{"id":"8f6d32a5-edc6-4b58-84df-2b388a85e51b","arxiv_id":"2506.17092","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"An explicit tower of abelian covers of the genus-2 curve y^2=x^6-1 has Dirac operator spectral gap uniformly bounded below by a positive constant.","lead":"This paper builds an explicit tower of arithmetic hyperbolic surfaces whose spinor Laplacians keep a uniform spectral gap while the genus grows without bound. The construction is a counterexample to the intuition, which is valid for the ordinary Laplacian, that large surfaces cannot keep such a gap.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 5.3 uses the false equality f_*f^*Lχ = Lχ^{⊗2}; the proof needs the norm map. The gap is repairable, so the CONDITIONAL verdict stands.","rationale":"The paper gives an explicit construction of a tower of spin arithmetic hyperbolic surfaces with a uniform spectral gap for the Dirac operator, and the overall strategy is sound. The stress-test identified the same load-bearing concern as the reader: Lemma 5.3 uses a push-pull identity for line bundles that is false under the standard sheaf-pushforward reading. This is indeed the point where the proof is least secure, because the main theorem flows through Theorem 3.1, whose proof requires the dichotomy χ∈{1,χ_2} that only follows from Lemma 5.3. However, the concern is a presentation-level error that is easily repaired by replacing f_* with the norm map N_f, which satisfies N_f(f^*L)≅L^{⊗2} for a degree-2 cover. The same notational ambiguity in Lemma 5.4 (where f'_*f_* should be f'_*f^*) further supports the need for a revision, but it does not affect the correctness of the argument once the intended maps are made explicit. Other potential issues considered were minor: the missing point at infinity in the proof of Lemma 5.4 is covered by the fact that differences of finite points generate Cl^0(E), and the infimum-versus-minimum imprecision in Proposition 2.3 is harmless because the relevant 2-power torsion points are dense. Since the identified gap is real but fixable, the CONDITIONAL verdict remains appropriate.","tokens_in":11875,"tokens_out":38807,"duration_ms":365521,"concrete_test":"Check the norm identity for this cover: for every line bundle L on E, verify N_f(f^*L) ≅ L^{⊗2}, where N_f is the norm map associated to f. Concretely, trivialize L over an open set of E where f is not ramified, pull back to the two sheets of Σ, take the product of the two values, and show it defines a trivialization of L^{⊗2}. Then re-run the proof of Lemma 5.3 with f_* replaced by N_f: if f^*Lχ is trivial, the norm of a trivialization gives a nowhere-vanishing section of Lχ^{⊗2}, forcing χ²=1, and the rest of the lemma follows as written.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The main theorem rests on Theorem 3.1, and the proof of Theorem 3.1 relies on Lemma 5.3 to pass from 'f^*Lχ is 2-torsion in Pic(Σ)' to 'χ is 2-torsion, so χ=1 or χ=χ_2'. The proof of Lemma 5.3 asserts the identity f_*f^*Lχ = Lχ^{⊗2}. For the degree-2 cover f:Σ→E, the sheaf pushforward f_*f^*Lχ is a rank-2 vector bundle on E, not a line bundle, so this equality is false as written. The correct statement is N_f(f^*Lχ) ≅ Lχ^{⊗2}, where N_f is the norm (transfer) map Pic(Σ)→Pic(E); the paper never defines N_f or mentions that f_* is being used in this nonstandard sense. Without this repair the dichotomy χ∈{1,χ_2} lacks justification: a nontrivial kernel character whose square is trivial would be invisible to the argument. The same notational ambiguity appears in Lemma 5.4, where f'_*f_* should read f'_*f^*. These are presentation errors rather than mathematical falsehoods: the norm identity is standard, and the divisorial computation in Lemma 5.4 is correct when the intended pullback–pushforward is used. The claimed spectral gap therefore remains plausible once the norm map is inserted.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs an explicit sequence of closed arithmetic hyperbolic surfaces Σ_n with genus tending to infinity and a uniform positive lower bound for the smallest eigenvalue of the spin Laplacian (equivalently, a spectral gap for the Dirac operator). The construction starts with the genus-2 curve Σ: y²=x⁶−1, equips it with a theta characteristic, and studies a one-parameter family of flat line bundles L_χ obtained by pullback from an elliptic curve E. The main algebraic claim (Theorem 3.1) is that for every character χ in this family, the twisted theta characteristic K_Σ^{1/2}⊗L_χ has no nonzero holomorphic sections. The proof combines a Fourier decomposition of abelian covers (Proposition 2.3), a criterion equating the vanishing of the spectral gap with the existence of holomorphic sections (Lemma 2.6), and a divisorial argument on hyperelliptic curves to rule out such sections. The arithmeticity of Σ is established by showing it is tiled by (2,6,6)-triangles and that the corresponding triangle group is commensurable with an arithmetic lattice. If the arguments are repaired as indicated below, the main theorem follows from the algebraic statement via the continuity and compactness arguments in Section 2.","tokens_in":12053,"tokens_out":26808,"duration_ms":251004,"significance":"The result is significant: it provides the first explicit construction of spin hyperbolic surfaces with genus going to infinity and a uniform spectral gap for the Dirac operator, complementing the coclosed 1-form gap on hyperbolic 3-manifolds of [Abd+24] and contradicting the oracle one might extrapolate from the function Laplacian case. The proof is transparent and largely self-contained, with the main reduction to an explicit algebraic geometry statement that is parameter-free and falsifiable. The paper also gives a clean example of the 'bass note spectrum' having a nonzero limit point for arithmetic spin surfaces. These strengths are substantial even though the current manuscript contains a few local but load-bearing technical gaps.","major_comments":[{"comment":"The proof of Lemma 5.3 asserts the identity f_*f^*L_χ = L_χ^{⊗2}. This is false: for the degree-2 cover f:Σ→E, the sheaf pushforward f_*(f^*L_χ) is a rank-2 vector bundle on E, not a line bundle. The intended statement is N_f(f^*L_χ) ≅ L_χ^{⊗2}, where N_f is the norm (transfer) map Pic(Σ)→Pic(E). Since the norm map is never defined or mentioned, the proof of injectivity of T_E→Pic(Σ) is incomplete as written. This is load-bearing because the proof of Theorem 3.1 uses Lemma 5.3 to conclude from f^*L_χ being 2-torsion that χ is 2-torsion, thereby restricting to χ∈{1,χ₂}. The repair is standard and local, but the manuscript must be corrected.","section":"Section 5, Lemma 5.3"},{"comment":"The statement of Lemma 5.4 reads 'f'_*f_* : Cl⁰(E)→Cl⁰(E′)'. This composition is not well-defined: f_* pushes forward divisors on Σ to divisors on E, so it cannot be applied to a divisor on E. The proof, however, computes f'_*(f^*(a,b)), i.e., it uses the pullback f^*, not f_*. The statement should be corrected to f'_*f^* : Cl⁰(E)→Cl⁰(E′). The subsequent use of Lemma 5.4 in the proof of Theorem 3.1 is consistent with the corrected statement, so this is a typo in the lemma statement, but it must be repaired for the manuscript to be coherent.","section":"Section 5, Lemma 5.4"},{"comment":"The text defines Σ_n = Σ_{T[2n]} and claims that the Σ_n form a tower of covers of Σ₀. This is not correct as written: the subgroups T[2n] of the torus are not nested as n grows (e.g., T[4] is not a subset of T[6]), so the covers are not linearly ordered by inclusion. The intended tower is obtained by taking Σ_n = Σ_{T[2^n]}, since T[2^n] ⊂ T[2^{n+1}]; Proposition 2.3 applies to all n, so the spectral gap argument is unaffected after this change. Please correct the definition and the statement of the tower in Theorem 1.1.","section":"Section 3, definition of Σ_n"}],"minor_comments":[{"comment":"There is a typo in 'By defintion' (should be 'By definition').","section":"Section 2.1"},{"comment":"The word 'speculatory' is nonstandard; consider using 'speculative'.","section":"Section 1.3"},{"comment":"The claim that the proof for finite-dimensional operators in [Kat95] 'works for compact operators in infinite dimension' should be justified by a short argument or a precise citation to Kato's Chapter 4, since the finite-dimensional proof uses characteristic polynomials and does not literally carry over to infinite dimensions.","section":"Section 2.4, Lemma 2.5"},{"comment":"In the computation of Div(f^*L_{χ₂}), the notation f^*[(1,0)-(0,0)] is correct, but the paper should clarify that the preimages under f of (0,0) are (0,i) and (0,-i), which follows from f(x,y)=(x²,xy); this is implicit but not stated.","section":"Section 5, proof of Lemma 5.2"},{"comment":"The phrase 'Let T be the image of TE in \\hat{H1(Σ,Z)} under the pullback map f^*' is written before Lemma 5.3 establishes injectivity; the paper may want to note that T is one-dimensional by the later injectivity, or prove injectivity earlier, to avoid a forward reference that might confuse the reader.","section":"Section 3"}],"recommendation":"major_revision","confidential_remarks":"The paper is well written and the main idea is sound and interesting. The gaps identified in Section 5 (the norm-map issue in Lemma 5.3 and the f^* vs f_* typo in Lemma 5.4) are local and repairable, as is the tower-definition issue in Section 3. I recommend major revision rather than rejection because the central chain of reasoning is correct after these repairs, and none of the issues require new mathematical ideas. The authors should also double-check the notational consistency between Lemmas 5.3 and 5.4 and the proof of Theorem 3.1 after the repairs."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe paper delivers a real result: the first explicit tower of spin hyperbolic surfaces with a uniform spectral gap for the Dirac operator. The proof strategy is clean and mostly sound. It reduces the spectral question to a statement about holomorphic sections of a theta characteristic twisted by characters, then rules out those sections using the explicit genus-2 curve y^2=x^6-1, a degree-2 map to an elliptic curve, and a small class-group computation. The base surface is shown to be arithmetic in Section 6. The comparison with [Ges+23] and with the 3-manifold result [Abd+24] is honest, and the speculative number-field section is clearly marked as such. Nothing in the central logical chain looks wrong to me.\n\nThe weak points are presentation-level, not mathematical. Lemma 5.3 asserts f_* f^* Lχ = Lχ^{⊗2}; as written this is the identity for the norm map on Picard groups, not the sheaf pushforward, which would give a rank-2 bundle. The intended argument is standard and the injectivity claim survives, but the notation has to be fixed. There is also a typo in Lemma 5.4: the map should be f'_* f^*, not f'_* f_*. And Proposition 2.3 says 'min' where the proof really uses an infimum or limit; this is minor.\n\nNone of these issues changes the verdict on the main theorem. Once the norm map is inserted into Lemma 5.3 and the typo in Lemma 5.4 is corrected, the proof goes through. The contradiction in the final step of Theorem 3.1 is genuine.\n\nThis paper deserves a serious referee. The flaws are exactly the kind a referee should catch in a first round, and they are repairable. I would cite it and I'd bring it up in reading group. Recommend: engage with it, and accept after a minor revision.","headline":"A genuinely new explicit construction with a small notation gap in the proof; worth refereeing and publishing after fixes.","tokens_in":12710,"tokens_out":12903,"would_cite":true,"duration_ms":123057,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["58J50","53C27","11F72"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that an explicit arithmetic tower of hyperbolic surfaces has a uniform spectral gap for the Dirac operator, with genus going to infinity.","keywords":["Dirac operator","spin hyperbolic surfaces","spectral gap","theta characteristic","arithmetic surfaces","abelian covers","twisted zeta function","spin structures"],"falsifier":"Compute $H^0(\\Sigma,K_\\Sigma^{1/2}\\otimes L_\\chi)$ for a $2^n$-torsion character $\\chi\\in T$ using the explicit divisors in Section 5; any nonzero section gives $\\lambda^{\\mathrm{spin}}_0(\\Sigma_{T[2^n]})=0$ and refutes Theorem 1.1. A cheaper check is to test the pushforward identity $f_*f^*L_\\chi=L_\\chi^{\\otimes2}$ under the paper's stated definitions on a concrete character; if the identity is false as equality of line bundles, Lemma 5.3 lacks a valid proof.","tokens_in":11560,"feed_emoji":"","tokens_out":18872,"duration_ms":169799,"temperature":0.7,"pith_summary":"This paper proves that a uniform spectral gap for the Dirac operator can occur on an explicit tower of arithmetic hyperbolic surfaces. The main theorem produces a spin surface $\\Sigma$, a tower of covers $\\cdots\\to\\Sigma_2\\to\\Sigma_1\\to\\Sigma_0=\\Sigma$ with genus tending to infinity, and a constant $c>0$ such that $\\lambda^{\\mathrm{spin}}_0(\\Sigma_n)\\ge c$ for every $n$. This matters because large injectivity radius forces the spinor gap to close, and because for the ordinary Laplacian on functions no such uniform gap can persist along growing-volume towers. The proof reduces the spectral statement to an algebraic-geometric vanishing statement: for every character in a chosen subtorus, a certain twisted $\\theta$-characteristic line bundle has no nonzero holomorphic sections.","feed_headline":"An explicit tower of spin surfaces with a uniform spectral gap","feed_subtitle":"One genus-2 curve generates covers whose smallest Dirac eigenvalue stays positive forever.","key_machinery":"The load-bearing mechanism is the character decomposition of the spin Laplacian on abelian covers, combined with the spin-structure/$\\theta$-characteristic correspondence. On a cover with character group $H$, the spinor space decomposes as $\\bigoplus_{\\chi\\in H}L^2(\\Sigma,S_+\\otimes L_\\chi)$, and the spin Laplacian preserves the summands; Proposition 2.3 identifies the bottom eigenvalue of the cover with the minimum, over all characters in the subtorus, of the bottom eigenvalues of the twisted operators, and reduces positivity to the vanishing of holomorphic sections of $S_+\\otimes L_\\chi\\simeq K_\\Sigma^{1/2}\\otimes L_\\chi$. The explicit curves and the double cover make this vanishing check a finite divisor computation; Lemma 4.1, about degree-two effective divisors on a genus-2 curve of the form $y^2=h(x)$, is what turns the hypothetical existence of a section into a contradiction.","core_discovery":"On the paper's own terms, the central discovery is Theorem 1.1: there exist $c>0$, a spin arithmetic hyperbolic surface $\\Sigma$, and a tower of covers with genus going to infinity such that $\\lambda^{\\mathrm{spin}}_0\\ge c$ on every cover. The construction is fully explicit. The base surface is the genus-$2$ curve $\\Sigma:\\ y^2=x^6-1$ with $\\theta$ characteristic $K_\\Sigma^{1/2}$, and the cover structure comes from the degree-$2$ map $f(x,y)=(x^2,xy)$ to the elliptic curve $E:\\ y^2=x^4-x$. Pulling back a one-parameter subtorus $T_E$ of the character variety of $E$ gives a one-dimensional subtorus $T\\subset\\widehat{H_1(\\Sigma,\\mathbb{Z})}$, and the tower is $\\Sigma_{T[2^n]}$. Theorem 3.1 asserts that $K_\\Sigma^{1/2}\\otimes L_\\chi$ has no nonzero holomorphic sections for any $\\chi\\in T$; Proposition 2.3 converts this into a uniform lower bound for the spin Laplacian on all the covers.","pith_inferences":["The same divisor-counting scheme could be tried on other explicit genus-2 curves equipped with a degree-2 map to an elliptic curve; the structure of Lemma 4.1 suggests the dichotomy between the trivial character and one 2-torsion character will persist whenever the relevant fibers lie in different fibers of the two-to-one projection to the $x$-line.","Because Proposition 2.3 is a general Fourier decomposition for abelian covers, the same reduction—uniform gap equivalent to vanishing of holomorphic sections of a fixed line bundle twisted by characters—should apply to other invariant operators and other line bundles, not only the spin Laplacian.","The number-field analogy in Section 1.3 can be read as a concrete prediction: if an analogue of the tower exists, there should be infinitely many quadratic unramified extensions $L/K$ for which the quotient of their zeta functions has a uniform zero-free strip to the right of $1/2$; the paper does not establish this, but the method suggests searching among abelian towers with controlled 2-torsion ","A fully effective version of the proof may be obtainable: compute the $2^n$-torsion points of $T_E$ symbolically, verify the divisor contradictions mechanically, and thereby extract an explicit numerical constant $c$ rather than an existential one."],"forward_implications":["The covers $\\Sigma_{T[2^n]}$ can be written down explicitly from $\\Sigma:y^2=x^6-1$, the double cover $f(x,y)=(x^2,xy)$, and the $2^n$-torsion subgroup of $T$; the construction requires no probabilistic input or search.","Arithmeticity does not obstruct a uniform spinor spectral gap: the base surface is arithmetic, every cover is arithmetic, and the gap $c$ is uniform across the tower.","Reformulated through the associated twisted zeta function, the result gives an infinite family of spin surfaces whose twisted zeta zeros stay uniformly away from the central point $1/2$.","Together with the universal upper bound of Appendix A, the theorem implies that the bass note spectrum of arithmetic spin surfaces has a nonzero limit point, so the values $\\lambda^{\\mathrm{spin}}_0$ accumulate away from $0$ along an explicit sequence."],"supporting_citations":[{"why":"Establishes the bijection between spin structures and theta characteristics, so the spinor bundle's holomorphic half can be treated as the line bundle $K_\\Sigma^{1/2}$.","marker":"[Ati71]"},{"why":"Supplies the perturbation-theory fact used in Lemma 2.5, that the spectrum of a compact operator varies continuously, which is needed to pass from torsion characters to all characters in the subtorus.","marker":"[Kat95]"},{"why":"Provides the negative result for the ordinary Laplace operator on functions along growing-volume towers, the contrast against which the uniform spinor gap is meaningful.","marker":"[Hub78]"}],"fun_headline_variants":["Explicit spin hyperbolic surfaces with uniform spectral gap","Genus-2 curve yields tower with Dirac eigenvalue bounded away from zero","Tower of arithmetic covers keeps spin gap uniform","Uniform Dirac gap from explicit tower of spin covers","Spin spectral gap on infinite tower of hyperbolic surfaces"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that $f_*f^*L_\\chi=L_\\chi^{\\otimes 2}$ for the double cover in Lemma 5.3; this identity holds only if $f_*$ is interpreted as the norm map on line bundles, a convention the paper never states, and with the standard sheaf-theoretic pushforward the left-hand side is a rank-2 vector bundle, not a line bundle.","fun_headline_variants_meta":{"raw":{"variants":["Explicit spin hyperbolic surfaces with uniform spectral gap","Genus-2 curve yields tower with Dirac eigenvalue bounded away from zero","Tower of arithmetic covers keeps spin gap uniform","Uniform Dirac gap from explicit tower of spin covers","Spin spectral gap on infinite tower of hyperbolic surfaces"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000916,"raw_usage":{"total_tokens":3869,"prompt_tokens":822,"completion_tokens":3047,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":438,"completion_tokens_details":{"reasoning_tokens":2972}},"tokens_in":438,"tokens_out":3047,"duration_ms":22467,"temperature":1.0,"reasoning_tokens":2972,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T19:15:38.647085+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute $H^0(\\Sigma,K_\\Sigma^{1/2}\\otimes L_\\chi)$ for a $2^n$-torsion character $\\chi\\in T$ using the explicit divisors in Section 5; any nonzero section gives $\\lambda^{\\mathrm{spin}}_0(\\Sigma_{T[2^n]})=0$ and refutes Theorem 1.1. A cheaper check is to test the pushforward identity $f_*f^*L_\\chi=L_\\chi^{\\otimes2}$ under the paper's stated definitions on a concrete character; if the identity is false as equality of line bundles, Lemma 5.3 lacks a valid proof.","supporting_citations":[],"review_version":1}