{"id":"43afcb4d-33e0-440a-a4bc-71cc6576f208","arxiv_id":"2611.16328","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Counting circles in bounded Kleinian circle packings with error exponent (delta+1)/2 matching Lax-Phillips orbit counting, with eigenvalues below the continuous spectrum supplying the main terms.","lead":"This paper proves a sharp asymptotic formula for the number of circles of radius at least 1/T in bounded Kleinian circle packings, where eigenvalues of the Laplace operator below the continuous spectrum supply the main terms. The error term matches the best-known Lax-Phillips exponent and improves earlier effective circle counts.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek/deepseek-v4-flash","headline":"The uniform spectral measure bound (Lemma 3.13) is the most delicate step; the construction of the test function g = (2b cos(U·))*(2b cos(U·)) and the appeal to the Kuznecov sum formula are not fully justified for the non-compact, infinite-volume setting of M = Γ\\H^3.","rationale":"The reader identified Lemma 3.13 as the weakest assumption and flagged the LLM-assisted wave equation argument. I agree that Lemma 3.13 is the most delicate step. However, the reader's concern focused on the possibility that the uniform bound might fail or require larger interval mass, and suggested independent verification of the spectral measure bound and residue sign. My stress-test goes deeper: the construction of the spectral measure m_S and the identity linking it to the surface integral of W_g rely on the kernel representation (3.8) for h_g(L) on the non-compact, infinite-volume quotient M. The paper does not justify the convergence of the sum over Γ in (3.8) for the kernel kg(dist(p,q)), which decays only like 1/sinh(dist). For geometrically finite groups with cusps, the sum over Γ of such kernels may not converge absolutely, and the resulting operator may not be bounded on L^2(M). This is a foundational gap that undermines Lemma 3.11 and hence Lemma 3.13. The residue positivity is a secondary issue; even if the residue is positive, the error exponent would still be unsupported if the spectral measure bound fails. Therefore, the verdict should be REJECT unless the kernel convergence is rigorously addressed. I partially agree with the reader because we both identify Lemma 3.13 as critical, but I find a more fundamental obstruction.","tokens_in":17029,"tokens_out":1327,"duration_ms":13514,"concrete_test":"Verify the convergence of the sum over Γ in (3.8) for the kernel kg(dist(p,q)) on M = Γ\\H^3, where Γ is geometrically finite with cusps and infinite volume. Specifically, for a fixed compactly supported g, show that for any u∈C_c^∞(M), the sum over γ∈Γ of kg(dist(p̃, γ q̃)) converges absolutely for almost every (p,q) and defines a function in L^2(M×M) that yields a bounded operator on L^2(M). Alternatively, provide a direct proof that the operator h_g(L) defined by the spectral theorem coincides with the integral operator with kernel Σ_γ kg(dist(p̃, γ q̃)). If this cannot be established, the spectral measure bound Lemma 3.13 is unsubstantiated.","verdict_should_be":"REJECT","load_bearing_attack":"The central claim (Corollary 1.7) rests on the error exponent (δ_Γ+1)/2, which in turn depends on the uniform bound m_S({λ≥1 : |√(λ-1)-U|≤1}) ≪ 1 (Lemma 3.13). This is the non-compact analogue of the Kuznecov sum formula. The proof constructs g = (2b cos(U·))*(2b cos(U·)) and uses Lemma 3.11 and (3.12) to bound the integral of h_g(λ) dm_S(λ) by ∥g∥_∞. However, Lemma 3.11 expresses ∫ h_g dm_S as ∫_S W_g dS, and (3.12) bounds this by area(S)(1+e^{2a})∥g∥_∞ only when supp g ⊂ (-a,a). Here a=1, so the bound is O(1). The critical gap is that the spectral measure m_S is defined via the functional ℓ on H^1(M) and the spectral resolution of L on L^2(M). For the non-compact, infinite-volume manifold M, the spectral theorem for L includes continuous spectrum, and the measure ⟨E_λ f_0, f_0⟩ may have a singular continuous component. The argument that the integral of h_g(λ) against dm_S(λ) equals ∫_S W_g dS relies on the identity W_g = L^{1/2} h_g(L) f_0 (3.10), which itself depends on the functional calculus for L on L^2(M). The step from (3.9) to (3.10) uses that L^{1/2} h_g(L) is bounded and self-adjoint, which holds if h_g is real and bounded on the spectrum. But the spectrum includes [δ_Γ(2-δ_Γ),∞), and h_g(λ) = (b̂(τ-U)+b̂(τ+U))^2 with τ=√(λ-1) is indeed bounded. However, the identity ⟨u, W_g⟩ = ⟨L^{1/2} h_g(L) u, f_0⟩ for all u∈C_c^∞(M) requires that h_g(L) u ∈ H^1(M) and that ℓ can be applied. The paper asserts that h_g(L) u is in C_c^∞(M) by (3.8), but the kernel expression (3.8) is derived for the wave kernel on H^3 and summed over Γ; the finite propagation speed argument on the quotient M requires that the sum over Γ converges in a suitable sense and that the resulting operator on L^2(M) coincides with h_g(L). For non-compact M, the convergence of the sum over Γ of the kernel is delicate, especially because the kernel kg(dist(p,q)) has slow decay (like 1/sinh dist).","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek/deepseek-v4-flash","summary":"The paper claims to prove an asymptotic formula for the number of circles of radius at least T^{-1} in a bounded Kleinian circle packing with symmetry group Γ. The main term is a finite sum over the eigenvalues of the Laplace operator below the continuous spectrum (including the base eigenvalue δ_Γ), and the error term is O(T^{(δ_Γ+1)/2} (log T)^{1/4}), matching the exponent in the Lax–Phillips orbit counting theorem. The proof uses spectral theory: the Dirichlet series Z_P(s) is expressed as an integral of an Eisenstein series over a totally geodesic surface S in M = Γ\\H^3, then meromorphically continued via the resolvent, and bounds on Z_P(s) are obtained from the spectral measure m_S of the surface integral functional. The key technical lemma is a uniform bound (Lemma 3.13) on the mass of m_S in intervals of length 2 in the variable √(λ-1). The paper also provides a mean-square estimate (1.9).","tokens_in":17703,"tokens_out":4205,"duration_ms":29658,"significance":"If correct, the result is a significant advance in the counting of circles in Kleinian circle packings. It gives a Selberg-type formula with the same error exponent as the Lax–Phillips orbit counting theorem, independent of the spectral gap, improving on previous effective results (e.g., Kontorovich–Lutsko, Kelmer–Kontorovich–Lutsko). The method is novel, combining spectral theory of the Laplacian on infinite-volume hyperbolic 3-orbifolds with a surface integral functional and its spectral measure. The paper is well-structured and the main steps are clearly laid out. The author acknowledges the use of LLM assistance for parts of the argument, which is transparent.","major_comments":[{"comment":"The uniform bound m_S({λ≥1 : |√(λ-1)-U|≤1}) ≪ 1 is the load-bearing estimate for the error exponent. The proof constructs g = (2b cos(U·))*(2b cos(U·)) and uses Lemma 3.11 and (3.12) to bound ∫ h_g dm_S by ∥g∥_∞. However, the step from (3.9) to (3.10) and the application of Lemma 3.11 rely on the identity W_g = L^{1/2} h_g(L) f_0, which in turn depends on the functional calculus for L on L^2(M). For the non-compact, infinite-volume manifold M, the spectral theorem includes continuous spectrum, and the measure ⟨E_λ f_0, f_0⟩ may have a singular continuous component. The argument that the integral of h_g(λ) against dm_S(λ) equals ∫_S W_g dS uses Lemma 3.11, which itself depends on (3.10) and the boundedness of L^{1/2} h_g(L). While h_g is bounded and rapidly decreasing, the identity (3.10) requires that h_g(L) u ∈ H^1(M) for u ∈ C_c^∞(M) and that ℓ can be applied. The paper asserts that h_g(L) u is in C_c^∞(M) by (3.8), but the kernel expression (3.8) is derived from the wave equation on H^3 and summed over Γ; the convergence and smoothness on the non-compact quotient M need justification, especially because the sum over Γ may not be locally finite uniformly in the support of u. The paper should discuss the functional calculus on L^2(M) for the operator L with continuous spectrum and justify that h_g(L) maps C_c^∞(M) into H^1(M) (or at least into Dom(L^{1/2})). Without this, the bound (3.12) and consequently Lemma 3.13 are not fully justified.","section":"§3, Lemma 3.13","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The bound (3.12) states |∫ h_g dm_S| ≤ area(S) (1+e^{2a}) ∥g∥_∞ when supp g ⊂ (-a,a). This is derived from Lemma 3.11 and the estimate |W_g| ≤ (1+e^{2a})∥g∥_∞. However, Lemma 3.11 expresses ∫ h_g dm_S as ∫_S W_g dS, and the bound on ∫_S W_g dS uses the pointwise bound on W_g and the finite area of S. But W_g is defined as a sum over planes (3.4), and the pointwise bound |W_g| ≤ (1+e^{2a})∥g∥_∞ is claimed from Lemma 3.1. Lemma 3.1 gives #{C: dist(p,Π_C)<a} ≤ 1+e^{2a}. However, W_g(p) = Σ_C g(ρ_C(p))/cosh ρ_C(p). For |g| ≤ ∥g∥_∞ and |cosh ρ_C(p)| ≥ 1, we have |W_g(p)| ≤ ∥g∥_∞ Σ_C 1_{ρ_C(p)<a}. The sum is exactly the count in Lemma 3.1, so the bound holds pointwise. This seems fine. The issue is the justification that ∫_S W_g dS = ∫ h_g dm_S, which relies on the functional calculus as above.","section":"§3, Lemma 3.11 and (3.12)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The positivity of the residue at s=δ_Γ is proved by constructing an explicit function w and showing ∫ η_{δ_Γ} ϕ_0 dvol < 0. The construction uses the assumption that Γ_∞ is trivial (Remark 1.12). The paper states in Remark 4.5 that if |Γ_∞|>1, the formulas gain a factor |Γ_∞|. The residue computation (4.9) then becomes Ress=δ_Γ Z_P(s) = - (|Γ_∞|/(4π)) (∫_S ϕ_0 dS) (∫_M η_{δ_Γ} ϕ_0 dvol). The sign of the second integral is still negative by the same argument, and ∫_S ϕ_0 dS > 0, so the residue remains positive. However, the construction of w in the proof uses the coordinate |p| and the function χ_0(|p|) which is Γ_∞-invariant only if χ_0 is radial and Γ_∞ acts by rotations. The paper should verify that the construction of w remains valid when Γ_∞ is non-trivial, or state that the torsion-free assumption is retained for the residue positivity and the general case follows by the alternative argument in Remark 4.10 (Landau's theorem). Since Remark 4.10 already gives an alternative proof of positivity (using Parker's identification and Landau's theorem), this is not a critical gap, but the reliance on the torsion-free assumption for the explicit computation should be clarified.","section":"§4, Proof of Theorem 1.5(i), residue positivity","authors_acknowledge":"Remark1.12.To simplify the notation, we assume from now on that Γ is torsion-free and thatP= ΓC 0 for someC 0 ∈ P.","authors_acknowledge_p":0.72},{"comment":"The bound (5.4) uses Lemma 3.13 to assert that the sets |τ-|t|| ≤ 1 and k < |τ-|t|| ≤ k+1 have bounded m_S-mass. Lemma 3.13 gives a uniform bound for intervals of length 2 in τ, i.e., m_S({λ: |τ-U|≤1}) ≪ 1. The argument then sums over k≥1 using that the mass in each annulus is O(1). This is valid if the intervals [U-1, U+1] for different U are disjoint; but the sum over k uses intervals of the form [|t|+k, |t|+k+1] and [|t|-k-1, |t|-k] (for k≥1). These are disjoint and each has length 1, but Lemma 3.13 only gives a bound for intervals of length 2. To cover intervals of length 1, one can note that m_S is a positive measure, so m_S([a,b]) ≤ m_S([a-1,b+1]) if the latter is contained in [1,∞). For intervals near 0, one must also handle λ<1 separately. The paper should address this minor technical point, but it is likely fixable.","section":"§5, Bound (5.4)","authors_acknowledge":null,"authors_acknowledge_p":null}],"minor_comments":[{"comment":"The sentence 'Since ∞ ∈ Ω(Γ), its stabilizer Γ_∞ is finite, and hence trivial under our assumption.' The finiteness of Γ_∞ is a known fact for geometrically finite groups, but the statement 'hence trivial under our assumption' is ambiguous: the assumption is that Γ is torsion-free, so a finite subgroup is trivial. This is correct, but the phrasing could be clearer.","section":"§1, Remark 1.12","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The kernel expression (3.8) is given for u ∈ C_c^∞(M). The paper should note that the sum over γ∈Γ is locally finite because the support of u is compact and the distance function is proper, so the kernel kg(dist(·, γ·)) has finite support in γ for fixed p in the support of u. This is standard but should be mentioned for completeness.","section":"§3, Equation (3.8)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The proof of Lemma 4.3 uses the wave formula for the resolvent and the positivity of the kernel for real s. The statement 'the integral converges in operator norm' is justified by the bound on sin(τ√(L-1))/√(L-1). However, the bound on L^2(M) uses the spectral theorem and the fact that the spectrum starts at δ_Γ(2-δ_Γ)>0, so the operator norm of sin(τ√(L-1))/√(L-1) is at most sinh(τ√(δ_Γ(2-δ_Γ)-1))/√(δ_Γ(2-δ_Γ)-1) for τ>0? Actually, for λ≥δ_Γ(2-δ_Γ), |sin(τ√(λ-1))/√(λ-1)| ≤ sinh(τ√(λ-1))/√(λ-1) ≤ e^{τ√(λ-1)}/√(λ-1). This grows exponentially in τ, so the integral representation for the resolvent as ∫_0^∞ e^{-(s-1)τ} sin(τ√(L-1))/√(L-1) dτ converges in operator norm only if Re(s-1) > sup_{λ∈spec(L)} √(λ-1). But the spectrum is unbounded, so this representation is not valid as an operator norm convergent integral. The standard representation for the resolvent on a complete manifold is via the heat kernel or the wave kernel with a different contour. The paper should provide a correct justification or cite a reference for the resolvent kernel formula. This is a critical point because Lemma 4.3 is used to meromorphically continue Z_P(s). If the resolvent representation is not justified, the continuation argument may be incomplete.","section":"§4, Lemma 4.3","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The claim that 'the integral converges in operator norm' for the representation (L-λ_s)^{-1} = ∫_0^∞ e^{-(s-1)τ} sin(τ√(L-1))/√(L-1) dτ is questionable. The operator sin(τ√(L-1))/√(L-1) is bounded but its norm grows like e^{τ√(Λ)} for large Λ if the spectrum is unbounded. The integral converges only if Re(s-1) is larger than the growth rate, which is not uniform. The standard approach is to use the wave kernel with a complex contour or to use the heat kernel. The paper should either correct this or provide a reference where this representation is justified for the Laplacian on a complete non-compact manifold with continuous spectrum. This is a load-bearing issue for the meromorphic continuation.","section":"§4, Lemma 4.3","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The paper states 'The same bound, with powers of |log r|, applies to its derivatives with respect to s.' This is vague; the bound on η_s and its derivatives should be made explicit if needed for the holomorphy of η_s as an L^2-valued function. It is likely fine, but the mention of |log r| is not explained.","section":"§4, Lemma 4.3","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The inequality |ℓ(h_s(L)η_s)|^2 ≤ ∥η_s∥_2^2 ∫_0^∞ |h_s(λ)|^2 dm_S(λ) uses the Cauchy-Schwarz inequality and the representation ℓ(h_s(L)η_s) = ⟨η_s, L^{1/2} h_{ar{s}}(L) f_0⟩. This requires that L^{1/2} h_{ar{s}}(L) f_0 ∈ L^2(M). Since h_s is bounded and decays, and f_0 ∈ L^2, this holds. But the step also uses that ℓ(u) = ⟨L^{1/2} u, f_0⟩ for u ∈ H^1(M). For u = h_s(L)η_s, we need h_s(L)η_s ∈ H^1(M). Since η_s ∈ L^2 and h_s(L) is bounded, h_s(L)η_s ∈ L^2, but to be in H^1 we need L^{1/2} h_s(L)η_s ∈ L^2. This is true if h_s(λ) decays sufficiently, e.g., |h_s(λ)| ≪ (1+λ)^{-1/2} for large λ. The paper should verify this decay for s in the relevant region. For s=1+ε+it, |h_s(λ)| = 1/|λ-λ_s|, which decays like 1/λ for large λ, so L^{1/2} h_s(L) is bounded. This is fine.","section":"§5, Equation (5.2)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The contour shift argument uses the pointwise bound |Z_P(1+ε+it)| ≪ |t|/ε for |t|≥1. This is derived in Theorem 1.5(ii). However, the bound for |t| small (|t|<1) is not given; the contour shift requires integrability on the line Re(s)=1+ε. The paper should address the behavior near t=0, perhaps using the fact that Z_P(s) has at most simple poles at s∈E and is otherwise holomorphic, so the integral over a bounded interval is finite. This is a minor omission.","section":"Appendix A","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The paper uses the notation ≪_{P,Γ} to indicate dependence on the packing and the group. In many places, the implied constants are not explicitly tracked. For the main results, this is acceptable, but for the bounds in Theorem 1.5(ii) and Lemma 3.13, the uniformity in U and t is important. The paper should state that the constants are independent of U and t (as claimed).","section":"General","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The construction of b with non-negative Fourier transform and bb(τ) ≥ 1 on [-1,1] is standard. The paper should note that such a function exists (e.g., by taking b = φ * φ where φ is a smooth even function supported in (-1/4,1/4) with ∫ φ = 1, then bb = (φ̂)^2 ≥ 0 and bb(0)=1, and by continuity bb(τ) ≥ 1/2 on [-1,1] after scaling). The statement 'at least one on [-1,1]' is fine.","section":"§3, Lemma 3.13","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The definition of η_s uses the Euclidean Laplacian Δ_Eucl. The calculation of (L-λ_s)(χ ht^s) should be justified. The paper states 'a direct calculation gives' but does not show it. This is acceptable for a research paper, but the reader might appreciate a brief derivation or a reference.","section":"§4, Equation (4.1)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The remark says 'the final expression in (4.7) and the formulas below for Z_P and its residues gain a factor |Γ_∞|.' This should be checked: if χ is Γ_∞-invariant, then the extension of χ ht^s to M is |Γ_∞| times the function on the quotient? Actually, the sum over Γ in the Eisenstein series already includes Γ_∞, so the relation between E(p,s) and the resolvent may need adjustment. The paper should provide more detail or cite a reference.","section":"§4, Remark 4.5","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The inequality |s-1|^2 |h_s(λ)|^2 ≤ 1/(ε^2+(τ-|t|)^2) is derived assuming λ≥1. For λ<1, the bound is different, but the contribution from eigenvalues below 1 is handled separately. This is fine.","section":"§5, Bound (5.3)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The bound |I(T)| ≪ T^{1+ε} U ε^{-1/2} is derived from the mean-square bound. The paper should explain the steps more clearly: using Cauchy-Schwarz and the L^2 bound on Z_P Ψ_U. The dependence on ε and U is plausible but should be checked.","section":"Appendix A, Equation (A.2)","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"major_revision"},"author_rebuttal":null,"desk_editor":{"model":"deepseek/deepseek-v4-flash","letter":"Sam,\n\nThis paper does something real. It improves the effective circle count for every bounded Kleinian circle packing from the prior exponent (3δ+2)/5 to (δ+1)/2, matching the Lax-Phillips orbit-counting error and removing dependence on the spectral gap. The main idea—express the Dirichlet series as an integral of an Eisenstein series over a totally geodesic surface, then use the spectral theorem to meromorphically continue it and bound it via a spectral measure—is clean and the execution is mostly careful. The residue positivity at δ is handled with a concrete test function, and the contour-shift in the appendix follows a known template. The paper is self-contained enough that a specialist could check the core computations.\n\nThe soft spot is exactly where the stress-test lands: Lemma 3.13, the uniform bound on the spectral measure m_S in intervals of length 2 in √(λ-1). This is the non-compact, infinite-volume analogue of the Kuznecov sum formula, and the proof constructs a test function g = (2b cos(U·))*(2b cos(U·)) and appeals to Lemma 3.11 and (3.12). The gap is that the identity linking the integral of h_g against dm_S to the surface integral of W_g relies on the functional calculus for L on L²(M) and on the convergence of the kernel sum over Γ in (3.8). For a non-compact quotient with continuous spectrum, the step from (3.9) to (3.10) and the claim that h_g(L)u ∈ C_c^∞(M) need more justification than the paper gives. The author discloses LLM assistance for exactly these sections, which is honest but means a referee should verify that the spectral measure bound holds as stated. If it does, the rest of the argument is solid. If it fails or requires a larger constant, the error exponent degrades.\n\nMinor: the assumption that Γ is torsion-free and Γ_∞ trivial is removed in a remark, but the handling of orbifold points in the spectral measure argument is not fully spelled out. The paper acknowledges this and says the results hold, but a referee might want a sentence on why the spectral measure bound is unaffected by finite stabilizers.\n\nWho this is for: spectral geometers and people working on circle/sphere packing counting. It is a genuine advance within a narrow but active subfield. The paper deserves a serious referee—the central claim is important enough and the proof is coherent enough that a desk reject would be a mistake. I would send it to a specialist who can check Lemma 3.13 and the wave-kernel convergence.\n\nRecommendation: accept after the spectral measure bound is independently verified. The LLM disclosure is a reason for extra scrutiny, not for rejection.","headline":"Sharp circle count for all bounded Kleinian packings, reaching the Lax-Phillips error exponent. The proof is coherent and the main advance is real. The load-bearing step is a spectral measure bound (Lemma 3.13) that the author partly produced with LLM assistance; that part needs independent verification.","tokens_in":18359,"tokens_out":705,"duration_ms":7832},"pith_extraction":{"msc":["11F72","11K38","30F40","58J50"],"pacs":[],"model":"deepseek/deepseek-v4-flash","headline":"A new counting formula shows that the eigenvalues of the Laplace operator below the continuous spectrum govern the number of circles down to a sharp error term, matching the Lax–Phillips orbit-counting exponent.","keywords":["Kleinian circle packings","circle counting","Laplace eigenvalues","Lax–Phillips theorem","Eisenstein series","spectral measure","geometrically finite hyperbolic 3-manifolds","Kuznecov sum formula"],"falsifier":"Compute, for a concrete bounded Kleinian circle packing (e.g., an Apollonian packing whose full symmetry group has no eigenvalues below 1), the difference N_P(T) − c $T^{{δΓ}}$ for T up to, say, 10⁶, using exact enumeration of circles; if the difference grows faster than C $T^{{(δΓ+1)/2}}$(log T)^{1/4} for a fixed constant C, the spectral measure bound of Lemma 3.13 would be refuted.","tokens_in":18636,"feed_emoji":"⭕","tokens_out":2190,"duration_ms":16048,"temperature":0.7,"pith_summary":"This paper proves an asymptotic formula for N_P(T), the number of circles of radius at least $T^{{-1}}$ in a bounded Kleinian circle packing with symmetry group Γ. The main terms are indexed by the finitely many eigenvalues of the Laplace operator on Γ\\H³ below the continuous spectrum, and the error term has exponent (δΓ+1)/2, the same as in the Lax–Phillips counting theorem for Γ-orbits. The result is new because it gives a Selberg-type spectral formula for circle counts, with an error exponent independent of the size of the spectral gap. Prior effective circle counts carried error exponents that depended on the spectral gap; this removes that dependence.","feed_headline":"Circles counted by spectral eigenvalues with sharp error term","feed_subtitle":"New formula matches Lax–Phillips exponent, independent of the spectral gap, for every bounded Kleinian packing.","key_machinery":"The central object is the Dirichlet series Z_P(s) = Σ_{C∈P} r(C)^s, expressed as Z_P(s) = (s−1)/(2π) ∫_S E(p,s) dS(p), an integral of an Eisenstein series over the totally geodesic surface S in Γ\\H³. The meromorphic continuation is obtained by writing the Eisenstein integral as ℓ((L−λ_s)^{-1} η_s) for a bounded functional ℓ on H¹(M), with η_s an explicit compactly supported error term. The critical analytic step is Lemma 3.13, a non-compact analogue of the Kuznecov sum formula: it bounds the spectral measure m_S in intervals of length 2 in the variable √(λ−1), giving m_S({λ≥1 : |√(λ−1)−U|≤1}) ≪ 1 uniformly in U. This bound, proved via a point-pair kernel construction g = (2b cos(U·))∗(2b cos(U·)) and the wave equation kernel (3.8), yields the mean-square estimates needed for the sharp error exponent.","core_discovery":"The paper establishes that the circle-counting function of every bounded Kleinian circle packing is governed by finitely many spectral parameters. Specifically, as T→∞, N_P(T) = Σ_{s₀∈E} c_{s₀} $T^{{s₀}}$ + O($T^{{(δΓ+1)/2}}$(log T)^{1/4}), where E = {δΓ} ∪ {s_j} is the set of eigenvalues of the Laplace operator below the continuous spectrum. The coefficients c_{s₀} are residues of the Dirichlet series Z_P(s) at these spectral parameters. The error exponent (δΓ+1)/2 equals that of the Lax–Phillips orbit-counting theorem and, unlike previous circle-counting error terms, does not depend on the spectral gap. The proof expresses Z_P(s) as an integral of an Eisenstein series over a totally geodesic surface, continues it meromorphically via the spectral theorem, and controls it on vertical lines using a spectral measure bound.","pith_inferences":["The independence of the error exponent from the spectral gap suggests the error in circle counting is dictated by the continuous spectrum (the radiation field) rather than by the slowest decaying eigenfunction, an intuitive analogue of the Lax–Phillips wave decomposition.","The spectral measure bound on m_S, being uniform in U, likely extends to twisted (character) versions and to weighted sums over packing circles with prescribed tangency properties, opening counting questions for sub-packings.","A testable extension: for a concrete concrete Apollonian packing where the full symmetry group is known to have no extra eigenvalues below 1 (as in Kelmer–Kontorovich–Lutsko), this paper predicts the exponent (δΓ+1)/2; an explicit numerical check on the first several million circles would either confirm or falsify the predicted error rate.","Connecting to lattice-point counting: the same proof scheme, replacing the surface S by a geodesic segment, should yield analogous spectral counting formulas for the lengths of closed geodesics with the Lax–Phillips exponent, a direction the author hints at through the Eisenstein-period formalism."],"forward_implications":["For any bounded Kleinian circle packing, if the symmetry group has no eigenvalues below the continuous spectrum other than the base eigenvalue, the count reduces to N_P(T) = c T^{δΓ} + O(T^{(δΓ+1)/2}(log T)^{1/4}), a power saving over Boyd's T^{δΓ+o(1)}.","The error exponent (δΓ+1)/2 improves existing effective circle counts, which carry exponents like (3δΓ+2)/5 for Apollonian packings.","The spectral measure bound of Lemma 3.13 is a non-compact Kuznecov-type estimate for totally geodesic surfaces, likely applicable to other counting problems in infinite-volume hyperbolic 3-manifolds.","The mean-square estimate (1.9) offers a quantitative statement on the fluctuations of N_P(T) around its spectral main terms, useful for equidistribution-type questions."],"supporting_citations":[{"why":"Lax–Phillips orbit-counting theorem, whose error exponent (δΓ+1)/2 is matched by the paper's Corollary 1.7.","marker":"[19]"},{"why":"Parker's identification of the circle-packing exponent with the critical exponent δΓ, used to identify the abscissa of convergence of Z_P(s).","marker":"[29]"},{"why":"Bunke–Olbrich's meromorphic continuation of Eisenstein series, a starting point for the continuation used here.","marker":"[4]"},{"why":"Guillarmou–Mazzeo's resolvent meromorphic continuation on geometrically finite hyperbolic manifolds, cited for the continuation of the Eisenstein series.","marker":"[12]"},{"why":"Pollicott's transfer-operator proof of the Apollonian circle-counting asymptotic, the method the author says was explained by Magee.","marker":"[31]"},{"why":"Sullivan's positivity result identifying the base eigenvalue δΓ(2−δΓ), used to place the spectrum and to show the residue at δΓ is positive.","marker":"[37]"},{"why":"Strichartz's results on the essential self-adjointness of the Laplacian and the domain of its square root, used to justify the functional-analytic framework.","marker":"[35]"},{"why":"The contour-integral and Mellin-inversion argument from the author's own earlier work is adapted in Appendix A to derive Corollary 1.7 from the bounds on Z_P(s).","marker":"[8]"},{"why":"Zelditch's Kuznecov sum formula, the compact analogue whose non-compact version is Lemma 3.13.","marker":"[39]"}],"fun_headline_variants":["Spectral eigenvalues count circles in Kleinian packings","Circle count formula matches Lax-Phillips error exponent","Every bounded packing yields finite spectral main terms","Sharp circle counting via Laplace eigenvalues and residues","No spectral gap needed for circle count asymptotics"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The whole error-term exponent rests on Lemma 3.13, the uniform bound on the spectral measure of the surface integration functional in √(λ−1)-intervals of length 2: if that bound failed, the error exponent would degrade and the claimed sharpness would drop to a weaker, spectral-gap-dependent rate.","fun_headline_variants_meta":{"raw":{"variants":["Spectral eigenvalues count circles in Kleinian packings","Circle count formula matches Lax-Phillips error exponent","Every bounded packing yields finite spectral main terms","Sharp circle counting via Laplace eigenvalues and residues","No spectral gap needed for circle count asymptotics"]},"model":"deepseek/deepseek-v4-flash","effort":"low","cost_usd":9.4e-05,"raw_usage":{"cost":9.3912e-05,"is_byok":false,"cost_details":{"upstream_inference_cost":9.3912e-05,"upstream_inference_prompt_cost":8.2628e-05,"upstream_inference_completions_cost":1.1284e-05},"total_tokens":970,"prompt_tokens":908,"completion_tokens":62,"prompt_tokens_details":{"audio_tokens":0,"video_tokens":0,"cached_tokens":0,"cache_write_tokens":0},"completion_tokens_details":{"audio_tokens":0,"image_tokens":0,"reasoning_tokens":0}},"tokens_in":908,"tokens_out":62,"cost_source":"openrouter_usage","duration_ms":2737,"image_pages":[],"temperature":1.0,"billed_attempts":1,"reasoning_tokens":0,"transport_provider":"openrouter","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-11-13T06:36:26.792938+00:00","model_set":{"reader":"deepseek/deepseek-v4-flash"},"pipeline_version":"pith-pipeline@v1.4.1-daily-deepseek-or-medium","input_kind":null,"input_chars":null,"input_total_chars":null,"passes_missing":null,"published_standing":null,"published_confidence":null,"objection_audit":null,"run_record":null,"basis":{"kind":"not_recorded","source":null,"chars_given":null,"chars_total":null},"receipt":null,"falsifier":"Compute, for a concrete bounded Kleinian circle packing (e.g., an Apollonian packing whose full symmetry group has no eigenvalues below 1), the difference N_P(T) − c $T^{{δΓ}}$ for T up to, say, 10⁶, using exact enumeration of circles; if the difference grows faster than C $T^{{(δΓ+1)/2}}$(log T)^{1/4} for a fixed constant C, the spectral measure bound of Lemma 3.13 would be refuted.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Lax–Phillips orbit-counting theorem, whose error exponent (δΓ+1)/2 is matched by the paper's Corollary 1.7."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Parker's identification of the circle-packing exponent with the critical exponent δΓ, used to identify the abscissa of convergence of Z_P(s)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Bunke–Olbrich's meromorphic continuation of Eisenstein series, a starting point for the continuation used here."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Guillarmou–Mazzeo's resolvent meromorphic continuation on geometrically finite hyperbolic manifolds, cited for the continuation of the Eisenstein series."},{"cited_title":"Pollicott,Apollonian circle packings, in Fractal Geometry and Stochastics V, Progr","cited_arxiv_id":null,"evidence_quote":"Pollicott's transfer-operator proof of the Apollonian circle-counting asymptotic, the method the author says was explained by Magee."},{"cited_title":"Sullivan,Related aspects of positivity in Riemannian geometry, J","cited_arxiv_id":null,"evidence_quote":"Sullivan's positivity result identifying the base eigenvalue δΓ(2−δΓ), used to place the spectrum and to show the residue at δΓ is positive."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Strichartz's results on the essential self-adjointness of the Laplacian and the domain of its square root, used to justify the functional-analytic framework."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The contour-integral and Mellin-inversion argument from the author's own earlier work is adapted in Appendix A to derive Corollary 1.7 from the bounds on Z_P(s)."},{"cited_title":"Zelditch,Kuznecov sum formulae and Szeg˝ o limit formulae on manifolds, Comm","cited_arxiv_id":null,"evidence_quote":"Zelditch's Kuznecov sum formula, the compact analogue whose non-compact version is Lemma 3.13."}],"lane_accuracy":{"status":"not yet measured","lane":{"model":"deepseek/deepseek-v4-flash","input_kind":null},"url":"https://pith.science/accuracy"},"review_version":1}