{"id":"37b68df3-142b-488d-9a8d-88da0b6e5fcf","arxiv_id":"2505.22350","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A new circle-average chaos decomposition reduces nodal-volume variance computations to four-Hermite expectations and extends to non-homothetic Gaussian fields on any Riemannian manifold.","lead":"This paper derives a new, simpler Wiener-Ito chaos expansion for the volume of zero sets of Gaussian random fields on arbitrary Riemannian manifolds, handling anisotropic fields that earlier methods could not. The main payoff is an exact variance formula requiring only four Hermite polynomials, plus explicit bounds and a quantitative form of Berry's cancellation.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.2's printed factor 2 is wrong: the proof in §3.3 and Corollary 1.3 both give coefficient Θ(a,b)/s_n, so the central formula as stated is false and must be corrected.","rationale":"Reading the paper in good faith, the method is sound: the product expansion of δ_0(f(x))||∇f(x)|| under the non-degeneracy assumption (1.10) is legitimate, the chi-variable expansion in Theorem 3.4 is correct, and the reduction to four Hermite polynomials in the variance is real. However, the central formula as printed is false. The proof in §3.3 unambiguously yields the coefficient Θ(a,b)/s_n, not 2Θ(a,b)/s_n, and every subsequent consistent formula—Corollary 1.3, Definition 3.9, Corollary 4.2, Proposition 1.15—uses the corrected constant. The factor 2 therefore contradicts the paper's own derivation and its own corollary. This is the most load-bearing issue because Theorem 1.2 is the main result and all quantitative statements inherit its constant. Corollary 2.1 shows a further prefactor discrepancy relative to a direct variance computation from Proposition 1.15, strengthening the need for a careful constant audit. The reader's weakest_assumption (condition (1.10)) is a standard and explicitly stated hypothesis, not a hidden flaw; it does not pose a correctness risk for the theorem as stated. The reader did flag the constant inconsistency in the rationale, so our concern is partially aligned, but the weakest_assumption field does not identify the same issue. The appropriate verdict remains CONDITIONAL: the paper should be accepted only after the factor 2 and related prefactor errors are corrected, since the structural claims are supported by the proof.","tokens_in":37890,"tokens_out":33999,"duration_ms":320088,"concrete_test":"Set q=0 in Theorem 1.2. Since H_0 ≡ 1 and Θ(0,0)=1, the printed formula gives E[L_f(M)] = (2/s_n)∫_M∫_{S(TxM)} ||u||_{g_f} du dx, whereas Corollary 1.3, Eq. (1.17), gives (1/s_n) times the same integral. Recomputing the coefficient in §3.3 as π/s_n · c_χ(2b) · (−1)^a/(2^a a!√(2π)) = Θ(a,b)/s_n settles that the factor 2 is spurious. This single q=0 comparison is sufficient to force the erratum.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central formula (1.15) contains a factor 2 that is not supported by the derivation. In §3.3, the coefficient is computed as the product of the delta expansion coefficient (−1)^a/(2^a a!√(2π)) and the chi-expansion coefficient A(n,2b) = (π/s_n)c_χ(2b), giving \\tilde Θ(a,b) = Θ(a,b)/s_n with Θ(a,b) as defined in (1.16). No factor 2 appears. This is confirmed by Corollary 1.3: the q=0 case of the corrected formula gives E[L_f(M)] = (1/s_n)∫∫||u||_{g_f} du dx, matching (1.17), while the printed factor 2 would give twice that. The same corrected constant is used in Definition 3.9, Corollary 4.2, and Proposition 1.15, so the error is localized to the theorem statement and to statements directly citing it (notably Corollary 2.1, which additionally appears to drop a square of the prefactor from Prop. 1.15 when computing Var(L_ϕ[2]/λ)). Because this factor changes every quantitative output of the claimed decomposition—expectation, variance, and bounds—it is load-bearing even though it is a typo. The structural claim, that the nodal volume measure admits the displayed chaos decomposition with the four-Hermite variance reduction, is well supported by the proof once the constant is corrected.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims an explicit Wiener-Itô chaos decomposition for the nodal volume measure of a C^2, unit-variance Gaussian random field with non-degenerate first jet on a compact Riemannian manifold, possibly with boundary. The main formula, Theorem 1.2, expresses the q-th chaos component as a finite sum of integrals over the unit tangent sphere of products of Hermite polynomials in f(x) and in the normalized derivative, with the Adler-Taylor metric appearing through the factor ||u||_{g_f}. The authors use this representation to reduce variance computations to expectations of products of four Hermite polynomials (Theorem 4.1), to give an exact variance formula (Corollary 4.2), to derive covariance bounds (Theorem 1.10), and to quantify Berry's cancellation for Riemannian random waves (Corollaries 2.1 and 2.2). The derivation is self-contained and uses standard delta and chi-variable expansions together with the diagram formula.","tokens_in":38237,"tokens_out":21858,"duration_ms":198427,"significance":"If the printed constants are corrected, the paper gives a substantial and plausible advance: it provides a coordinate-free chaos expansion valid beyond homothetic fields, reduces the complexity of nodal-volume variance computations in any dimension, and introduces the frequency and eccentricity parameters as deterministic, non-fitted functionals of the field. The structural claim, namely that the nodal volume measure admits the displayed chaos decomposition with a four-Hermite variance formula, is well supported by the proof. The paper also contains no free parameters and the main formulas are falsifiable against known cases such as spheres and Berry's field. However, as printed, several quantitative statements are wrong by constant factors, so the manuscript cannot be accepted without correction.","major_comments":[{"comment":"The coefficient in the central formula (1.15) is wrong by a factor of 2. The computation in §3.3, Eq. (3.15), gives \\tilde Θ(n,a,b) = Θ(a,b)/s_n, and Corollary 1.3, Eqs. (1.17)-(1.18), Definition 3.9, Eq. (3.28), and Proposition 1.15, Eq. (1.44), all use the corrected constant without the factor 2. The erroneous factor 2 is repeated in Corollary 3.8, Eqs. (3.23)-(3.26), and in Definition 3.9, Eq. (3.29). Because this constant enters every quantitative output of the paper—expectation, variance, and bounds—the theorem statement must be corrected and the corrected constant must be propagated through all dependent formulas.","section":"Theorem 1.2, Eq. (1.15); §3.3; Corollary 1.3; Corollary 3.8; Definition 3.9"},{"comment":"The variance formula in Corollary 2.1 drops the square of the prefactor from Proposition 1.15. Using the corrected constant, for a random wave with orthonormal eigenfunctions and σ^2 = #{λ_i ∈ I}, one obtains \\tilde L_ϕ[2]/λ = - s_{n-1}/(2 s_n √n σ^2 λ^2) ∑ γ_i^2(λ^2 - λ_i^2), whose variance is s_{n-1}^2/(2 s_n^2 n σ^4 λ^4) ∑(λ_i^2 - λ^2)^2. The printed Eq. (2.3) instead has s_{n-1}/(2 s_n √n σ^4 λ^4) ∑(λ_i^2 - λ^2)^2, missing the factor s_{n-1}/(s_n √n). The qualitative statement about vanishing of the second chaos is unaffected, but the quantitative constant in Berry's cancellation is wrong.","section":"Corollary 2.1, Eq. (2.3); Proposition 1.15, Eq. (1.44)"},{"comment":"The displayed identity in Eq. (5.6) is false. For q=4, the left-hand side q!(Σ|Θ(a,b)|)^2 equals 25/6, while the right-hand side 2^{-q} binom(q,q/2) (Σ binom(a+b,b)|2b-1|)^2 equals 27/2. Since the proof of Lemma 5.2 needs only an upper bound, the argument can likely be repaired by replacing this identity with a valid estimate, and the lemma itself appears true; however, as printed the proof of this lemma is not correct.","section":"Section 5.1, Eq. (5.6), Lemma 5.2"}],"minor_comments":[{"comment":"The statement of c_χ(2b) has the denominator (b−1)b!, which contradicts Theorem 3.4 and the derivation in Eq. (C.6); the correct denominator appears to be (2b−1)b!.","section":"Appendix C.2, Lemma C.2"},{"comment":"The proof multiplies the distribution-valued expansion of δ_0(f(x)) with the L^2 expansion of ||∇f(x)||; since the δ_0 expansion is not an L^2 expansion, a short regularization or approximation argument would make the product step fully rigorous.","section":"§3.3, proof of Theorem 1.2"},{"comment":"The notation H_{n-1}(dv) for the spherical Hausdorff measure is confusing because H_q is also used for Hermite polynomials; using Vol_{n-1}(dv) or dσ(v) would avoid ambiguity.","section":"Eq. (1.27) and Section D"},{"comment":"The definitions of \\tilde L_ϕ(M){q} in Eq. (1.39) include the factor 2, while Proposition 1.15 is computed without it; the constant audit requested above should also resolve this inconsistency.","section":"Throughout Section 1.4"}],"recommendation":"major_revision","confidential_remarks":"The factor-2 error in the central theorem and the missing prefactor square in Corollary 2.1 are mechanical but pervasive, so I would ask the authors for a complete audit of all constants in Theorems 1.2, 3.4, Corollaries 1.3, 2.1, 3.8, 4.2, Definition 3.9, Proposition 1.15, and Lemma 5.2 before the paper can be accepted."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read it. The core result is good and likely correct after a constant fix. The genuinely new thing is a Wiener-Itô chaos decomposition for nodal volume measures that does not require the field to be homothetic, and it reduces variance computations to expectations of four Hermite polynomials in any dimension. That is a real step beyond the homothetic-only expansions in [19,55], and the quantitative Berry cancellation on general manifolds is a worthwhile payoff. The derivation is self-contained: the circle-average identity is not new, but the systematic use for non-homothetic fields on manifolds with boundary is. No fitted parameters, no circularity; the self-citations are background.\n\nNow the soft spots. As printed, Theorem 1.2 has a factor 2 in front of Θ(a,b)/s_n that the proof does not support. The product of the delta coefficient (−1)^a/(2^a a!√(2π)) and A(n,2b) = (π/s_n)cχ(2b) gives Θ(a,b)/s_n exactly, not 2Θ(a,b)/s_n. Corollary 1.3 confirms this: its q=0 case gives E[L] = (1/s_n)∫∫∥u∥ du dx. So the theorem statement is false as written, and anything relying directly on it—notably Corollary 2.1—inherits the wrong constant. Corollary 2.1 also appears to drop the square of the prefactor from Prop. 1.15 when writing Var(Lϕ[2]/λ). These are calculation errors, not structural ones; the four-Hermite variance formula, the covariance bound, and the eccentricity/frequency framework should all survive a correction. The sphere-measure convention between normalized and unnormalized forms needs a careful pass.\n\nThe non-degenerate first-jet assumption (1.10) is essential: it gives the independence used in the product delta expansion and in the diagram formula. That is a genuine limitation, but it is standard and the paper states it clearly.\n\nBottom line: the paper deserves a serious referee. Send it out, but the referee should insist the constants be fixed and the main theorem restated with the corrected coefficient. I would cite the corrected version.","headline":"A genuinely new chaos expansion for nodal volumes with a wrong constant in the printed main theorem; correction needed before use.","tokens_in":38706,"tokens_out":6221,"would_cite":true,"duration_ms":60783,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60G60","60H07"],"pacs":[],"model":"deepseek-v4-flash","headline":"The nodal volume of any sufficiently regular Gaussian field on a compact Riemannian manifold has an explicit Wiener–Itô chaos decomposition.","keywords":["Gaussian nodal volume","Wiener–Itô chaos expansion","Hermite polynomials","Adler–Taylor metric","Riemannian random waves","second-chaos cancellation","variance bounds","nodal sets"],"falsifier":"For a concrete non-isotropic field, such as the unit-variance normalization of a Riemannian random wave on a flat torus with two spectral intervals, compute the second chaos component by Monte-Carlo projection of the nodal volume and compare it with both the $q=2$ case of Theorem 1.2 and the formula in Corollary 1.3; because the two displayed formulas differ by a factor of $2$, this numerical check would identify the correct normalization.","tokens_in":37735,"feed_emoji":"📐","tokens_out":12565,"duration_ms":110460,"temperature":0.7,"pith_summary":"This paper proves that the volume of the zero set of a smooth Gaussian field on any compact Riemannian manifold (with or without boundary) splits into Wiener–Itô chaos components through an explicit formula, with no assumption that the field is isotropic or otherwise tied to the manifold's geometry. The formula expresses the nodal volume density as a sum of products of two Hermite polynomials—one in the field value, one in the normalized directional derivative—integrated over tangent directions and weighted by the field's own metric. The payoff is computational: the variance of any chaos component becomes an expectation of a product of just four Hermite polynomials in any dimension, replacing the standard expansion's products of $2+2n$ Hermite polynomials. From this the paper obtains an exact variance formula, variance bounds, and a quantitative version of the cancellation of the second chaos, now extended from symmetric settings to all compact manifolds.","feed_headline":"A new chaos formula cuts nodal-volume variance to four Hermite terms","feed_subtitle":"No more multi-Hermite products: variance in any dimension from four Hermite terms.","key_machinery":"The key object is a new chaos expansion of the chi variable $\\|\\xi\\|$, the length of a standard Gaussian vector $\\xi\\in\\mathbb R^n$, written as an integral over the unit sphere: $\\|\\xi\\|=\\sum_{b}A(n,2b)\\int_{S^{n-1}}H_{2b}(\\langle \\xi,v\\rangle)\\,dv$, with $A(n,2b)=\\pi c_\\chi(2b)/s_n$. Multiplying this by the classical expansion of the Dirac delta gives the nodal density $\\delta_0(f(x))\\|\\nabla f(x)\\|$ as a sum of products $H_{2a}(f(x))H_{2b}(\\langle d_xf,u\\rangle/\\|u\\|_{g_f})$, integrated over the tangent-unit sphere with weight $\\|u\\|_{g_f}$. The variance computation then uses a diagram formula for four Hermite polynomials under the condition $C_{12}=C_{34}=0$, which holds exactly because $f(x)$ and $d_x f$ are independent at each point; this is what reduces the problem to four Hermite factors.","core_discovery":"The central discovery is Theorem 1.2: for a $C^2$ Gaussian field $f$ on a compact Riemannian manifold $(M,g)$ of dimension $n$ with unit variance and non-degenerate differential at every point, the $q$-th chaos component of the nodal volume measure $L_f$ vanishes for odd $q$, and for even $q$ it is the measure $$L_f(dx)[q]=\\sum_{\\substack{a,b\\in\\mathbb N\\\\ a+b=q/2}}\\frac{2\\Theta(a,b)}{s_n}\\int_{S(T_xM)}H_{2a}(f(x))H_{2b}\\!\\left(\\frac{\\langle d_xf,u\\rangle}{\\|u\\|_{g_f}}\\right)\\|u\\|_{g_f}\\,du\\,dx,$$ where $\\Theta(a,b)=(-1)^{a+b-1}/(2^{a+b}(2b-1)a!b!)$, $s_n$ is the volume of the unit $n$-sphere, and $\\|u\\|_{g_f}^2=\\mathbb E\\{|d_x f(u)|^2\\}$ is the norm of the field's associated metric. The proof multiplies the chaos expansion of $\\delta_0(f(x))$ by a new spherical-integral expansion of the chi variable $\\|d_x f\\|$, separating the field value from its gradient without choosing an orthonormal basis. As a direct corollary, the second chaos is an explicit quadratic form in $f$ and the normalized gradient, and the variance of each component is a double integral over $M\\times M$ of expectations of four Hermite polynomials depending only on the full covariance of the first jet.","pith_inferences":["If the four-Hermite reduction proves stable under small eccentricity, it should make quantitative central limit theorems for nodal volumes of non-isotropic random waves accessible, since the fourth-chaos bound would control the total variance.","The frequency and eccentricity parameters are defined directly from the covariance function, so they could be estimated in simulations or from spectral data; measuring them for a given manifold would predict whether the homothetic approximation is accurate.","The varifold extension to cosine-transform integrands suggests a concrete numerical test: for a non-isotropic field, compute the chaos components of the nodal intersection with a fixed hypersurface and compare them with the formula that depends only on first-jet covariances."],"forward_implications":["The variance of every even chaos component $L_f(M)[q]$ can be written as an explicit integral involving only $C(x,y)$, the first-jet covariances, and the second mixed derivative $C''_{x,y}(u,v)$; no summation over multi-indices of length $n$ is required.","The variance upper bound $|\\mathbb E\\{L_f(dx)[q]L_f(dy)[q]\\}|\\le 2^q(\\lambda(f,x)\\lambda(f,y)/n)\\|j''_{x,y}C\\|^q_{g_f}dxdy$ gives ready-made control of every chaos component in terms of the field-adapted norm of the jet covariance.","For a homothetic field satisfying an eigenvalue equation, the second chaos has the simple form $-\\lambda s_{n-1}/(2s_n\\sqrt n)(\\|f\\|^2_{L^2}-\\|\\lambda^{-1}\\nabla f\\|^2_{L^2})$, which vanishes for zero-boundary or Neumann eigenfunctions, recovering the second-chaos cancellation.","For weakly monochromatic Riemannian random waves on arbitrary manifolds, the second-chaos variance is bounded by $c_n\\ell^{1-n}\\eta_\\ell(O(\\eta_\\ell^2/\\ell^2)+\\varepsilon(\\phi_\\ell))$, giving a quantitative form of the cancellation beyond the sphere and the flat torus.","The same spherical-integral mechanism extends the chaotic decomposition to non-zero levels $f^{-1}(t)$ and to integrals over the nodal set whose angular dependence is a cosine transform, i.e., to certain random varifold functionals."],"supporting_citations":[{"why":"supplies the coarea representation of the nodal volume measure that the expansion starts from.","marker":"[31]"},{"why":"provides the Wiener–Itô chaos decomposition framework and the orthogonality relations for Hermite polynomials.","marker":"[63]"},{"why":"gives the delta-function chaos expansion and the standard Hermite expansion that the new formula compresses.","marker":"[54]"},{"why":"defines the field-associated metric used as the weight in the formula.","marker":"[1]"},{"why":"provides the Laguerre-polynomial representation of nodal volumes in symmetric settings, which the new expansion connects to via the spherical average.","marker":"[55]"},{"why":"establishes the second-moment finiteness that makes the $L^2$ chaos decomposition convergent.","marker":"[37]"}],"fun_headline_variants":["Four Hermite terms for nodal volume variance in any dimension","New chaos decomposition simplifies nodal volume variance to four terms","Nodal volume chaos: variance from four Hermite products, any manifold","Generalized nodal volume variance via four Hermite expectations","Reduced Hermite count: nodal volume variance from four terms"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything rests on the field having unit variance and a non-degenerate differential at every point; if the derivative degenerates on a positive-measure region, the value and gradient are no longer independent and the product expansion, as well as the four-Hermite variance formula, breaks down.","fun_headline_variants_meta":{"raw":{"variants":["Four Hermite terms for nodal volume variance in any dimension","New chaos decomposition simplifies nodal volume variance to four terms","Nodal volume chaos: variance from four Hermite products, any manifold","Generalized nodal volume variance via four Hermite expectations","Reduced Hermite count: nodal volume variance from four terms"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000882,"raw_usage":{"total_tokens":3893,"prompt_tokens":1110,"completion_tokens":2783,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":726,"completion_tokens_details":{"reasoning_tokens":2714}},"tokens_in":726,"tokens_out":2783,"duration_ms":20700,"temperature":1.0,"reasoning_tokens":2714,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T13:11:53.399320+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a concrete non-isotropic field, such as the unit-variance normalization of a Riemannian random wave on a flat torus with two spectral intervals, compute the second chaos component by Monte-Carlo projection of the nodal volume and compare it with both the $q=2$ case of Theorem 1.2 and the formula in Corollary 1.3; because the two displayed formulas differ by a factor of $2$, this numerical check would identify the correct normalization.","supporting_citations":[{"cited_title":"Federer.Geometric Measure Theory","cited_arxiv_id":null,"evidence_quote":"supplies the coarea representation of the nodal volume measure that the expansion starts from."},{"cited_title":"Nourdin and G","cited_arxiv_id":null,"evidence_quote":"provides the Wiener–Itô chaos decomposition framework and the orthogonality relations for Hermite polynomials."},{"cited_title":"Marinucci, G","cited_arxiv_id":null,"evidence_quote":"gives the delta-function chaos expansion and the standard Hermite expansion that the new formula compresses."},{"cited_title":"Laguerre Expansion for Nodal Volumes and Applications","cited_arxiv_id":"2312.09962","evidence_quote":"provides the Laguerre-polynomial representation of nodal volumes in symmetric settings, which the new expansion connects to via the spherical average."},{"cited_title":"Gass and M","cited_arxiv_id":null,"evidence_quote":"establishes the second-moment finiteness that makes the $L^2$ chaos decomposition convergent."}],"review_version":1}