{"id":"9302412f-a3cf-4fa4-a79a-08fa274965e5","arxiv_id":"2607.21590","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A polynomial near the Gaussian is realized as an eigenfunction of a real-analytic localization domain, and local maximizers of Gaussian time-frequency concentration are disks.","lead":"This paper builds a first inverse theory for time-frequency localization: given a wave shape close to the Gaussian, it constructs a phase-space region that makes that shape an exact eigenmode. It also proves that local concentration maximizers are disks and that a known stability exponent is optimal.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1's unsymmetrization step uses the false identity P#(z)=conjugate of P(z); with P# defined as coefficientwise conjugation, P#(∂w/π)G equals ∫ P P# e^{πzw}, not ∫ |P|^2 e^{πzw}.","rationale":"The reader's weakest assumption targeted the nested-strip implicit-function estimates and the invertibility of the linearized moment map. My concern is different and more elementary: one displayed identity in the proof of Theorem 1 is algebraically false under the paper's own definition of P#. The reader did note 'notation ambiguities around complex conjugates,' so there is partial overlap, but the reader treated that as an editorial artifact rather than as a load-bearing flaw. In my reading it is the hinge of the unsymmetrization: without the equality P#(∂w/π)G = ∫ |P|^2 e^{πzw}, the constructed symmetrized identity does not imply that G−λP is annihilated by P#(∂w/π), Lemma 3.2 never applies, and Eq. (3.25) is not proved. The proposed check settles the matter by direct computation. If the check confirms, the central advertised inverse construction is unsupported as written, and Theorem 3 loses its main input. I am not asserting the theorems are false; I am asserting that the current proof has a concrete, testable gap that moves the appropriate verdict below CONDITIONAL.","tokens_in":55772,"tokens_out":14969,"duration_ms":145060,"concrete_test":"Take N=1, P(z)=1+εz, and U=D_r with r=r_λ. Evaluate the disputed equality before Eq. (3.25) at w=0. The paper's left side is ∫_{D_r}(1+\\bar ε z)(1+ε z)e^{-π|z|^2}dA = λ + O(ε^3), because ∫ z dA = ∫ \\bar z dA = ∫ z^2 dA = 0. The integral involving |P|^2 is ∫_{D_r}(1+ε z)(1+\\bar ε \\bar z)e^{-π|z|^2}dA = λ + |ε|^2 ∫_{D_r}|z|^2 e^{-π|z|^2}dA. These differ by |ε|^2 times a positive constant. This one computation decides the issue: the equality used before (3.25) is false in the simplest nontrivial case, independent of the strip estimates.","verdict_should_be":"REJECT","load_bearing_attack":"The proof of Theorem 1 fails at the unsymmetrization step because of a false conjugate identity. The paper defines (Eq. 3.22) P#(z) = \\overline{P(\\bar z)} = 1 + Σ \\bar b_k z^k and states (Eq. 3.24) that P#(∂w/π)e^{πzw} = P#(z)e^{πzw} = \\overline{P(z)}e^{πzw}. But \\overline{P(z)} = 1 + Σ \\bar b_k \\bar z^k, so the final equality holds only on the real axis. Consequently, with G(w)=∫_U P(z)e^{-π|z|^2}e^{πzw}dA(z), one obtains P#(∂w/π)G(w) = ∫_U P(z)P#(z)e^{-π|z|^2}e^{πzw}dA(z), not ∫_U |P(z)|^2 e^{-π|z|^2}e^{πzw}dA(z). The whole unsymmetrization depends on identifying these two integrals: the symmetrized identity (3.27) and P#(∂w/π)(λP)=λQ would imply P#(∂w/π)(G−λP)=0 only if P#(∂w/π)G equaled the left side of (3.27). It does not. Lemma 3.2 therefore never applies, and (3.25) is unsupported. Theorem 3, which is built directly on Theorem 1, inherits the gap. This is not a notational subtlety: the displayed equality is algebraically false for non-real z.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes an inverse theory for Gaussian time–frequency localization operators: for a prescribed near-Gaussian polynomial f0, it claims to construct a real-analytic domain U such that f0 is an eigenfunction (Theorem 1); it then uses this construction to prove optimality of the exponent 1/2 in the quantitative stability inequality of [22] (Theorem 3), to give two new proofs of the Abreu–Dörfler rigidity theorem (Theorem 2), and to prove that local maximizers of the Gaussian Faber–Krahn functional are disks, yielding a new proof of the Nicola–Tilli theorem (Theorems 4 and 5). The technical core is a Bargmann–Fock moment identity, a disk inverse theorem for the symmetrized problem, and an 'unsymmetrization' step using a coefficientwise-conjugate differential operator.","tokens_in":56160,"tokens_out":18602,"duration_ms":168983,"significance":"If Theorem 1 were valid, it would be the first general inverse construction of localization domains from prescribed eigenfunctions, and it would provide a bridge between inverse potential theory, quadrature domains, and time-frequency analysis. The local-maximizer argument in Section 6, which shows that the derivative of a first eigenfunction is again an eigenfunction and then forces the domain to be a disk, is structurally interesting and potentially important. The paper is also transparent about its external inputs and uses standard tools without circularity. However, the central inverse theorem rests on an algebraically false conjugation identity in the unsymmetrization step. That error is load-bearing: it invalidates Theorem 1 and, through it, Theorem 3. The remaining parts (Theorems 2, 4, 5) may be independent, but the paper's main advertised contribution is not established.","major_comments":[{"comment":"The unsymmetrization step contains a false identity. Equation (3.22) defines P#(z) = 1 + Σ \\bar b_k z^k, i.e. P#(z) = \\overline{P(\\bar z)}. This is not equal to \\overline{P(z)} = 1 + Σ \\bar b_k \\bar z^k for non-real z. Consequently, the displayed computation (3.24), P#(∂w/π)e^{πzw} = P#(z)e^{πzw} = \\overline{P(z)}e^{πzw}, is false: the multiplier is P#(z), not \\overline{P(z)}. Applying P#(∂w/π) to G(w)=∫_U P(z)e^{-π|z|^2}e^{πzw}dA(z) therefore gives ∫_U P(z)P#(z)e^{-π|z|^2}e^{πzw}dA(z), not ∫_U |P(z)|^2 e^{-π|z|^2}e^{πzw}dA(z). The identification of P#(∂w/π)G with the left side of the symmetrized identity (3.27) is thus unjustified, and the conclusion H=G-λP satisfies P#(∂w/π)H=0 does not follow. Lemma 3.2 is never applied, and (3.25) is unsupported. This is not a notational subtlety; the equality fails on every non-real z, and U is a complex domain. The unsymmetrization mechanism is the","section":"§3, Eqs. (3.22)–(3.24) and Remark 3.2"},{"comment":"Theorem 3 is built directly on Theorem 1. The proof begins with 'By Theorem 1, applied with λ=λ0 and Pε(w)=1+εw²', which yields the domain Uε and the eigenvalue identity (5.17). Since the proof of Theorem 1 is invalidated by the false identity in §3, the existence of the family Uε is not established. The subsequent first-variation computation, the identification of the second-harmonic mode, and the distance/deficit asymptotics all depend on (5.17). Thus the claimed optimality of the exponent 1/2 is unsupported. A repaired proof of Theorem 1, or a direct construction of the family Uε, would be needed before Theorem 3 can be assessed.","section":"§5.2, proof of Theorem 3"},{"comment":"Even independently of the false identity, the manuscript only proves the symmetrized identity (3.3), i.e. ∫_{U_b} |P_b|² e^{-π|z|²}e^{πzw}dA = λ Q_b(w). The desired eigenfunction identity (3.25) is an entirely different weighted quadrature identity with the holomorphic polynomial P, not |P|². The text asserts in Remark 3.2 that the symmetrization (3.25)⇒(3.27) is immediate, but the converse direction—recovering the unsymmetrized identity from the symmetrized one—is precisely the step that fails. This is a separate structural gap: the disk inverse theorem solves a problem about |P_b|², and no valid argument shows that the same domain solves the problem about P_b.","section":"§3, Proposition 3.1 and Remark 3.2"}],"minor_comments":[{"comment":"The abstract contains the ungrammatical phrase 'whose central idea that of is free-boundary problem'; similarly, §1 has duplicated words ('makes ... makes'). These should be corrected.","section":"Abstract and §1.4"},{"comment":"The word 'auxilliary' should be 'auxiliary'.","section":"§4.2"},{"comment":"The proof refers to 'INSERT HERE' for a discussion of the Reynolds transport formula. This is an unresolved placeholder and must be completed before publication.","section":"§5.1, proof of Lemma 5.3"},{"comment":"The Fock reproducing kernel is written as e^{πzw} with no conjugation. In the standard Bargmann–Fock inner product the reproducing kernel is e^{πz\\bar w}. If the paper intentionally uses a different convention, it should be stated explicitly; as written, the moment identities and the Taylor expansion in w mix holomorphic and antiholomorphic variables, which is confusing and potentially inconsistent.","section":"Throughout, e.g. Lemma 2.1(i), Eq. (3.3)"}],"recommendation":"reject","confidential_remarks":"The paper is ambitious and contains some independently interesting material, especially the local-maximizer rigidity argument in Section 6 and the two proofs of the Abreu–Dörfler theorem in Section 4. However, the central advertised result, Theorem 1, is not proved because the unsymmetrization step relies on the false identity P#(z)=\\overline{P(z)}. This is not a typo or a missing estimate; it is an algebraic error at the pivot of the proof. The dependent sharpness result (Theorem 3) therefore also fails. If the author can supply a correct unsymmetrization or a direct proof of Theorem 1, the paper may be worth reconsidering; in its current form it is not suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know three things about this paper. First, the central inverse construction (Theorem 1) is not established: the unsymmetrization step uses a false identity that the stress-test note correctly identifies. Second, the paper contains more than that—Theorems 2 and 4, and the new proof of Nicola–Tilli, are independent, plausible, and worth serious attention. Third, the manuscript is visibly unfinished (literal \"INSERT HERE\", normalization inconsistencies), so reading it requires patience.\n\nThe stressful note is right. The paper defines P#(z) as the coefficientwise conjugate, 1+Σ\\bar b_k z^k, which equals \\overline{P(\\bar z)}. But in (3.24) it claims P#(∂w/π)e^{πzw} = \\overline{P(z)}e^{πzw}. That equality is false away from the real axis: the multiplier is \\overline{P(\\bar z)}, not \\overline{P(z)}. Consequently, applying P#(∂w/π) to G gives ∫ P(z)\\overline{P(\\bar z)}e^{-π|z|^2}e^{πzw}dA(z), not ∫ |P(z)|^2 times the same exponential. The symmetrized identity (3.27) therefore does not imply the unsymmetrized one (3.25), and Lemma 3.2 is never brought to bear. Theorem 3, which builds directly on Theorem 1, inherits this gap. The reader's report was too generous here: the flaw is in the displayed identity, not in a subtle estimate.\n\nWhat the paper does well: the disk inverse step (Proposition 3.1) is explicit and the Wiener-algebra/Fourier-multiplier setup is clean. The two proofs of the Abreu–Dörfler rigidity (Theorem 2) are substantial, and the second proof, avoiding boundary smoothness, is elegant. The local rigidity of maximizers (Theorem 4) and the Fock-space profile decomposition leading to a new proof of Nicola–Tilli are genuinely interesting, and they appear to rest on sound reasoning. Those parts deserve a careful reader.\n\nThe soft spots are the unpolished state (the placeholder, a translation-normalization inconsistency in Theorem 4 Step 5, notation ambiguities around conjugates) and the fact that the flagship theorem is currently unsupported. The passage at (3.22)–(3.25) is not a minor typo; it breaks the proof.\n\nWho is this for? Specialists in time-frequency analysis and inverse spectral problems. It deserves a serious referee: the ideas are original and much of the paper is salvageable and valuable, but a referee should require the unsymmetrization step to be fixed—or the claim narrowed—before publication. I would not cite Theorem 1 or Theorem 3 in their current form, but I would cite the local rigidity and the new proof of Nicola–Tilli if they survive checking.","headline":"The central inverse construction is unproven due to a false conjugate identity in the unsymmetrization step; the rest of the paper has independent value and deserves serious refereeing.","tokens_in":846,"tokens_out":1009,"would_cite":false,"duration_ms":40449,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["42B10","42C15","30H20","35R35","35R30","47B35","49Q10"],"pacs":[],"model":"deepseek-v4-flash","headline":"In Gaussian time–frequency localization, a near-Gaussian eigenfunction determines a nearby real-analytic domain on which it is an eigenfunction, making the localization domain recoverable from prescribed spectral data.","keywords":["time–frequency localization","Bargmann–Fock space","free-boundary problems","inverse spectral problem","quadrature domains","Faber–Krahn inequality","Hermite functions","concentration–compactness"],"falsifier":"Compute the moment system (3.4) numerically for f0 = h0 + ε h_2 with ε = 10^{-3} and λ = 1/2, iterating the implicit-function corrections on fine radial graphs; if the iteration fails to converge to a smooth radial graph (or produces self-intersections), the linearized invertibility in Proposition 3.1 does not extend to an actual solution of the infinite system.","tokens_in":55642,"feed_emoji":"🎯","tokens_out":5158,"duration_ms":47916,"temperature":0.7,"pith_summary":"This paper develops an inverse theory for Gaussian time–frequency localization: instead of asking which waveform a given domain localizes best, it asks which domain a given near-Gaussian waveform localizes. The central result is that for any prescribed polynomial f0 = h0 + Σ a_k h_k with small coefficients and any λ ∈ (0,1), there exists a real-analytic domain U_λ, a small perturbation of a disk, for which f0 is an eigenfunction of the localization operator with eigenvalue λ. This is the first general inverse construction of localization domains, and a sympathetic reader would care because it turns the eigenfunction into recoverable geometric data about the unknown boundary. The same 'eigenfunction-as-geometry' principle yields two proofs that Hermite eigenfunctions force the domain to be a disk, proves that the exponent 1/2 in the quantitative stability inequality cannot be improved, and delivers a new proof that disks are the global maximizers of Gaussian concentration.","feed_headline":"Near-Gaussian eigenfunctions fix their own localization domains","feed_subtitle":"The first inverse construction: prescribing spectral data yields the domain, sharpening disk rigidity and stability.","key_machinery":"The central object is the moment identity ∫_U P(z) e^{π w z} e^{-π|z|^2} dA(z) = λ P(w) (and its basis form (1.2)), which encodes the eigenvalue equation in the Bargmann–Fock space. The free boundary of U is written as an analytic radial graph φ over the Daubechies disk, in a Wiener-type Banach algebra A^R_τ of holomorphic strip functions; the linearized moment map D_φ F(0,0) is a Fourier multiplier diagonal in angular modes, shown to be an isometric isomorphism onto the weighted sequence space Y_τ. An analytic implicit-function theorem on nested analytic strips (with quantitative remainder estimates (3.11)–(3.12)) produces the perturbed domain. Secondary machinery: a Paley–Wiener divisibili","core_discovery":"The paper's central claim is that an eigenfunction can be read as geometric data: near the Gaussian, prescribing the eigenfunction and eigenvalue determines the localization domain. After the Bargmann transform, the eigenvalue equation becomes an infinite weighted moment system against the Fock weight, and the boundary of the unknown domain is parametrized as an analytic radial graph over the Daubechies disk. The key step is that the linearized moment map at the disk is an isometric isomorphism from the Wiener-type space of radial graphs onto a weighted sequence space, so an analytic implicit-function theorem produces a real-analytic domain U_λ for any polynomial f0 = h0 + Σ_{k=1}^N a_k h_k","pith_inferences":["The construction suggests a general principle: for Gaussian STFT localization, small perturbations of the Gaussian are 'flexible' — they can be realized as eigenfunctions on many nearby domains, whereas exact Hermite data are rigid. This flexibility/rigidity dichotomy may extend to other special functions in Fock spaces.","The isometric linearization provides quantitative control of the inverse map (eigenfunction perturbation → boundary perturbation), which could be made explicit to give a stability theorem for the inverse problem itself.","The moment-system formulation is a Gaussian-weighted quadrature-domain structure; extending the finite-moment system to infinite-order quadrature identities might yield new rigidity statements for non-polynomial eigenfunctions.","The local-rigidity-plus-profile-decomposition route to Nicola–Tilli suggests that other phase-space concentration functionals (e.g., wavelet transforms) could be treated by local rigidity alone, without rearrangement inequalities."],"forward_implications":["Prescribed near-Gaussian spectral data determine a real-analytic localization domain, so one can engineer both the waveform and the phase-space region that localizes it.","Hermite eigenfunctions force the domain to be a disk in the simply connected case, so the disk is the only domain consistent with such spectra.","The exponent 1/2 in both stability inequalities is optimal: no larger exponent can hold uniformly.","Local maximizers of the Gaussian Faber–Krahn problem are disks; together with profile decomposition this recovers the global Nicola–Tilli theorem.","The two proofs of the Abreu–Dörfler rigidity include one with no boundary regularity hypothesis, showing the disk conclusion is null-set invariant."],"fun_headline_variants":["Eigenfunctions near Gaussian determine their own domains","First inverse construction of localization domains","Spectral data fixes the boundary of the domain","Near-Gaussian eigenfunctions encode domain geometry","Analytic domains from prescribed eigenfunctions"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The whole inverse construction rests on the claim that the linearized moment map at the Daubechies disk is an isometric isomorphism from the space of analytic radial graphs onto the weighted sequence space Y_τ, together with the nonlinear remainder estimates on nested analytic strips; if the infinite moment system hides compatibility conditions invisible at the linearized level, or if the strip-width loss cannot absorb polynomial growth in the mode index j, Theorem 1 fails ev","fun_headline_variants_meta":{"raw":{"variants":["Eigenfunctions near Gaussian determine their own domains","First inverse construction of localization domains","Spectral data fixes the boundary of the domain","Near-Gaussian eigenfunctions encode domain geometry","Analytic domains from prescribed eigenfunctions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000255,"raw_usage":{"total_tokens":1441,"prompt_tokens":808,"completion_tokens":633,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":552,"completion_tokens_details":{"reasoning_tokens":568}},"tokens_in":552,"tokens_out":633,"duration_ms":6439,"temperature":1.0,"reasoning_tokens":568,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T07:01:12.015872+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the moment system (3.4) numerically for f0 = h0 + ε h_2 with ε = 10^{-3} and λ = 1/2, iterating the implicit-function corrections on fine radial graphs; if the iteration fails to converge to a smooth radial graph (or produces self-intersections), the linearized invertibility in Proposition 3.1 does not extend to an actual solution of the infinite system.","supporting_citations":[],"review_version":1}