.e., roughly 2.6 orders of magnitude, not more than three. If the intent is to compare the endpoint to the pump linewidth (2π×43.2 kHz), the measured 193 kHz is still a factor of 4.5 above the pump, so 'toward the pump-linewidth scale' is appropriate but 'more than three orders' is not. This quantitative claim should be corrected.","section":"Conclusions and abstract"}],"minor_comments":[{"comment":"The caption reports 'τ_i = 31(1) ns' and 'γ_i = 2π×15(1) MHz' for the interferometric dataset; these two numbers are inconsistent, since τ = 31 ns corresponds to γ_i/2π ≈ 10.3 MHz via Eq. (1).","section":"Fig. 2 caption"},{"comment":"The inset uses the labels f and f1 without defining them in the caption; they should be tied explicitly to Eq. (13) so the reader can follow which model is the single-Lorentzian and which is the two-Lorentzian fit.","section":"Fig. 4 inset"},{"comment":"The fit-regime discrimination depends on a 'manually chosen critical value' χ̄. The paper should state the numerical value of χ̄ and show that the conclusions are insensitive to reasonable changes in it; otherwise the division into f1 and f2 regimes is not reproducible.","section":"Supplemental Sec. S6, Eq. (S.19)"},{"comment":"The derivation of the SU(1,1) evolution is attributed to Ref. [83], which is an 'In preparation' work. The authors should either move the derivation into the supplement or cite a published source, because the referee and readers cannot currently verify the step leading to Eq. (S.10).","section":"Supplemental Sec. S5"},{"comment":"The paper does not provide a data table for the individual linewidth points shown in Fig. 4, nor a data-availability statement. Given that one central data point is disputed, a table with all detunings, methods, fitted coherence times, and uncertainties would substantially improve transparency.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The strongest issue is the unresolved factor-of-50–80 discrepancy between the interferometric and heterodyne measurements at δeff ≈ 1.27γ. If the interferometer is correct, the near-threshold narrowing stops around 10 MHz and the abstract's 'toward the pump-linewidth scale' is unsupported; if the heterodyne is correct, the interferometric method is not functioning as a cross-check and its use in Fig. 4 as 'final verification' is misleading. I would ask the authors to provide a direct stability test or a side-by-side measurement with known systematic corrections, and to make the theory-experiment comparison in Fig. 4 quantitative by supplying the missing detuning mapping."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nRead this one for the experiment: it is the first continuous mapping of the idler linewidth from the spontaneous-FWM regime to the OPO threshold in an integrated Si3N4 microresonator, and it is backed by a linearized Lindblad model (Eq. 12) that generalizes the known FWM result of Vernon and Sipe by including dispersion. The theory has no fitted linewidth parameters—γ, g, ζ2/γ, and Pin are all independently measured or estimated. That is solid work, and the experimental effort spans four techniques: coincidences, g(2), interferometry, and heterodyne.\n\nThe soft spot is not in the theory. At δeff ≈ 1.27γ, the interferometric measurement gives τ_i = 31(1) ns, i.e., a few tens of MHz (the caption’s 2π×15 MHz does not match the 31 ns it quotes), while the heterodyne at the same detuning gives τ_i = 1.6(1) μs, i.e., 2π×193(5) kHz. That is a factor of 50–80 between two supposedly cross-checking methods at one operating point. The paper calls the interferometric points a “final verification” and says they agree with the trend, but that cannot be true at the point where both exist. Either the interferometer suffers from slow phase drift—plausible, but it needs to be shown, not asserted—or the heterodyne result is not representative. The abstract’s “toward the pump-linewidth scale” rests on the heterodyne alone.\n\nTwo smaller issues. The conclusions claim “more than three orders of magnitude” narrowing; the data span 85 MHz to 193 kHz, about 2.6 orders. And the theory–experiment comparison in Fig. 4 uses different x-axes (δeff/γ vs ℑρ̄₀/γ) with the mapping between them asserted rather than derived; the agreement is therefore qualitative in shape, not a quantitative point-by-point test. The single-Lorentzian fitting assumption is a moderate concern, but the authors analyze it themselves and show it mostly affects the far-detuned points.\n\nThe qualitative story—narrowing from cavity-limited toward pump-limited as threshold is approached—is probably right, and the parameter-free theory is a useful addition. But the inter-method discrepancy is load-bearing and must be resolved before the quantitative claims can be trusted.\n\nSend it to peer review. A good referee will ask for a reconciliation of the two phase-sensitive methods, a corrected caption, and a toned-down “orders of magnitude” claim. I would not cite the near-threshold linewidth until that is done.","headline":"A useful experimental mapping and a parameter-free theory, but an ~80x discrepancy between the interferometric and heterodyne linewidths at the same detuning leaves the central quantitative claim unresolved.","tokens_in":29937,"tokens_out":5826,"would_cite":false,"duration_ms":57173,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Signal and idler photons in a silicon nitride microresonator narrow from cavity-limited to pump-limited linewidth as the device approaches optical parametric oscillation, and the paper traces this to a linearized Lindblad model with…","keywords":["microresonator photon-pair source","four-wave mixing","optical parametric oscillation","linewidth narrowing","first-order coherence","silicon nitride","Lindblad master equation","heterodyne spectroscopy"],"falsifier":"Measure the heterodyne spectrum of the idler mode at high signal-to-noise for effective detunings between the OPO threshold ($\\delta_{\\rm eff}\\simeq 1.25\\gamma$) and the bifurcation point ($\\delta_{\\rm eff}\\simeq 1.31\\gamma$), and fit it against the product of two Lorentzians with widths $\\gamma_\\pm = \\gamma \\pm 2\\bar{\\rho}$ versus a single Lorentzian; the model predicts the narrow component $\\gamma_-$ dominates and that a single-Lorentzian fit overestimates the width, with the discrepancy growing as $\\delta_{\\rm eff}$ moves away from threshold.","tokens_in":28603,"feed_emoji":"⚛️","tokens_out":6103,"duration_ms":63233,"temperature":0.7,"pith_summary":"This paper tracks how the spectral linewidth of photon pairs generated by four-wave mixing in a silicon nitride microresonator behaves as the device is driven from the spontaneous quantum regime toward optical parametric oscillation. It finds that the emission linewidth narrows continuously from a cavity-limited scale of roughly 85 MHz to the scale of the pump laser linewidth, about 193 kHz, just before threshold. The authors trace this narrowing to a single mechanism: a linearized Lindblad model with dispersion, whose spectrum is a product of two Lorentzians with widths $\\gamma_\\pm = \\gamma \\pm 2\\bar{\\rho}$; near threshold the narrower component dominates and the linewidth approaches the pump linewidth rather than zero because of pump phase noise. Mapping this transition matters for designing integrated sources whose photons have the narrow, phase-stable spectra needed for quantum memories and long-distance quantum communication.","feed_headline":"Microresonator photon linewidth shrinks from 85 MHz to 193 kHz","feed_subtitle":"Tracking the SFWM-to-OPO transition shows cavity-limited emission narrows to the pump laser's linewidth.","key_machinery":"The central object is the linearized three-mode Lindblad master equation restricted to the pump, signal, and idler modes, with sideband detunings set by chromatic dispersion so that the signal-idler detuning $\\delta_s$ replaces the pump detuning. The quantity that carries the spectral transition is $\\bar{\\rho} = \\sqrt{g^2 n_0^2 - (\\delta_s + 2g n_0)^2}$: when $\\bar{\\rho}$ is imaginary the spectrum splits into two Lorentzians of equal width $\\gamma$ separated by $2|\\bar{\\rho}|$; when $\\bar{\\rho}$ becomes real the two Lorentzians sit at zero frequency with widths $\\gamma_\\pm = \\gamma \\pm 2\\bar{\\rho}$. The linewidth narrowing is the dominance of the $\\gamma_- = \\gamma - 2\\bar{\\rho}$ component near threshold, with $\\gamma_-$ formally vanishing at the OPO threshold. The effective detuning $\\delta_{\\rm eff}$, measured in real time by sideband probing, lets the experiment map the theory's detuning axis onto the experimental operating point.","core_discovery":"The paper establishes that in a silicon nitride microring driven by a continuous-wave pump, the spectral linewidth of the signal and idler photons emitted by spontaneous four-wave mixing narrows continuously as the effective pump-cavity detuning $\\delta_{\\rm eff}$ is reduced toward the OPO threshold, from the cavity-lifetime-limited scale of about $2\\pi\\times 85(4)$ MHz (idler) down to $2\\pi\\times 193(5)$ kHz near threshold. The authors show that a three-mode Lindblad master equation, linearized around the pumped steady state and including chromatic dispersion, yields an emission spectrum of the form $\\nu_s(\\omega) = g^2 n_0^2 / \\bigl[|\\gamma/2 - \\bar{\\rho} + i\\omega|^2 |\\gamma/2 + \\bar{\\rho} + i\\omega|^2\\bigr]$, with $\\bar{\\rho} = \\sqrt{g^2 n_0^2 - (\\delta_s + 2g n_0)^2}$. When $\\bar{\\rho}$ is imaginary the spectrum splits into two equal-width Lorentzians separated by $2|\\bar{\\rho}|$; when $\\bar{\\rho}$ becomes real the two Lorentzians sit at zero frequency with widths $\\gamma_\\pm = \\gamma \\pm 2\\bar{\\rho}$, and the narrower component $\\gamma_- = \\gamma - 2\\bar{\\rho}$ governs the observed narrowing, formally vanishing at the OPO threshold. This prediction is supported by four complementary measurements—temporal coincidences, second-order autocorrelation, Mach–Zehnder interferometry, and heterodyne beats—and by numerical cumulant-expansion solutions that reproduce the narrowing trend and its sensitivity to pump-power fluctuations.","pith_inferences":["Extension: because the model's linewidth floor is set by pump phase noise, reducing the pump laser linewidth should directly narrow the emitted photons near threshold; this is testable by injecting a sub-kHz-linewidth pump.","Extension: the product-of-two-Lorentzians structure implies that near threshold the photon-pair wavepacket develops a slow, near-exponential tail with a distinct narrow component; this could be probed by measuring biphoton correlation asymmetry in the real-$\\bar{\\rho}$ regime with higher timing resolution.","Extension: the wavelength-dependent coupling that causes the observed signal-idler linewidth asymmetry could be deliberately engineered to produce unequal linewidths on demand, which may be useful for asymmetric quantum-network tasks.","Extension: the linearized theory predicts a cusp in the linewidth at the bifurcation point where $\\bar{\\rho}$ crosses zero; a direct measurement of the linewidth's derivative across $\\delta_{\\rm eff}\\approx 1.31\\gamma$ could test the square-root eigenvalue splitting without relying on the fit models."],"forward_implications":["The emission linewidth narrows continuously by more than three orders of magnitude across the spontaneous-to-stimulated transition, so a single device can provide both cavity-limited broadband pairs and narrowband, coherent emission depending on operating point.","Far from threshold the linewidth is set by the cavity lifetime, while near threshold it approaches the pump laser linewidth, making pump phase noise the practical floor for source coherence.","The single-exponential (single-Lorentzian) fitting assumption overestimates the linewidth far from threshold, but becomes increasingly accurate as the system approaches the OPO threshold, where the narrow $\\gamma_-$ component dominates.","The coincidence and autocorrelation methods lose validity near and above threshold, leaving heterodyne detection as the only reliable way to extract linewidths in the OPO regime.","The numerical cumulant-expansion simulations reproduce the observed narrowing trend and show that the linewidth is highly sensitive to pump-power fluctuations near threshold, matching the experimental scatter."],"supporting_citations":[{"why":"Supplies the linearized quantum-fluctuation framework for Kerr microresonators below threshold that the paper adapts to the three-mode model.","marker":"[27]"},{"why":"Provides the mean-field and linearization treatment of strongly driven microring four-wave mixing whose sideband equations of motion are extended here to include dispersion.","marker":"[64]"},{"why":"Supplies the numerical cumulant-expansion solver used to simulate linewidths beyond the linearized regime.","marker":"[73]"},{"why":"Defines the Wiener–Khintchine relation connecting the first-order correlation function to the emission spectrum.","marker":"[49,50]"},{"why":"Establishes the standard Lorentzian single-exponential assumption for microresonator photon-pair linewidths that the paper tests and refines.","marker":"[51,52]"},{"why":"Supplies the derivations of the SU(1,1) operator equations, the analytic correlation function, and the fit functions used to extract linewidths.","marker":"[67]"},{"why":"Provides the expression for the Kerr coupling strength g whose temperature dependence is checked in the supplemental material.","marker":"[84]"}],"fun_headline_variants":["Photon linewidth shrinks 440-fold in silicon nitride microring","From 85 MHz to 193 kHz: microring photon linewidth collapse","Cavity-limited to pump-limited: microring photon linewidth evolution","Photon linewidth narrows toward pump scale near OPO threshold"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The linewidth numbers all come from fitting each measured decay as a single exponential (a single Lorentzian spectrum), even though the paper's own analytic spectrum is a product of two Lorentzians; if the true decay is biexponential, the extracted widths—including the cavity-limited reference scale—are systematically off.","fun_headline_variants_meta":{"raw":{"variants":["Photon linewidth shrinks 440-fold in silicon nitride microring","From 85 MHz to 193 kHz: microring photon linewidth collapse","Cavity-limited to pump-limited: microring photon linewidth evolution","Photon linewidth narrows toward pump scale near OPO threshold"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000666,"raw_usage":{"total_tokens":3054,"prompt_tokens":974,"completion_tokens":2080,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":590,"completion_tokens_details":{"reasoning_tokens":1998}},"tokens_in":590,"tokens_out":2080,"duration_ms":17894,"temperature":1.0,"reasoning_tokens":1998,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T18:57:42.060170+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the heterodyne spectrum of the idler mode at high signal-to-noise for effective detunings between the OPO threshold ($\\delta_{\\rm eff}\\simeq 1.25\\gamma$) and the bifurcation point ($\\delta_{\\rm eff}\\simeq 1.31\\gamma$), and fit it against the product of two Lorentzians with widths $\\gamma_\\pm = \\gamma \\pm 2\\bar{\\rho}$ versus a single Lorentzian; the model predicts the narrow component $\\gamma_-$ dominates and that a single-Lorentzian fit overestimates the width, with the discrepancy growing as $\\delta_{\\rm eff}$ moves away from threshold.","supporting_citations":[{"cited_title":"Quantum dynamics of kerr optical frequency combs below and above threshold: Spontaneous four-wave mixing, entanglement, and squeezed states of light,","cited_arxiv_id":null,"evidence_quote":"Supplies the linearized quantum-fluctuation framework for Kerr microresonators below threshold that the paper adapts to the three-mode model."},{"cited_title":"Strongly driven nonlinear quantum optics in microring resonators,","cited_arxiv_id":null,"evidence_quote":"Provides the mean-field and linearization treatment of strongly driven microring four-wave mixing whose sideband equations of motion are extended here to include dispersion."},{"cited_title":"QuantumCumulants.jl: A Julia framework for generalized mean-field equations in open quantum systems,","cited_arxiv_id":null,"evidence_quote":"Supplies the numerical cumulant-expansion solver used to simulate linewidths beyond the linearized regime."},{"cited_title":"Supplemental material,","cited_arxiv_id":null,"evidence_quote":"Supplies the derivations of the SU(1,1) operator equations, the analytic correlation function, and the fit functions used to extract linewidths."},{"cited_title":"Optical hyperparametric oscillations in a whispering-gallery- mode resonator: Threshold and phase diffusion,","cited_arxiv_id":null,"evidence_quote":"Provides the expression for the Kerr coupling strength g whose temperature dependence is checked in the supplemental material."}],"review_version":1}