{"id":"cab3a156-5591-4d31-8c87-6a5859aa9e7e","arxiv_id":"2607.20397","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Over an algebraically closed field of characteristic 0, the automorphism group of a curve with simple Jacobian must be cyclic of prime-power or two-prime order, a generalized quaternion group, or trivial, and every such group occurs.","lead":"This paper gives the complete list of finite groups that can appear as automorphism groups of curves whose Jacobians are simple, and constructs explicit curves realizing each group. It settles a long-standing question in arithmetic geometry and produces new examples of unlikely intersections in moduli spaces of abelian varieties.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 6.1's quaternion construction rests on unproved degeneration computations (Prop 6.13/Cor 6.14); a wrong blow-up or torsion-specialization formula would break the induction and the existence half of Theorem 1.2.","rationale":"The central theorem splits into a classification half and a construction half. The classification half is supported by standard results (Wolf, BGG93, Serre, Feit-Thompson) and seems internally consistent. The construction half for cyclic groups of order p^n is written out in detail in Section 4, including explicit stable-model computations and torsion-specialization arguments. The construction for generalized quaternion groups in Section 6 is substantially shorter and self-avowedly omits many proofs, with Proposition 6.13 and Corollary 6.14 left by analogy. These are not cosmetic omissions: Theorem 6.1's induction, Lemma 6.16, and the final L→L+1 step all depend on the exact component structure of the degenerate fiber and on the 2-torsion specialization formulas in (6.3). Since the quaternion groups G_{2^n} for n≥2 are exactly the cases not already handled by CLV23, these unverified computations carry the new existence claim. I found no clear error in the written parts, but the reliance on omitted delicate degeneration computations is a real soft spot. This matches the reader's conditional verdict. A focused recomputation for small parameters would settle whether the analogy with Section 4 is faithful. No change to the verdict: the paper should remain CONDITIONAL until the Section 6 computations are expanded or independently verified.","tokens_in":52727,"tokens_out":19758,"duration_ms":164787,"concrete_test":"Independently recompute Proposition 6.13 for the minimal new case n=2, L=1, S={1}: write the family (6.2), perform the blow-up/normalization exactly as in Proposition 4.16, identify the special fiber and its component intersections, and compute the Néron-model specialization of the basis D^±_{α,i}, P_i from (6.3) by constructing the multidegree-zero line-bundle extension on the stable model. Verify the four displayed formulas in Proposition 6.13, especially η(D^-_{α,i}) = (0,0, [(ζ^i_4 a^{-1/4},0)-(0,0)]) and η(P_i) = ([(ζ^i_{8},0)-∞_1],0,0). Then repeat the check for L=2 with S={1} and S={2} to test Corollary 6.14's contradiction. If any formula differs, Theorem 6.1's induction collapses; if all match, the main omitted step is confirmed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The classification half appears coherent, but the constructive half for generalized quaternion groups is under-supported. Section 6 explicitly says \"the techniques and proofs in Sections 6.2 and 6.3 are very similar to those in Sections 4.3 and 4.4, we omit many of the details.\" In particular, Proposition 6.13 -- the admissible degeneration with J_{T',t} ≅ Jac(C1)×Jac(C2)^2 and the explicit specialization of the 2-torsion basis (6.3) -- is not proved. Corollary 6.14 says its proof is identical to Corollary 4.18, and Theorem 6.12 says its proof is conceptually identical to Theorem 4.10. These statements are load-bearing: Theorem 6.1's induction uses Proposition 6.13 to specialize isotypic components, to control the endomorphism algebra of degenerate fibers, and via Corollary 6.14 to rule out nontrivial decompositions of J^al. Since CLV23 only covers n=1, the entire new quaternion case in Theorem 1.2(iii) depends on these omitted computations. The degeneration involves a multistep blow-up, the identification of the special-fiber components and their intersection, and nontrivial line-bundle twisting in the Néron model; an error in any of these, including the exact rational form of the second component (twist vs. C2), would invalidate the torsion-counting contradictions and the induction. This is a genuine gap in support, not a disagreement with consensus.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper determines, for curves of genus at least 2 over an algebraically closed field of characteristic 0 with simple Jacobian, all possible automorphism groups, and proves that every group in the list occurs. The classification part proceeds by showing that the action of Aut(C) on H^0(C,Ω^1) is fixed-point-free, then applying Wolf's classification together with superelliptic genus restrictions; the resulting list is cyclic p-power, cyclic product of two distinct primes except 6, generalized quaternion, or trivial. The constructive part is substantial: for cyclic p-power groups the author gives explicit families y^p = x∏(x^{p^{n-1}}-a_l), proves generic endomorphism algebra Q(ζ_{p^n}) by a combination of CM-type arguments, specialization techniques, and induction on the number of branch points, and obtains density-one simplicity and automorphism-group statements; for pq-cyclic groups the Catalan curve is shown to be the unique example; for generalized quaternion groups the author constructs hyperelliptic families y^2 = x(x^{2^{n+1}}-1)∏(x^{2^n}-a_l)(x^{2^n}-a_l^{-1}) and claims the analogous endomorphism-algebra computation. The paper also derives density-one results using height-counting and Masser's specialization bound.","tokens_in":53082,"tokens_out":10167,"duration_ms":97558,"significance":"If correct, this is a complete answer to a natural and long-studied question, and it is significantly stronger than prior results, which were largely limited to cyclic prime order or the classical quaternion group of order 8. The modular-unlikely-intersection consequences are also new. The paper is honest about what is proved and what is deferred: the cyclic constructions are written out in detail, including the delicate specialization and torsion-basis computations, and the author explicitly flags the parts of the quaternion case that are left as analogues of earlier arguments. This is a high-value contribution to the arithmetic geometry of Jacobians.","major_comments":[{"comment":"The new quaternion construction for n≥2 is not proved at the points where it is load-bearing. The section preamble states that many proofs are omitted; Proposition 6.13 is justified only as 'conceptually similar' to [CLV23, Lemma 6.3.2] and Proposition 4.16, Corollary 6.14 says its proof is 'identical' to Corollary 4.18, and Theorem 6.12 says its proof is 'identical' to Theorem 4.10. These are not cosmetic omissions. Proposition 6.13 supplies the admissible degeneration J_{T',t} ≅ Jac(C1)×Jac(C2)^2 and, crucially, the explicit specialization of the 2-torsion basis (6.3), including the new points P_i and the asymmetric assignment of D^±_{α_l,i} to the two copies of Jac(C2). Corollary 6.14 uses those formulas to rule out nontrivial isotypic decompositions of J^al, and Theorem 6.1 relies on this to prove simplicity. Since [CLV23] treats only n=1, the entire existence part for G_{2^n} with n","section":"Section 6.3, Prop. 6.13, Cor. 6.14, Thm. 6.12"}],"minor_comments":[{"comment":"The index ranges in the displayed formulas should be 0≤i≤2^{n+1}-1 and 0≤i≤2^n-1, not '1≤i≤2n+1' and '1≤i≤2n', to match the basis in (6.3).","section":"Section 6.3, formulas in Prop. 6.13(2)"},{"comment":"The notation G_n in the hyperelliptic-automorphism table conflicts with the generalized quaternion group G_{2^n} used later. Please disambiguate, e.g. by renaming the table entries or the quaternion notation.","section":"Theorem 2.5 and Section 6"},{"comment":"The height notation is introduced twice; in Definition 1.4 'H' is used before the formal definition, and the 'K L' in the text should be K^L. Minor but should be cleaned up.","section":"Definition 1.4 / Definition 2.18"},{"comment":"The reduction 'every cyclic subgroup of order n of Aut(P^1) is conjugate to z↦ζ_n z' is only true over an algebraically closed field. Since automorphisms are geometric and uniqueness is over the algebraic closure, the argument is acceptable, but the text should say 'over an algebraic closure of Q' to avoid the false rational-statement.","section":"Section 5, proof of Theorem 5.1"},{"comment":"The discussion of [ACF25] and [Fit24] is interesting but unrelated to the main theorem; consider moving it to a separate note or removing it, as it distracts from the main thread.","section":"Remark 4.27"}],"recommendation":"major_revision","confidential_remarks":"The stress-test concern about Section 6 is justified: the manuscript itself repeatedly says the proofs are omitted or 'identical' to earlier ones, and those omitted arguments are load-bearing for the quaternion existence part. I do not see a fatal flaw in the classification half or in the cyclic-constructions; the paper is coherent and the strategy is credible. The appropriate path is major revision, not rejection, because the missing details are of a type that the author has already supplied in Section 4 and can in principle be written out. If Proposition 6.13, Corollary 6.14, and Theorem 6.12 receive full proofs and no other technical issue emerges, I would support acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The headline: this is a serious paper that likely has the right general shape, but the quaternion existence half is under-supported. The reader's conditional verdict is about right.\n\nWhat is actually new: a complete classification answer to Question 1.1, filling gaps left by Zarhin, CLV23, and the Catalan curve literature. The classification half (Theorem 3.1) is clean: fixed-point-free representation, Wolf's list, superelliptic genus restrictions. The cyclic prime-power constructions in Section 4 are detailed, with CM-type arguments for L=1 and specialization/induction for larger L. The uniqueness of the C_pq example (Theorem 5.1) is a genuine addition. The paper also gives density-one families and notes the unlikely intersections angle.\n\nThe soft spot is exactly where the reader and the stress-test point: Section 6. Theorem 6.1, and hence Theorem 1.2(iii), relies on Proposition 6.13 (an admissible degeneration to Jac(C1) x Jac(C2)^2) and Corollary 6.14. The paper says the proofs are 'conceptually identical' or 'similar' to Section 4, but the computations involve a multistep blow-up and explicit 2-torsion specialization. In Section 4 the analogous Proposition 4.16 is actually proved in detail; Section 6 is not. Since CLV23 only handled n=1, all new quaternion cases depend on these omitted details. If the rational form of the second component (twist vs. C2) or the torsion specialization is wrong, the induction and the existence claim break. This is a real gap in support, not a stylistic quibble.\n\nThe paper is open about the omission, which is honest, but a main theorem should not rest on that. The remainder of the paper is, as far as I can tell, sound: the classification chain is coherent, and the density-one machinery (Masser-style specialization, Lemma 2.21) is standard.\n\nWho should read it: anyone working on automorphism groups of curves, simple Jacobians, or CM constructions in moduli. It deserves a serious referee. I would send it to review and ask the author to write out Section 6.3 (or at least provide the blow-up computations and the torsion specialization verification) before acceptance.\n\nRecommendation: engage; conditional accept after expansion of Section 6.","headline":"A serious classification paper that likely has the right shape, but the quaternion existence half rests on omitted degeneration proofs — send it to review with a request to expand Section 6.","tokens_in":53589,"tokens_out":2849,"would_cite":true,"duration_ms":27452,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14H40","14K12","14H37","14K05","14H45"],"pacs":[],"model":"deepseek-v4-flash","headline":"A curve whose Jacobian is simple has automorphism group cyclic, quaternion, or trivial — and every such group occurs.","keywords":["simple Jacobian","automorphism group of a curve","superelliptic curves","fixed-point-free representations","endomorphism algebras","specialization","complex multiplication","quaternion algebras"],"falsifier":"Compute the endomorphism algebra of the Jacobian of a concrete small member of the predicted families, e.g., C: y^3 = x(x^8 − 2)(x^8 − 3) over Q (p=3, n=2, L=2); the theorem predicts End^0(Jac(C)) ≅ Q(ζ_9). Using reduction modulo a prime ℓ ≠ 3 and counting points, or a period computation, any indication of an endomorphism not in Q(ζ_9), or a decomposition into lower-dimensional factors, would falsify the density-one claim for that family.","tokens_in":52614,"feed_emoji":"🔄","tokens_out":8244,"duration_ms":66819,"temperature":0.7,"pith_summary":"The paper settles the classification of automorphism groups of smooth projective curves of genus at least two over an algebraically closed field of characteristic zero whose Jacobian is simple (i.e., not isogenous to a product of lower-dimensional abelian varieties). The answer is short: the group must be cyclic of prime-power order, cyclic of order pq for distinct primes with pq ≠ 6, a generalized quaternion group, or trivial — and every group on this list actually occurs. The classification half is quick: simplicity forces every nontrivial automorphism to have a rational quotient, so the action on holomorphic differentials is fixed-point-free and the classification of finite groups with such representations applies. The existence half is the hard part: explicit superelliptic families are constructed, and a new specialization argument shows that 100% of the members (in a height-density sense) have simple Jacobian with the prescribed endomorphism ring. If correct, the results complete the answer to the question of which finite groups can act on a curve without splitting its Jacobian, and they produce many new examples of curves whose Jacobians carry unusually large endomorphism algebras.","feed_headline":"Simple Jacobians force cyclic, quaternion, or trivial automorphisms","feed_subtitle":"Complete classification, with explicit families realizing each group for 100% of parameter values.","key_machinery":"The load-bearing tool is a specialization lemma: for an admissible degeneration of a family of curves to a reducible fiber, the geometric endomorphism algebra of the generic Jacobian embeds into that of the special fiber, preserving dimensions of idempotent kernels, and torsion subgroups specialize bijectively. This enables an induction on the number of branch points: degenerate to a union of lower-genus curves, compare isotypic components, and rule out unexpected factors. The base cases are handled by complex multiplication theory, reading the CM type off the explicit equation; for the quaternion family, the endomorphism algebra is shown to be a definite quaternion division algebra. The cla","core_discovery":"Theorem 1.2: if C is a curve of genus ≥ 2 over an algebraically closed field of characteristic 0 with simple Jacobian, then Aut(C) is isomorphic to C_{p^n} for some prime p and n ≥ 1, or C_{pq} for distinct primes with pq ≠ 6, or the generalized quaternion group G_{2^n}, or the trivial group. Conversely, for every group in this list, there exists such a curve. The proof decouples into a short classification (simplicity implies fixed-point-free action on regular differentials, the classification of finite groups admitting such representations, plus superelliptic genus constraints) and a long existence argument that constructs families of curves y^p = x·∏(x^{p^{n-1}} - a_l) for the cyclic case","pith_inferences":["The specialization lemma likely has broader use: it converts a generic simplicity statement into a check on degenerate fibers, and the torsion bijection is a new tool for controlling torsion in families of abelian varieties; one could apply it to other families of superelliptic curves or to Prym varieties.","The connection to the classification of fixed-point-free representations suggests a geometric principle: curves with simple Jacobians are precisely those whose automorphism groups are space-form groups, the same groups that act freely on spheres. For higher-dimensional varieties or for curves over non-closed fields, analogous constraints might follow from a fixed-point-free condition on differenti","The density-one statement is measured by height and leaves open whether the exceptional set is finite or Zariski-closed; testing small parameter values (e.g., the families with L=2) could reveal whether the simplicity condition holds for all but finitely many fibers, a stronger statement than the paper claims.","The p=2, n>2 case shows a precise role of CM types: when the generic CM type has nontrivial stabilizer, the generic Jacobian is not simple, which is exactly why the theorem requires L≥3 in that case; this predicts that any family improving the L-bound would have to break the CM-stabilizer symmetry."],"forward_implications":["Every group in the classification is realized by a family of curves over Q; for 100% of the parameter values (ordered by height) the Jacobian is geometrically simple and its endomorphism ring is explicitly determined (Z[ζ_{p^n}] for cyclic p^n, and a definite quaternion order for G_{2^n}).","The curve y^p = x^q − 1 is the only curve over Q (up to isomorphism) with automorphism group C_{pq}, pq ≠ 6, and simple Jacobian; no such curve exists for pq = 6.","The constructed families give infinitely many new examples of unlikely intersections in moduli spaces of principally polarized abelian varieties: for genus g ≥ 9, the locus of Jacobians whose endomorphism ring contains Z[ζ_{p^n}] has negative expected dimension, yet the families land in it.","When the family has at least two branch-point parameters (L ≥ 2), the curves realize infinitely many distinct isomorphism classes and their Jacobians infinitely many distinct isogeny classes over Q.","The results pin down the possible automorphism groups of curves with simple Jacobians, resolving the motivating question completely; in particular the only non-abelian groups that can occur are the generalized quaternion groups G_{2^n}."],"fun_headline_variants":["Simple Jacobians restrict automorphisms to cyclic, quaternion, or trivial","Automorphism groups of simple Jacobian curves classified completely","Only cyclic, quaternion, or trivial groups act on simple Jacobian curves","Explicit families realize every allowed automorphism group for simple Jacobians"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The existence proofs rest on the specialization lemma, which assumes the chosen families admit semistable regular models over A^1 with Néron models whose special fibers are abelian varieties, and that endomorphism algebras and torsion subgroups behave faithfully under specialization; if any constructed degeneration violates these assumptions, the density-one simplicity conclusions do not follow.","fun_headline_variants_meta":{"raw":{"variants":["Simple Jacobians restrict automorphisms to cyclic, quaternion, or trivial","Automorphism groups of simple Jacobian curves classified completely","Only cyclic, quaternion, or trivial groups act on simple Jacobian curves","Explicit families realize every allowed automorphism group for simple Jacobians"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.0009,"raw_usage":{"total_tokens":3642,"prompt_tokens":608,"completion_tokens":3034,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":352,"completion_tokens_details":{"reasoning_tokens":2958}},"tokens_in":352,"tokens_out":3034,"duration_ms":20310,"temperature":1.0,"reasoning_tokens":2958,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T10:05:44.886466+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the endomorphism algebra of the Jacobian of a concrete small member of the predicted families, e.g., C: y^3 = x(x^8 − 2)(x^8 − 3) over Q (p=3, n=2, L=2); the theorem predicts End^0(Jac(C)) ≅ Q(ζ_9). Using reduction modulo a prime ℓ ≠ 3 and counting points, or a period computation, any indication of an endomorphism not in Q(ζ_9), or a decomposition into lower-dimensional factors, would falsify the density-one claim for that family.","supporting_citations":[],"review_version":1}