{"id":"2c6a6ee9-dd99-4ce7-84aa-cb51c7d9f224","arxiv_id":"2507.06865","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For odd hyperelliptic Jacobians, an explicit pure-spinor Kummer embedding and duplication map yield a density-one canonical height lower bound matching Lang-Silverman.","lead":"The authors give explicit formulas, for every genus g, for the Kummer variety of an odd hyperelliptic Jacobian and its duplication map, built from pure spinors. They use these formulas to prove a density-one lower bound for canonical heights of rational points, a statistical form of the Lang-Silverman conjecture.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 5.11 leans on the unpublished reduction-height bound from [22], and its reduction to P notin 2J misses odd-order torsion.","rationale":"The Reader identified the unpublished [22] reduction-height estimate as the weakest point, and I agree that this is the most load-bearing input: without it the density-one canonical-height theorem has no proof. I add a second, statement-level flaw: the proof's reduction to P notin 2J does not cover odd-order torsion, and the theorem as written is false for any curve with nonzero rational torsion. This does not necessarily destroy the intended Lang-Silverman statement for non-torsion points, but it means the theorem and abstract need correction. These considerations reinforce the Reader's CONDITIONAL verdict rather than moving it.","tokens_in":65414,"tokens_out":24630,"duration_ms":298931,"concrete_test":"Independently re-derive [22, Cor. 3.11 and Thm. 4.10] and check that the reduction-height lower bound for all P outside 2J is correct; as a side test, evaluate the stated inequality on g=1, f(x)=x^3-432, P=(12,36), which has order 3 and canonical height 0, demonstrating that the literal theorem needs a torsion exclusion.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The density-one canonical-height theorem is assembled in Section 5.3 from Lemma 5.13, Proposition 5.12, and an imported density-one statement that, for almost all f, every P in J(Q)-2J(Q) satisfies eh(P)>-epsilon log Ht(f), attributed to [22, Cor. 3.11, Thm. 4.10]. That external bound is the only source of the '100 percent' conclusion, but [22] is an unpublished preprint and no proof or independent verification is supplied here; if the reduction-height estimate fails, Theorem 5.11 collapses. Separately, the proof says 'it suffices' to handle P outside 2J(Q). This reduction is invalid for odd-order torsion: such P lies in 2J(Q), and every repeated halving keeps it in 2J(Q), so the quadratic-height argument never reaches the imported class. Since the canonical height vanishes on torsion, the theorem as literally stated is false for any f with nonzero rational torsion; for example, g=1, f(x)=x^3-432, P=(12,36) has 2P=-P, so bh=0 while the claimed lower bound is positive. The theorem must either be restated for non-torsion points or supplemented with a proof that density-one many f have no odd-order rational torsion.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops an explicit theory of the Kummer embedding for odd hyperelliptic Jacobians. For f(x)=x^{2g+1}+c_1x^{2g}+...+c_{2g+1} of nonzero discriminant, the authors construct a morphism Ψ:J→OGr(g,V) and show that composing with the spinor embedding of OGr(g,V) realizes the |2Θ|-linear system, giving a canonical basis for H^0(J,O_J(2Θ)) in terms of a Mumford representation. They further describe the theta group action via explicit matrices, prove the existence of quartic duplication polynomials, and show that the Kummer variety is cut out scheme-theoretically by quadrics and quartics. In the arithmetic part, they compare several height functions and prove a lower bound for the canonical height in terms of the naive height (Theorem 5.4), and state a density-one lower bound for canonical heights (Theorem 5.11) attributed to a combination of the present results with results from the authors' previous preprint [22].","tokens_in":65643,"tokens_out":11013,"duration_ms":107185,"significance":"If the height theorems were fully established, the paper would provide a density-one form of the Lang–Silverman conjecture for odd hyperelliptic Jacobians, a major advance. The explicit spinor/Clifford-algebra construction of the Kummer embedding and the duplication map is a substantial and original contribution: it gives, for all g, a canonical coordinate system on |2Θ|, explicit theta-group matrices, and a practical duplication algorithm, with a worked genus-4 example checked against Cantor's algorithm in Magma. These parts contain no fitted constants and are developed from scratch with careful sign conventions. The paper is therefore significant even aside from the height applications, and the explicit nature of the constructions is a definite strength.","major_comments":[{"comment":"The reduction to points outside 2J(Q) is invalid for torsion points. The proof says 'In view of the quadratic property of the height, it suffices to show that ... for all P ∈ J(Q) − 2J(Q)'; however, for a non-trivial point P of finite order, repeated halving never leaves 2J(Q), and the canonical height vanishes. Concretely, for g=1 and f(x)=x^3−432, the point P=(12,36) satisfies 2P=−P, so P∈2J(Q), and \\widehat h_Θ(P)=0, while the asserted lower bound is (1−ϵ)\\log Ht(f)>0 for small ϵ. Thus Theorem 5.11 as stated fails for any f with non-trivial rational torsion, and the proof provides no estimate showing that such f form a density-zero exceptional set. The theorem must either be restricted to non-torsion points or supplemented with a proof that 100% of f in F(X) have no non-trivial rational torsion.","section":"§5.3, proof of Theorem 5.11"},{"comment":"The proof imports two black-box statements from the authors' unpublished preprint [22]: [22, Corollary 3.11] and [22, Theorem 4.10], which together supply the density-one assertion that for almost all f, every P∈J(Q)−2J(Q) satisfies \\widetilde h(P)>−\\epsilon \\log Ht(f). This external result is the only source of the '100%' conclusion for the points outside 2J(Q); no proof or independent verification is given in the present paper. If the reduction-height estimate in [22] fails, Theorem 5.11 collapses. The paper should either incorporate the necessary argument or state the theorem as conditional on [22].","section":"§5.3, proof of Theorem 5.11"},{"comment":"The abstract claims the result for '100% of monic, degree 2g+1 polynomials f(x)∈Z[x]' with the bound in terms of \\log|\\Delta(f)|, but the body (Theorem 1.4 = Theorem 5.11) is proved only for the subfamily f(x)=x^{2g+1}+c_2x^{2g-1}+\\cdots+c_{2g+1}, i.e., c_1=0, and with the bound in terms of \\log Ht(f). Under the ordering Ht(f)=\\max|c_i|^{1/i} used throughout, the c_1=0 subfamily has density zero among all monic degree 2g+1 polynomials, since c_1 ranges over O(X) values. Moreover, the sentence 'using the lower bound |\\Delta(f)| \\ll Ht(f)^{2g(2g+1)}' is backwards: this is an upper bound on |\\Delta|, whereas passing from a lower bound in terms of \\log Ht to one in terms of \\log|\\Delta| would require a lower bound of the form |\\Delta(f)| \\gg Ht(f)^{2g(2g+1)} on a density-one set, which is not proved. The abstract's headline claim is therefore unsupported by the arguments in the paper.","section":"Abstract and Theorem 1.4"}],"minor_comments":[{"comment":"The displayed Mumford triple for the example reads '(x4 + 4 + x3 + x2 + 2 + x + 3, ...)'; the ordering of terms is confusing and the constant appears to be misplaced. Please rewrite in standard polynomial notation.","section":"§4.3"},{"comment":"Proposition 3.18 is stated with 'We omit the proofs of these statements'; since this proposition is part of the paper's claimed results, it would be better to include at least a sketch or a precise reference for each assertion.","section":"§3.6"},{"comment":"The notation F_δ(X) is introduced after Theorem 5.11 but used in Proposition 5.12; consider moving the definition immediately before its first use to avoid confusion.","section":"§5.3"}],"recommendation":"major_revision","confidential_remarks":"The constructive spinor/duplication part of the paper is a substantial and likely correct contribution, and the explicit genus-4 example is a nice verification. However, the advertised height theorem is not established as stated: the proof of Theorem 5.11 does not handle torsion points, depends on the authors' unpublished preprint [22] for the core density statement, and the abstract overstates the family (c_1=0 versus all monic polynomials) and the height (log Ht versus log|Δ|). I would encourage the editors to invite a revision that fixes the torsion reduction, makes the dependence on [22] explicit, and corrects the abstract; the underlying spinor theory is worth publishing."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, the good news. The spinor construction in Sections 2–4 is the real contribution. For every g, the authors give an explicit Kummer embedding K_f → P^{2^g−1}, with a canonical Q-basis, coordinates in terms of the Mumford representation, explicit theta-group actions, and quartic duplication polynomials. Previous explicit coordinates only existed for g=2,3. The g=4 example over F5 checks out against Magma, and the formulas reduce correctly to the known Cassels–Flynn and g=3 cases. This part is careful, self-contained, and will be cited.\n\nThe height results are more uneven. Theorem 5.1 (h ≥ h†) and Theorem 5.4 (explicit lower bound for the canonical height in terms of the naive height) are clean and appear to be proved in detail. I believe those.\n\nTheorem 5.11, the 'density 1 form of Lang–Silverman', is where things get shaky. Three problems, in increasing order of seriousness.\n\nFirst, the abstract claims 100% of all monic degree 2g+1 polynomials, but the theorem and Section 5.3 only treat the c1 = 0 subfamily. Under the stated Ht ordering, that subfamily has density zero. No transfer argument is supplied. That is an overclaim, and it should be fixed in the abstract or the theorem.\n\nSecond, the proof imports the reduction-height bound from the authors' own unpublished preprint [22] as a black box. If [22, Cor. 3.11 and Thm. 4.10] are wrong, Theorem 5.11 collapses. This is not circular—the Kummer machinery is independent—but it is a serious dependency.\n\nThird, and most importantly, the reduction step in the proof of Theorem 5.11 says it suffices to handle P outside 2J(Q), because the canonical height is quadratic. That is invalid for odd-order torsion. Such P lies in 2J(Q), and every halving keeps it there. Since torsion has zero canonical height and the claimed lower bound is positive for large Ht(f), the theorem as literally stated is false for any f with nontrivial rational torsion—take g=1, f(x)=x^3−432, P=(12,36). The fix is straightforward: either state the theorem for non-torsion points, or show that a density-one family has no odd-order rational torsion. As written, the theorem overreaches and the proof has a genuine gap.\n\nBottom line: the Kummer/spinor part deserves publication, and the paper deserves a serious referee. But the referee should push for a corrected abstract, a proof or honest restriction for the c1=0 family, a resolution of the torsion issue, and ideally an update on [22]. I would not desk-reject, but I would not accept without revision.","headline":"The explicit Kummer embedding for all g is a solid, citeable contribution; the density-one canonical-height theorem is not proved as stated.","tokens_in":66190,"tokens_out":4800,"would_cite":true,"duration_ms":51794,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11G50","14H40","14K25"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper gives an explicit pure-spinor description of the Kummer embedding for every odd hyperelliptic Jacobian, and uses it to prove a density-one lower bound for canonical heights.","keywords":["odd hyperelliptic curves","Kummer varieties","pure spinors","canonical heights","Lang–Silverman conjecture","theta groups","density one"],"falsifier":"For a positive-density sequence of polynomials $f(x)=x^{2g+1}+c_2x^{2g-1}+\\cdots+c_{2g+1}\\in\\mathbb{Z}[x]$ with $Ht(f)\\to\\infty$, construct a rational point $P\\in J_f(\\mathbb{Q})$ given by a Mumford triple and evaluate the paper's duplication polynomials to compute $\\widehat{h}_\\Theta(P)$; if such points have $\\widehat{h}_\\Theta(P)<((3g-1)/2-\\epsilon)\\log Ht(f)$ infinitely often, the density-one theorem is false.","tokens_in":65192,"feed_emoji":"📐","tokens_out":12128,"duration_ms":123454,"temperature":0.7,"pith_summary":"This paper claims that for odd hyperelliptic curves $y^2=f(x)$ with monic $f$ of degree $2g+1$ and nonzero discriminant, the Kummer variety $J_f/\\{\\pm 1\\}$ embeds into $\\mathbb{P}^{2^g-1}$ by an explicit, canonical map built from pure spinors. It proves that $100\\%$ of the integer polynomials in the subfamily $f(x)=x^{2g+1}+c_2x^{2g-1}+\\cdots+c_{2g+1}$ have the property that every non-trivial rational point $P\\in J_f(\\mathbb{Q})$ satisfies $\\widehat{h}_\\Theta(P)\\ge ((3g-1)/2-\\epsilon)\\log Ht(f)$, a density-one form of the Lang–Silverman conjecture; the abstract states the same conclusion for all monic polynomials. The explicit Kummer map supplies a canonical basis of $H^0(J_f,\\mathcal{O}_{J_f}(2\\Theta))$, duplication polynomials for the map $[2]$, and quadric-and-quartic equations for the Kummer image. A reader should care because the description is uniform in $g$ and turns height comparisons on hyperelliptic Jacobians into concrete polynomial algebra in the Mumford triple.","feed_headline":"100% of odd hyperelliptic Jacobians obey a height bound","feed_subtitle":"Pure spinors make the Kummer embedding explicit, turning Lang–Silverman into a density-one theorem for every genus.","key_machinery":"The load-bearing object is the morphism $\\Psi\\colon J_f\\to OGr(g,V)$, where $V=k[x]/(f(x))$ carries the bilinear form $\\psi(a,b)=\\tau(ab)$ and $OGr(g,V)$ is the orthogonal Grassmannian of isotropic $g$-planes. Its composition with the spinor embedding realizes the complete linear system $|2\\Theta|$: points of the Kummer variety correspond to pure spinors, i.e. lines in the spin representation $S$ whose annihilator is a maximal isotropic subspace, and the coordinates are Pfaffians in the coefficients of the Mumford triple. The Clifford group of $V$ supplies matrices for the $\\theta$ group $G(2\\Theta)$, and the duplication map $[2]$ is represented by explicit quartic polynomials; this linear-algebraic package is what allows the naive, reduction, and canonical heights to be compared uniformly in $g$.","core_discovery":"The central discovery is that the $|2\\Theta|$ linear system of an odd hyperelliptic Jacobian is governed by the same quadratic space $V=k[x]/(f(x))$ that carries the Mumford representation: the Kummer embedding $K_f\\to\\mathbb{P}^{2^g-1}$ is the composite of a morphism $\\Psi\\colon J_f\\to OGr(g,V)$ with the pure-spinor embedding $\\Sigma\\colon OGr(g,V)\\to\\mathbb{P}(S)$, and both factors are computable from the triple $(U,V,R)$. Clifford multiplication gives the $\\theta$-group action on $H^0(J_f,\\mathcal{O}_{J_f}(2\\Theta))$, explicit matrices for lifts of $2$-torsion points, and quartic polynomials representing duplication. The main arithmetic consequence is Theorem 1.4: for fixed $g\\ge1$ and $\\epsilon>0$, $100\\%$ of polynomials $f(x)=x^{2g+1}+c_2x^{2g-1}+\\cdots+c_{2g+1}\\in\\mathbb{Z}[x]$ of nonzero discriminant satisfy $\\widehat{h}_\\Theta(P)\\ge((3g-1)/2-\\epsilon)\\log Ht(f)$ for every non-trivial $P\\in J_f(\\mathbb{Q})$; the body proves this for the $c_1=0$ family, and the abstract's all-monic statement is asserted rather than proved.","pith_inferences":["A reader should not count the abstract's all-monic statement as established: the body's Theorem 1.4 is stated only for the $c_1=0$ subfamily, and no argument transferring the density result to all monic polynomials under the $Ht$ ordering appears.","If the imported reduction-height estimate of the earlier preprint is verified, the same method should apply to any height-ordered subfamily of monic polynomials that satisfies an analogous 'not too small, too often' estimate; the all-monic version would then follow.","The Pfaffian coordinates and duplication polynomials should be directly implementable for genus at least $5$, extending the experimental range of explicit Kummer geometry beyond the currently worked cases $g=2,3$.","The rank-at-most-one determinantal description of the Kummer image may be useful for sieve-style counting of rational points, because it replaces transcendental theta-function comparisons by algebraic inequalities in the Mumford coefficients."],"forward_implications":["For the $c_1=0$ family, the naive height $h(P)$ is well-defined and satisfies $h(P)\\ge(g-\\epsilon)\\log Ht(f)$ for every non-trivial rational point (Theorem 1.2).","For every $f$ in the family and every $P\\in J_f(\\mathbb{Q})$, the canonical height satisfies $\\widehat{h}_\\Theta(P)\\ge \\tfrac12 h(P)-\\tfrac{1}{12}g(7g+5)\\log Ht(f)+c(g)$ (Theorem 1.3).","As stated in the abstract, $100\\%$ of all monic degree $2g+1$ polynomials would have $\\widehat{h}_\\Theta(P)\\ge(\\tfrac{3g-1}{4g(2g+1)}-\\epsilon)\\log|\\Delta(f)|$ for every non-trivial $P$.","The Kummer image is cut out scheme-theoretically by quadrics and quartics, with the ideal of quadrics independent of $f$, giving an explicit criterion for whether a point of projective space lies in $K_f(\\mathbb{Q})$ and lifts to $J_f(\\mathbb{Q})$.","Duplication is computed by quartic polynomials, so the canonical height of a rational point can in principle be evaluated by repeatedly applying these polynomials to its Mumford triple."],"supporting_citations":[{"why":"Supplies the density-one reduction-height estimate that Theorem 5.11 imports; without it the density proof does not run.","marker":"[22]"},{"why":"Supplies the squarefree-discriminant density results used to prove that the subfamily $F_\\delta(X)$ has density one.","marker":"[2]"},{"why":"Provides the genus-two explicit Kummer basis and surface equation that the spinor formulas reproduce and extend.","marker":"[7]"},{"why":"Provides the Mumford theta-function formalism and Thomae-type formulas used in the archimedean height comparison.","marker":"[28]"},{"why":"Supplies the duplication-polynomial method for bounding the canonical height that Theorem 1.3 refines for all $g$.","marker":"[35]"},{"why":"Identifies $J[2]$ with the centralizer of multiplication-by-$x$ in $SO(V)$, used to lift the theta group into the Clifford group.","marker":"[1]"},{"why":"Gives the Arakelov-theoretic bound $\\widehat{h}\\ge \\tfrac12 h^\\dagger+O(1)$ that this paper makes explicit with $f$-dependent constants.","marker":"[17]"},{"why":"Gives the genus-three explicit height theory that the uniform-$g$ construction generalizes.","marker":"[36]"}],"fun_headline_variants":["Density-one height bound for odd hyperelliptic Jacobians via pure spinors","Pure spinor Kummer map gives explicit density-one height bound","Explicit Kummer embedding from pure spinors yields Lang–Silverman for almost all","Density-one Lang–Silverman for odd hyperelliptic Jacobians via spinors","Spinor-theoretic Kummer map proves density-one height bound for Jacobians"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is an unpublished density-one estimate from the authors' earlier preprint that, for almost all polynomials in the family, every rational point that is not twice another rational point has reduction height at least roughly $\\log Ht(f)$; if that estimate fails, the canonical-height theorem collapses.","fun_headline_variants_meta":{"raw":{"variants":["Density-one height bound for odd hyperelliptic Jacobians via pure spinors","Pure spinor Kummer map gives explicit density-one height bound","Explicit Kummer embedding from pure spinors yields Lang–Silverman for almost all","Density-one Lang–Silverman for odd hyperelliptic Jacobians via spinors","Spinor-theoretic Kummer map proves density-one height bound for Jacobians"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000847,"raw_usage":{"total_tokens":3762,"prompt_tokens":1097,"completion_tokens":2665,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":713,"completion_tokens_details":{"reasoning_tokens":2562}},"tokens_in":713,"tokens_out":2665,"duration_ms":21315,"temperature":1.0,"reasoning_tokens":2562,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T18:53:28.413400+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a positive-density sequence of polynomials $f(x)=x^{2g+1}+c_2x^{2g-1}+\\cdots+c_{2g+1}\\in\\mathbb{Z}[x]$ with $Ht(f)\\to\\infty$, construct a rational point $P\\in J_f(\\mathbb{Q})$ given by a Mumford triple and evaluate the paper's duplication polynomials to compute $\\widehat{h}_\\Theta(P)$; if such points have $\\widehat{h}_\\Theta(P)<((3g-1)/2-\\epsilon)\\log Ht(f)$ infinitely often, the density-one theorem is false.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the genus-two explicit Kummer basis and surface equation that the spinor formulas reproduce and extend."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the Mumford theta-function formalism and Thomae-type formulas used in the archimedean height comparison."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the duplication-polynomial method for bounding the canonical height that Theorem 1.3 refines for all $g$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the Arakelov-theoretic bound $\\widehat{h}\\ge \\tfrac12 h^\\dagger+O(1)$ that this paper makes explicit with $f$-dependent constants."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the genus-three explicit height theory that the uniform-$g$ construction generalizes."}],"review_version":1}