{"id":"4b3a157e-6b19-4dc9-a8e5-f83576226d95","arxiv_id":"2607.25133","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Hyperboloidal slicing plus Chebyshev discretization computes optical QNMs in dispersive media, and pseudospectra in L2 versus energy norms reveal norm-dependent spectral instability.","lead":"The paper adapts hyperboloidal compactification and pseudospectrum tools from black-hole physics to one-dimensional open optical cavities with Lorentz dispersion. It shows how to compute quasi-normal modes without artificial absorbers and how the choice of norm changes apparent resonance stability.","discovery_kind":"new_application","skeptic_critique":{"model":"moonshotai/kimi-k3","headline":"The discretized pencil provably contains cavity-independent eigenvalues (ω=im, integer spacing) that arise from degeneration of the radiation condition at ℐ+, yet only self-convergence — no benchmark against the exact 1D-cavity resonance equation — distinguishes physical QNMs from formulation-intris","rationale":"Good-faith read: this is a methods paper transferring a mature gravitational-physics toolkit (hyperboloidal compactification + pseudospectra, refs [1,2,8,17,19–21]) to 1D dispersive optics. The genuinely supported core — a working multi-domain Chebyshev implementation of the Maxwell–Lorentz pencil (Eqs. 16–28) and the demonstration that pseudospectra are norm-dependent (Figs. 8–12, L2 vs energy) — is sound as methodology and honestly presented; the authors themselves flag the antibound modes' status as unresolved and list Kramers–Kronig compliance as future work. My concern overlaps the reader's weakest assumption only partially: the reader questioned whether one height function correctly encodes outgoing conditions for the dispersive system (a modeling-alignment question, real but benign here since ε∞=μ0=1 in the exterior makes the Bizoń–Mach slices asymptote to the true characteristics at high frequency, and the compact-support restriction is explicit in §II.A). I instead identify a sharper, internally demonstrable issue: the formulation provably generates a family of eigenvalues that are not cavity resonances (integer-spacing ω=im follows from degeneration of the radiation condition at ℐ+, independent of cavity parameters), and the paper's only convergence evidence is self-referential, so neither the physical identification of the Regge branches nor the headline stability contrast of Fig. 14 is secured. This is not fatal — the Regge branches, Weyl scaling (Fig. 15), and spacing convergence (Fig. 5) are plausibly correct — and it is exactly the kind of gap the reader's CONDITIONAL verdict already contemplates (external baselines, code release, scoping of antibound modes). Hence UNCHANGED: CONDITIONAL stands, with the benchmark above as the decisive, nearly costless condition for lifting it. Note also that the paper gets credit for honest scope-setting and for correct handling of the norm/normalization distinction in §IV.D, which is more careful than most of the optics QNM literature.","tokens_in":25040,"tokens_out":10118,"duration_ms":306183,"concrete_test":"For the non-dispersive slab of Figs. 8/10 (ε_II=2, air exterior, length L), write the exact QNM condition: match A e^{−iωx}+B e^{iωx} inside to single outgoing branches outside at both interfaces, giving a 4×4 determinant (equivalently e^{2inωL}=±(n−1)/(n+1)-type transcendental equations), solvable to machine precision by root-finding. Overlay roots on the Fig. 4 spectrum: (i) do the Regge branches agree root-by-root? (ii) does the exact pole set contain anything at ω=im (parameter-independent integers)? If (i) passes and (ii) is empty, the framework is validated and the imaginary family is demoted to artifact, voiding the Fig. 14 contrast. If (i) fails, the QNM computation itself has a systematic error.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's own analyticity argument in §III.B.1 shows more than the authors draw out. At ω=im (m∈ℕ), BOTH free-space branches become analytic at the compactified endpoints: (1+y)^{-iω}=(1+y)^m and (1−y)^{-iω}=(1−y)^m are polynomials. This means the outgoing condition degenerates at these frequencies — it no longer excludes anything. A dimension count then gives that every ω=im is an eigenvalue for ANY compactly supported cavity: at these frequencies each exterior contributes a 2-dim analytic solution space, the cavity a 2-dim space, and the 4 interface conditions (Eq. 26) leave a solution space of dimension ≥2 regardless of cavity length and permittivity. So the \"antibound\" family is structurally tied to the tanh compactification map, not to the resonator; genuine antibound/virtual poles of a 1D slab would sit at parameter-dependent locations. The paper half-acknowledges this (\"eigenvalues do include scattering resonances but potentially also other 'resonant' frequencies\", §I) but then builds the headline stability contrast — Regge branches stable, purely imaginary modes split under a 10^{-6} permittivity perturbation (Fig. 14) — partly on this family. If these eigenvalues are formulation artifacts, that contrast is physically vacuous. Compounding this, the only validation offered is self-convergence (Eq. 29 measures |1−ω_n/ω_N| against the largest-N run). Spectral collocation converges to eigenvalues of the CONTINUUM compactified operator — which, per the above, includes non-resonance eigenvalues — so Fig. 3 cannot certify that the Regge branches are the true scattering resonances either. The extreme sensitivity reported at 3i (perturbation ~10^{-18}, §IV.E.1) is itself characteristic of pseudospectra near essential spectrum of hyperboloidal operators, a known pathology in the gravitational literature the method is imported from. In 1D this is all cheaply checkable: the constant-permittivity slab of Figs. 8/10 has an exact transcendental resonance condition. Its 1","agreement_with_reader":"partial"},"referee_report":{"model":"moonshotai/kimi-k3","summary":"The manuscript transfers the hyperboloidal/compactified-slice QNM methodology from gravitational perturbation theory to a 1D open optical cavity with Lorentz/Drude dispersion. The Maxwell–Lorentz system is rewritten in Bizoń–Mach-type coordinates, reduced to first order in time, assembled as a non-Hermitian generalized pencil on three Chebyshev subdomains with interface matching, and solved without PMLs. The authors report two spectral families (purely imaginary ω=im and logarithmic Regge QNM branches), show self-convergence of one eigenvalue and of one pseudospectral level, compute L2 and “energy-norm” pseudospectra, and argue that Regge QNMs are comparatively robust while purely imaginary modes are unstable under a 10^{-6} permittivity perturbation.","tokens_in":25427,"tokens_out":5908,"duration_ms":202264,"significance":"If validated, the work is a useful and timely bridge between hyperboloidal QNM methods and optical resonator theory: it gives an explicit non-Hermitian pencil rather than a fitted model, treats outgoing conditions geometrically rather than by absorbing layers, and makes a falsifiable stability claim by directly perturbing the permittivity. The detailed coordinate transformation, first-order reduction, interface conditions, Radau/Lobatto grid choices, and the pseudospectrum convergence spot-check are genuine strengths, as is the emphasis that pseudospectral stability is norm-dependent. The central methodology appears defensible; the main risk is not the framework but the physical identification of part of the computed spectrum and the quantitative weight placed on the energy-norm comparison.","major_comments":[{"comment":"The analyticity argument given for f±(y)=2^{iω}(1±y)^{-iω} implies a stronger degeneration than the manuscript draws out: at ω=im both free-space branches become polynomial/regular at the compactified endpoints, so the outgoing condition excludes nothing there. A dimension count for the three-domain problem then suggests cavity-independent eigenvalues at ω=im for essentially any compactly supported cavity. Before these modes are called “antibound” and used in the Fig. 14 stability contrast, please test whether they move with cavity length/permittivity and whether they persist under a different height function h(y) or compactification g(y); if not, they should be relabeled as compactification/ℐ+ eigenvalues and removed from the physical conclusion.","section":"§III.B.1, Eq. (30); Fig. 4; Fig. 14; Conclusion A item 1"},{"comment":"The only validation is self-convergence against the largest-N run (Eq. (29)) and a single pseudospectral level at 1+2i. Spectral collocation can converge to the continuum compactified operator while still converging to formulation-intrinsic eigenvalues of the kind above. The Regge-branch claim needs at least one independent benchmark: the exact transfer-matrix resonance equation for a piecewise-constant 1D cavity with outgoing conditions in x, an equivalent complex-scaling/PML calculation, or an error-controlled resonance computation. Please report agreement for several Regge QNMs and state explicitly whether the ω=im family is absent, present, or parameter-dependent in that benchmark.","section":"§III.A, Eq. (29); §IV.G, Fig. 13"},{"comment":"The energy-norm pseudospectrum is central to the norm-dependence claim, but the functional setting is under-specified. Eq. (35) is written with E² rather than |E|², omits the ω0 polarization-potential term and the dispersive Brillouin factor ∂(ωε)/∂ω, and for Γ>0 is not a conserved energy; H is reconstructed nonlocally through an integration matrix after Eq. (46). For the first-order pencil the Gram matrix, positivity on complex fields, and adjoint used in s_min are not defined, and Eq. (32) perturbs only L although a physical δε perturbation changes both L and M. Without these details, the quantitative L2-vs-energy contrast in Figs. 10–12 is not yet reproducible.","section":"§IV.B–F, Eqs. (31)–(35), (44)–(46)"},{"comment":"The outgoing construction relies on a single Bizoń–Mach height function tuned to one asymptotic velocity, while Eq. (3) allows the exterior regions to be Lorentz/Drude media. The compact-support/homogeneous-exterior restriction should be made precise: in the numerical examples the exterior must be nondispersive and nonabsorbing (e.g. ωp=Γ=0 outside, constant ε), or the authors must explain how one asymptotic group velocity encodes outgoing waves for a dispersive exterior over the whole frequency range shown. A concrete check would be to let exterior Drude parameters tend to vacuum and show spectral insensitivity; otherwise the ℐ+ regularity argument is not controlled.","section":"§II.A after Eq. (11); §II, Eq. (3)"}],"minor_comments":[{"comment":"Clarify the Fourier convention early and repeatedly: with ∂t↦iω, decay corresponds to Im ω>0, opposite to the common optics e^{-iωt} convention. This affects the words “antibound,” “decaying,” and the interpretation of Figs. 4–7.","section":"Global notation"},{"comment":"Units/nondimensionalization are inconsistent: ε∞ appears as relative permittivity while μ0 and ε0 are retained; ε∞μ0−y² compares dimensioned and dimensionless quantities unless c=ε0=μ0=1 is stated. Eq. (A4) contains “ω−0²” and Eq. (A6) mixes ∆, c, ε∞ and wp.","section":"Eq. (18), Eq. (44), Appendix A"},{"comment":"The generalized-pseudospectrum definitions mix λ and iω in the same displayed equation, and the third characterization perturbs L only. Please align notation with zM−L and state whether perturbations to M are included.","section":"Eq. (32)"},{"comment":"Several figure references are off: the permittivity-perturbation spectrum is Fig. 14 (not Fig. 13) and the Weyl scaling is Fig. 15 (not Fig. 14). Eq. (29) uses n+1 points while the pencil dimension is described with N+1; define N consistently.","section":"§III, §V cross-references"},{"comment":"Typos include “jl fluid dynamics,” “This works present,” “far-disctance,” “tratament,” “descretization,” “inseting,” “orthoganality,” “Kronig/König,” and Appendix B5 “[−1,−1]” should be [−1,1]. In B10 the endpoint statement duplicates x0; distinguish the left/right Radau endpoint conventions.","section":"Copyediting"},{"comment":"No code, grid sizes, tolerance, ε contour values, complex-plane sampling, or cavity lengths a,b are tabulated. Please provide the parameters for every figure and, ideally, a small reproducible script for the generalized pencil and singular-value evaluation.","section":"Reproducibility"}],"recommendation":"major_revision","confidential_remarks":"The paper fits a journal interested in open-wave/QNM methods and non-selfadjoint spectral stability. The research program is continuous with the authors’ prior gravity/optics work and thesis, but the manuscript should stand on its own validation. I would suggest handling by an editor comfortable asking a scattering-resonance expert to check specifically whether the ω=im family is a resonance/virtual pole or a compactification eigenvalue; that point is technical and potentially decisive for one of the two headline conclusions."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The one thing to know: this is a careful methods transfer of hyperboloidal compactification plus pseudospectra from the authors’ gravity program into 1D Lorentz/Drude optics, with a real point about norm choice. The stability story they advertise is weaker than the abstract suggests.\n\nWhat is actually new is the joint packaging: multi-domain Chebyshev pencil (Lobatto cavity, Radau exteriors), auxiliary-field Maxwell–Lorentz system on Bizoń–Mach slices, concrete spectra, L2-versus-energy pseudospectra, a small permittivity perturbation, and a Weyl-counting check. The pieces are known ([1,2], Ansorg/Macedo, Burgess–König, Trefethen), but the optical assembly is done properly. Derivations of the first-order reduction, interface matching, and grid choices are clear; Fig. 3 shows spectral convergence; they compare norms instead of pretending L2 is physical. That is honest craft.\n\nThe soft spot that matters is the pure-imaginary family. Their own free-space argument (§III.B.1) shows that at ω=im both exterior branches become analytic (polynomials) at y=±1, so the outgoing condition degenerates. A dimension count then makes every such frequency an eigenvalue for any compactly supported cavity—tied to the tanh map, not to resonator poles. They half-flag that eigenvalues can exceed scattering resonances, then still build the Regge-stable / imaginary-fragile contrast (Fig. 14) partly on that family. If those modes are formulation modes, that contrast is not an optical stability result. Validation is only self-convergence against the largest-N run; there is no check against the exact transcendental condition for a constant-ε slab, which is cheap in 1D. Extreme sensitivity near 3i also matches known hyperboloidal essential-spectrum pathology from the gravity literature they import. Minor limits: 1D only, no code, antibound physics left open, Kramers–Kronig deferred.\n\nWho it is for: people already doing non-Hermitian photonics or geometric QNM methods. Math and citations look solid; circularity is low. It deserves a serious referee who will demand a slab benchmark and a quarantine of the imaginary family—not a desk reject. Engage for the pipeline and the norm lesson; do not treat the stability headline as settled until the artifact issue is cleaned up.","headline":"Useful 1D hyperboloidal–pseudospectrum pipeline for dispersive optics, but the stability headline partly rests on cavity-independent imaginary eigenvalues and lacks an external resonance benchmark.","tokens_in":25790,"tokens_out":586,"would_cite":false,"duration_ms":29845,"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":"Hyperboloidal coordinates plus pseudospectra let optical cavity resonances be computed and their stability judged in a way that depends on the norm used to measure perturbations.","keywords":["quasi-normal modes","hyperboloidal compactification","pseudospectrum","non-Hermitian operators","optical cavities","Lorentz–Drude model","spectral stability","Chebyshev spectral methods"],"falsifier":"Compute the same one-dimensional dispersive cavity both with the hyperboloidal multi-domain Chebyshev pencil and with an independent method (for example complex scaling or a well-converged PML), then apply a controlled permittivity perturbation of known size; if the two methods disagree on which eigenvalues move and by how much, or if the energy-norm pseudospectrum fails to predict the observed shifts, the central claim fails.","tokens_in":25388,"feed_emoji":"📡","tokens_out":984,"duration_ms":22648,"temperature":0.7,"pith_summary":"Open optical cavities leak energy, so their natural resonances (quasi-normal modes) are complex frequencies of a non-Hermitian operator and can be highly sensitive to small changes. This paper transfers a geometric compactification from gravitational-wave theory—hyperboloidal time slices that reach null infinity—to one-dimensional dispersive electromagnetism, so outgoing radiation is built into the coordinate system rather than imposed by artificial layers. With a Lorentz-model permittivity written through auxiliary fields, Chebyshev spectral methods then yield the resonance spectrum, including both the usual logarithmic Regge branches and a set of purely imaginary eigenvalues. Pseudospectrum contours in both the L2 and energy norms show how far those eigenvalues can move under small perturbations, and a direct permittivity perturbation of order 10^{-6} confirms that Regge-branch modes stay put while the imaginary ones shift and acquire real parts. The central message is that whether an optical resonance looks stable is inseparable from the functional setting in which the perturbation is measured, giving a practical route to stability and sensitivity analysis in open photonic systems.","feed_headline":"Norm choice decides when optical resonances look stable","feed_subtitle":"Hyperboloidal slices plus energy-norm pseudospectra show which cavity modes survive tiny material noise","key_machinery":"Hyperboloidal compactification (Bizoń–Mach-type height function and spatial map) that turns outgoing boundary conditions into regularity at the compactified endpoints, paired with the ε-pseudospectrum of the resulting generalized non-Hermitian eigenvalue pencil measured in different norms.","core_discovery":"Combining the hyperboloidal formulation of the Maxwell–Lorentz system with pseudospectrum analysis makes it possible both to compute optical quasi-normal modes for compact cavities and to quantify their spectral stability, and that assessment depends strongly on the choice of scalar product (L2 versus energy norm).","pith_inferences":["If energy-norm fragility of the imaginary eigenvalues survives in 2-D and 3-D cavities, device designers may need to treat far-field or material noise as a first-class design constraint rather than a numerical nuisance.","The observed clustering of equal-imaginary-part modes in the energy pseudospectrum suggests a selection rule for which resonances dominate late-time transients—testable by ringdown simulations.","Disagreement between L2 and energy contours offers a cheap diagnostic for whether a published optical QNM spectrum is likely to be experimentally robust."],"forward_implications":["Optical QNMs can be normalized rigorously in compactified coordinates, removing the usual divergence that blocks mode-volume calculations.","Energy-norm pseudospectra give a physically grounded map of which resonances will drift under realistic material or geometric noise.","Purely imaginary “antibound” eigenvalues are flagged as the most fragile and should be checked in time-domain expansions before they are discarded or kept.","A Weyl-type counting law appears to hold for the non-selfadjoint optical operator, linking cavity length to high-frequency resonance density.","The same geometric-plus-pseudospectrum pipeline can be ported to higher-dimensional resonators and to causal dispersive models that obey Kramers–Kronig relations."],"fun_headline_variants":["Norm choice shapes stability of optical cavity resonances","Hyperboloidal QNMs plus pseudospectra tie stability to the norm","Energy norm versus L2 alters optical resonance stability readout","Pseudospectra show optical QNM stability depends on scalar product","Compact-cavity optical modes: stability hinges on chosen norm"],"cache_read_input_tokens":16512,"weakest_assumption_plain":"For a compact cavity in a uniform exterior, one fixed height function tuned to a single asymptotic group velocity correctly encodes all outgoing radiation at the compactified ends, so the discrete eigenvalues that appear are the physical optical resonances rather than coordinate artifacts.","fun_headline_variants_meta":{"raw":{"variants":["Norm choice shapes stability of optical cavity resonances","Hyperboloidal QNMs plus pseudospectra tie stability to the norm","Energy norm versus L2 alters optical resonance stability readout","Pseudospectra show optical QNM stability depends on scalar product","Compact-cavity optical modes: stability hinges on chosen norm"]},"model":"grok-4.5","effort":"low","cost_usd":0.003654,"raw_usage":{"total_tokens":1160,"prompt_tokens":772,"num_sources_used":0,"completion_tokens":87,"cost_in_usd_ticks":36544000,"prompt_tokens_details":{"text_tokens":772,"audio_tokens":0,"image_tokens":0,"cached_tokens":128},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":301,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":772,"tokens_out":87,"duration_ms":6438,"temperature":1.0,"reasoning_tokens":301,"cache_read_input_tokens":128,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-31T00:26:08.187768+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Compute the same one-dimensional dispersive cavity both with the hyperboloidal multi-domain Chebyshev pencil and with an independent method (for example complex scaling or a well-converged PML), then apply a controlled permittivity perturbation of known size; if the two methods disagree on which eigenvalues move and by how much, or if the energy-norm pseudospectrum fails to predict the observed shifts, the central claim fails.","supporting_citations":[],"review_version":1}