{"id":"85039540-a9b2-4c84-bc56-ed233b9101d6","arxiv_id":"2506.18750","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"For singular isothermal ellipsoid lenses that produce four gravitational-wave images, three-image search templates reduce waveform mismatch by roughly a factor of two to five relative to two-image templates, and geometric optics holds for lens masses above about 10^5 solar masses at ground-based…","lead":"This paper studies how gravitational waves are lensed by elliptical galaxies, which can create two or four copies of a signal. It finds that using three-image templates to search for these lensed signals reduces waveform mismatch by factors of 1.5 to 5 compared to standard two-image templates, and maps when the simple geometric-optics approximation breaks down.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"GO-validity threshold for SIE rests on an O(1/w) expansion validated only for SIS; the 10^5 M_sun criterion is unverified against full wave optics for elliptical lenses.","rationale":"The strongest claim has two parts: (i) the GO-validity mass threshold for SIE, and (ii) the waveform-mismatch gain from three-image templates. Part (ii) is well supported: Appendix A gives a PSD-independent derivation in the extreme-GO limit showing the overlap increases monotonically with the number of included images (Eqs. A17-A19), and the numerical mismatch results are consistent with that ordering. Part (i) is the load-bearing piece. The threshold rests on a first-order asymptotic expansion, Eq. (16), checked only against the SIS exact solution. The SIE differs qualitatively: four images, astroid caustics, and divergent third and fourth derivatives near folds and cusps. The expansion is non-uniform near those regions, so the Delta-based criterion may not mark the true GO-to-wave-optics transition. The paper concedes the difficulty of full wave optics for SIE, but then builds event classifications and detection recommendations on that unvalidated criterion. The reader identified the same weakness, and I agree. The CONDITIONAL verdict remains appropriate; the concern is genuine and should be settled by a concrete wave-optics check. Secondary issues (single ellipticity, unspecified PSD, idealized templates) are real but less central, because the mismatch advantage has independent analytical support and is likely robust to those choices.","tokens_in":24143,"tokens_out":16280,"duration_ms":161454,"concrete_test":"Numerically evaluate the full SIE diffraction integral F_wo(w) for e = 0.2 at representative four-image source positions (e.g., y = 0.05, phi = 0 deg, 45 deg, 90 deg) using a Picard-Lefschetz or Levin-type solver, over w spanning 0.1 max_j |Delta^(j)| to 10 max_j |Delta^(j)|. Compare |F_GO/F_wo - 1| at the predicted threshold w = max_j |Delta^(j)| and at twice that threshold. If the fractional error is not below 1 at the threshold, the quasi-GO criterion Eq. (27) is not validated for SIE and the quoted mass threshold must be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central quantitative claim — that GO is valid for SIE lenses above MLz ~ 10^5 M_sun, Eq. (27) — rests entirely on the quasi-GO expansion of Takahashi, Eq. (16), truncated at O(1/w). This expansion is validated in Fig. 1 only for the SIS lens, against the exact wave-optics series Eq. (20). For the SIE, no independent wave-optics computation is provided. The paper explicitly states in Sec. II that computing F for SIE in the wave-optics regime 'remains a challenging task' and therefore assumes the quasi-GO framework carries over. That assumption is load-bearing: the threshold MLz(min) = |Delta^(j)|/(8 pi fmin) and the resulting event classification (Fig. 7, Table I) inherit any error in the first-order expansion. Near the astroid caustic, the Hessian eigenvalue of an image tends to zero, so Delta^(j) diverges and the O(1/w) expansion is not uniform. The SIS check does not exercise this cusp/fold regime, since the SIS has only a radial caustic with a different singularity structure. If the first-order Delta criterion misestimates the true breakdown scale for SIE, the 10^5 M_sun threshold and the fraction of events assigned to wave-optics versus GO classes (e.g., 59% WO at 10^5 M_sun) would shift. A related wrinkle: the SIS validation text says max_j Delta^(j) rather than max_j |Delta^(j)|; Eq. (27) correctly uses absolute values, but the figure caption as written could mislead the threshold location.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper extends gravitational-wave (GW) lensing studies from axisymmetric models to the singular isothermal ellipsoid (SIE) lens model. Using Takahashi's O(1/w) quasi-geometrical-optics expansion, it derives a lens-mass threshold, Eq. (27), below which the GO approximation for a given SIE image is claimed to fail, and uses this threshold to map GO validity in the two-image and four-image regimes. It then classifies four-image events into 14 macrolensing/microlensing classes and computes waveform mismatches between simulated four-image signals and two-image (2L) and three-image (3A, 3B) templates, reporting that three-image templates reduce median mismatches by factors of 1.5-5. An analytic extreme-GO overlap calculation in Appendix A is presented to explain the template ranking.","tokens_in":24475,"tokens_out":16818,"duration_ms":171461,"significance":"If the central criterion is correct, the paper provides the first quantitative GO-validity mass scale for elliptical GW lenses and a concrete, physically motivated template-ranking result relevant to future lensed-GW searches. The paper's strengths include a parameter-free analytic overlap derivation in Appendix A, clear use of glafic to obtain SIE image properties, and a falsifiable threshold statement. However, the central mass threshold relies on an extrapolation of a first-order expansion to a regime not independently validated by wave-optics computations, and the mismatch numbers are obtained in an idealized setting. With these points addressed, the paper would be a useful contribution to the lensed-GW template literature.","major_comments":[{"comment":"The central GO-validity criterion is not yet adequately supported for the SIE. The O(1/w) expansion in Eq. (16) is tested only against the exact SIS wave-optics result in Fig. 1, and Sec. II explicitly states that computing F(w) for the SIE in the wave-optics regime remains challenging; the paper therefore assumes the expansion carries over. This matters because Eq. (17) has denominators proportional to powers of the Hessian eigenvalues |λ^(a)|, so Δ^(j) diverges when an image approaches the critical curve, and the SIS test—which has only a radial pseudo-caustic—does not exercise the fold/cusp geometry of the SIE. Moreover, the treatment of the y→yr limit is internally inconsistent: Fig. 4 and the text state that M^(2)_Lz(min)→∞, yet also assert that GO remains valid because |µ^(2)|^{1/2}Δ^(2)→0. For the SIS benchmark, Eqs. (22a) and (23) give |µ_-|^{1/2} Δ_- = -1/[8 y^{3/2}(1-y)^{3/2}], which diverges as y→1; hence the claimed limit is not a general property and is not proven for the SIE. A magnification-weighted validity condition, or a direct SIE wave-optics check at representative source positions near the fold and cusp, is needed before Eq. (27), Fig. 7, and Table I can be taken as quantitative.","section":"§V, Eq. (27) and Fig. 4"},{"comment":"The mismatch comparison is an idealized 'known lens parameters' statement, and the reported reduction factors of 1.5-5 should be understood as upper bounds on template performance, not as search-ready template-bank results. The text says the templates are parameterized using the properties of the brightest corresponding images within the true four-image signal, and Eq. (31) maximizes only over tc and φc; no search over lens parameters such as e, y, φ, or M_Lz is performed. The noise power spectral density Sn(f) entering Eq. (33) is never specified, and the numerical results use only e=0.2, fmin=20 Hz, and one source mass configuration. The relative ranking ϵ3A < ϵ3B < ϵ2L may be robust, but the absolute mismatch values and the claim of reducing mismatches from O(10^{-1}) to O(10^{-2}) should either be explicitly qualified as ideal, or supported by a search over lens parameters for representative events.","section":"§VII, Eq. (33), Table II"}],"minor_comments":[{"comment":"The validity condition is written as w > max_j Δ^(j), but Δ^(j) can be negative (for the SIS second image, Δ_- is negative), so the largest correction is controlled by max_j |Δ^(j)|. Please use absolute values consistently in the text, Eq. (24), and the Fig. 1 caption. Eq. (27) already does this, which highlights the inconsistency.","section":"§III, Eq. (24) and Fig. 1"},{"comment":"The signal self-overlap expression in Eq. (A7) appears to have the wrong trigonometric functions: cross terms involving the saddle images 3 and 4 and image 1 should contain sin(2πf Δt) rather than cos(2πf Δt), as in Eq. (A11). Since the extreme-GO approximation drops all oscillatory terms, the final overlap estimates are unaffected, but the numerical implementation of the mismatch should be checked.","section":"Appendix A, Eq. (A7)"},{"comment":"The noise power spectral density Sn(f) is never specified. Please name the exact PSD and detector configuration used in the pycbc filter package, since the absolute mismatch values depend on it.","section":"§VII, Eq. (33)"},{"comment":"The caption refers to 'eccentricity e' while the text defines e as ellipticity; please use a single terminology throughout.","section":"§IV, Fig. 3 caption"},{"comment":"References [57] and [58] are the same paper by R. Takahashi (2004) and should be merged.","section":"References [57] and [58]"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is close to publishable but the two load-bearing concerns—the unvalidated extension of the quasi-GO criterion to the SIE and the idealized mismatch comparison—require substantive revision. There is no sign of duplicate publication; the relationship to the authors' prior work [34] is clearly stated."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe short version: this is a worthwhile paper for the GW lensing community. It takes Takahashi's quasi-GO expansion, applies it to the SIE lens, produces a mass threshold for GO validity, and shows that three-image templates beat two-image templates by factors of 1.5 to 5 in waveform mismatch. The three-image template result is new and practically relevant.\n\nWhat is genuinely good: the extreme-GO overlap derivation in Appendix A is a clean, parameter-free calculation that explains the numerical ordering (3A < 3B < 2L for most source positions). The 14-class classification is thorough, and the paper uses glafic for image properties and cites the relevant literature. No citation red flags.\n\nNow the soft spots, in order.\n\nFirst, the GO-validity threshold. The criterion is w > max_j |Delta^(j)| from Takahashi's first-order expansion in 1/w. That expansion is checked against exact wave optics only for the SIS, which has no fold or cusp. For the SIE, near the tangential caustic the Hessian eigenvalue goes to zero, so the expansion is not uniform. The paper acknowledges that computing wave-optics F for the SIE is hard, but the abstract's \"10^5 M_sun\" claim inherits an unverified assumption. This is a real caveat, not a fatal flaw. The authors should state the assumption more prominently and ideally test one SIE configuration against a numerical wave-optics calculation.\n\nSecond, the mismatch analysis is idealized. The templates are built from the true image properties of the signal, so the reported mismatches are a best-case scenario, not what a search achieves when parameters are unknown. The noise PSD is never specified, so Table II is not reproducible as written. Only one ellipticity (e=0.2) and one source are explored. These are fixable with additional runs or explicit caveats.\n\nThird, a minor inconsistency: Sec III and Fig 1 write the condition as max_j Delta^(j) rather than max_j |Delta^(j)|. Eq (27) correctly uses absolute values; the caption could mislead.\n\nBottom line: the paper is a solid, useful extension, and the three-image template result is worth having on record. It deserves a serious referee and likely publication after the authors clarify the quasi-GO assumption and make the mismatch study reproducible. I would send it to peer review, not desk reject.","headline":"Solid, useful extension of quasi-GO to SIE lensing; the 10^5 M_sun threshold is plausible but rests on an expansion validated only for SIS, and the mismatch study is idealized.","tokens_in":25006,"tokens_out":6702,"would_cite":true,"duration_ms":67766,"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":"Elliptical galaxy lenses can split gravitational waves into four images, and three-image templates match those signals far better than two-image templates.","keywords":["gravitational lensing","gravitational waves","singular isothermal ellipsoid","geometrical optics","wave optics","waveform mismatch","template banks","microlensing"],"falsifier":"Numerically evaluate the full diffraction integral for the SIE lens at source positions near the fold, cusp, and Einstein cross for frequencies around $w \\approx \\Delta^{(j)}$, and compare the resulting amplification factors with the quasi-GO prediction; if the fractional error $|F_{\\rm GO}/F_{\\rm WO}-1|$ stays below unity at masses below $10^5\\,M_\\odot$, or exceeds unity above it, the threshold in Eq. (27) is falsified.","tokens_in":1884,"feed_emoji":"🌌","tokens_out":2005,"duration_ms":94587,"temperature":0.7,"pith_summary":"This paper asks when the simple geometrical-optics picture of gravitational-wave lensing remains valid for elliptical galaxy lenses and how many images a search template needs. Using the singular isothermal ellipsoid (SIE) lens model, an elliptical generalization of the isothermal sphere that can produce up to four images, the authors show that the geometrical-optics approximation holds above roughly $10^5\\,M_\\odot$ for ground-based detector frequencies, with wave-optics effects becoming significant at lower masses or near caustics. They then show that templates built from three of the four images reproduce the lensed waveform much better than the standard two-image templates, reducing mismatch from $O(10^{-1})$ to $O(10^{-2})$ by factors between 1.5 and 5. If correct, this provides a quantitative rule for when discrete-image templates are valid and argues that future lensed-GW searches should include three-image template families.","feed_headline":"Three-image templates cut mismatch in lensed-GW searches","feed_subtitle":"For elliptical lens potentials, adding a third image cuts waveform mismatch by factors of 1.5 to 5.","key_machinery":"The central object is the first-order quasi-geometrical-optics expansion of the lensing amplification factor, Eq. (16) from reference [58], in which each image carries a correction $i\\Delta^{(j)}/w$; the real coefficient $\\Delta^{(j)}$ (Eq. 17), built from third and fourth derivatives of the Fermat potential at the image, measures how soon diffraction erodes the stationary-point approximation. Setting $\\Delta^{(j)}/w \\sim 1$ turns into the per-image minimum lens mass $M_{Lz,\\min}^{(j)} = |\\Delta^{(j)}|/(8\\pi f_{\\min})$ (Eq. 27). The second part of the machinery is the template comparison: the 2L, 3A, and 3B amplification factors of Appendix A, whose non-oscillatory overlaps predict the ordering $\\epsilon_{3A} < \\epsilon_{3B} < \\epsilon_{2L}$ when the fourth-image flux ratio $I_4 < 1$.","core_discovery":"For the SIE lens, the paper establishes a per-image breakdown criterion from a first-order quasi-geometrical-optics expansion of the diffraction integral: image $j$ remains in the geometrical-optics regime when the dimensionless frequency $w=8\\pi M_{Lz}f$ exceeds the coefficient $\\Delta^{(j)}$, so the whole lens is in the geometrical-optics regime when $M_{Lz} > \\max_j |\\Delta^{(j)}|/(8\\pi f_{\\min})$ (Eq. 27). For ground-based detectors with $f_{\\min}=20\\,\\mathrm{Hz}$, this yields a threshold of roughly $10^5\\,M_\\odot$, with higher masses required near caustics. In the four-image region, the paper classifies events into 14 microlensing and macrolensing classes depending on which inter-image time delays exceed the in-band event duration. Waveform mismatch analysis shows that three-image templates, type 3A (two minima and one saddle) and type 3B (one minimum and two saddles), outperform two-image (2L) templates by factors of 1.5 to 5, with 3A best near the major cusp, 3B best near the minor cusp, and, for lens masses above $10^7\\,M_\\odot$, diffraction effects are negligible and mismatches drop by an additional factor of roughly three.","pith_inferences":["The same per-image $\\Delta^{(j)}$ criterion could be applied to other non-axisymmetric lens models, such as SIE potentials with external shear or power-law ellipsoids, to test whether the $10^5\\,M_\\odot$ threshold is robust.","If three-image templates are folded into search pipelines, the detection volume for lensed events inside the tangential caustic should grow; this could be tested with injected lensed signals in realistic noise before any real event is found.","For future space-based detectors with lower frequency floors, the minimum lens mass scales as $1/f_{\\min}$, so wave-optics corrections become relevant for more massive lenses than for the LVK band, and the paper's class taxonomy maps directly onto that regime.","If most galaxy-scale lenses are approximately SIE-like, the mass-dependent class fractions in Table I imply that the first catalog of lensed gravitational-wave events could be used to infer the lens mass function from the observed mix of event classes."],"forward_implications":["For lens masses below roughly $10^5\\,M_\\odot$ at ground-based frequencies, or for sources close to caustics, geometrical-optics templates will be inadequate and wave-optics effects must be included.","At lens masses above $10^7\\,M_\\odot$, diffraction effects become negligible for ground-based detectors, and waveform mismatches fall by an additional factor of roughly three.","Using three-image templates reduces waveform mismatch from $O(10^{-1})$ to $O(10^{-2})$, typically by factors of 1.5 to 5 compared with two-image templates.","The relative merit of 3A versus 3B templates is set by the brightness of the fourth image: 3A wins when $I_4<1$, while 3B wins when $I_4>1$.","Four-image lensed events can be sorted into 14 microlensing and macrolensing classes, and the fraction of events in each class shifts dramatically with lens mass, with classes 6 and 7 dominating at $10^7\\,M_\\odot$."],"supporting_citations":[{"why":"Supplies the first-order quasi-geometrical-optics expansion (Eq. 16) and the coefficient $\\Delta^{(j)}$ that defines the breakdown of the geometrical-optics approximation.","marker":"[58]"},{"why":"Introduces the singular isothermal ellipsoid lens model whose two- and four-image behavior the paper analyzes.","marker":"[42–44]"},{"why":"Provides the SIE lensing potential and the numerical code used to compute image positions, magnifications, and Morse indices.","marker":"[44]"},{"why":"Defines the diffraction-integral amplification factor whose high-frequency stationary-point limit is the geometrical-optics approximation being tested.","marker":"[16]"},{"why":"Earlier treatment of SIE lens and image parameters whose relationships the paper follows for time delays and flux ratios.","marker":"[34]"},{"why":"Supplies the post-Newtonian inspiral waveform and the mismatch definition used in the template comparison.","marker":"[61]"},{"why":"Provides the filtering routines used to compute the mismatches between lensed signals and templates.","marker":"[64]"}],"fun_headline_variants":["Lensed GWs: three-image templates cut mismatch by 5x","Elliptical lensing: add third image to slash GW mismatch","Triple-image GW templates beat binary for elliptical lenses","Three images, not two: sharper lensed GW searches"],"cache_read_input_tokens":27008,"weakest_assumption_plain":"The mass threshold rests on trusting the first-order quasi-geometrical-optics correction to mark where the geometrical-optics approximation breaks down, even though that correction is checked against exact wave optics only for the spherical lens and not for the elliptical one.","fun_headline_variants_meta":{"raw":{"variants":["Lensed GWs: three-image templates cut mismatch by 5x","Elliptical lensing: add third image to slash GW mismatch","Triple-image GW templates beat binary for elliptical lenses","Three images, not two: sharper lensed GW searches"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000648,"raw_usage":{"total_tokens":3037,"prompt_tokens":1067,"completion_tokens":1970,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":683,"completion_tokens_details":{"reasoning_tokens":1899}},"tokens_in":683,"tokens_out":1970,"duration_ms":16033,"temperature":1.0,"reasoning_tokens":1899,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T18:43:29.869857+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Numerically evaluate the full diffraction integral for the SIE lens at source positions near the fold, cusp, and Einstein cross for frequencies around $w \\approx \\Delta^{(j)}$, and compare the resulting amplification factors with the quasi-GO prediction; if the fractional error $|F_{\\rm GO}/F_{\\rm WO}-1|$ stays below unity at masses below $10^5\\,M_\\odot$, or exceeds unity above it, the threshold in Eq. (27) is falsified.","supporting_citations":[{"cited_title":"This correction is proportional to the term ∆(j) given by Eq","cited_arxiv_id":null,"evidence_quote":"Supplies the first-order quasi-geometrical-optics expansion (Eq. 16) and the coefficient $\\Delta^{(j)}$ that defines the breakdown of the geometrical-optics approximation."},{"cited_title":"Takahashi and T","cited_arxiv_id":null,"evidence_quote":"Defines the diffraction-integral amplification factor whose high-frequency stationary-point limit is the geometrical-optics approximation being tested."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Earlier treatment of SIE lens and image parameters whose relationships the paper follows for time delays and flux ratios."},{"cited_title":"Matsunaga and K","cited_arxiv_id":null,"evidence_quote":"Supplies the post-Newtonian inspiral waveform and the mismatch definition used in the template comparison."},{"cited_title":"Gravitational Lenses With More Than Four Images: I. Classification of Caustics","cited_arxiv_id":"astro-ph/0002401","evidence_quote":"Provides the filtering routines used to compute the mismatches between lensed signals and templates."}],"review_version":2}