{"id":"07e108c9-d810-4666-9e71-215b5f372aee","arxiv_id":"1908.06675","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Every finitely generated group is the automorphism group of a Riemann surface (compact for finite groups), proved via maximal non-arithmetic triangle groups and Margulis' commensurability theorem.","lead":"This paper gives a short algebraic proof of Greenberg's theorem for finitely generated groups: every such group is the automorphism group of a Riemann surface, compact when the group is finite. The proof builds the surface from a quotient of a hyperbolic triangle group and then uses a commensurability argument to control the full automorphism group.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified; the proof is internally sound and the delicate normalizer equality follows correctly from the cited Margulis/Singerman/Takeuchi input.","rationale":"I traced the proof step by step. The existence of N as a maximal torsion-free surface subgroup follows from Macbeath's epimorphism and primitivity of PSL2(q) on the projective line. The construction of θ with M=kerθ and the conclusion NΔ(M)=N are valid, including the WLOG choice of B1∉K via a symplectic change of basis. The final normalizer equality is the delicate point: N(M) is a Fuchsian group containing the cocompact groups N and N^g, so N and N^g are commensurable; since Comm(N)=Comm(Δ)=Δ by Margulis plus maximality of Δ, every g∈N(M) lies in Δ, hence N(M)=N. This is correct. The reader's weakest assumption about reliance on external classification and commensurability theorems is accurate, but these are established results used correctly; the paper also gives concrete triples satisfying the needed hypotheses. Thus I have no load-bearing correctness concern, and the verdict should remain unchanged.","tokens_in":5053,"tokens_out":47213,"duration_ms":501628,"concrete_test":"Independently verify in GAP that PSL2(311) contains a generating triple (x,y,z) of orders 2,3,13 with xyz=1, and that the inverse image of PSL2(311)_∞ in Δ(2,3,13) is a torsion-free surface group of genus 15; this checks the most concrete cited input of the proof.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I find no internal gap. The construction yields a maximal torsion-free surface subgroup N of Δ; the argument NΔ(M)=N is sound via maximality of N and B1∉K, and the final equality N(M)=NΔ(M) follows because Margulis gives Comm(Δ)=Δ for non-arithmetic maximal Δ, so any g∈N(M) commensurates N and hence lies in Δ. The only soft spot is that the proof leans on substantial external results (Macbeath for the PSL2(q) epimorphism; Singerman/Takeuchi for a maximal non-arithmetic triangle group; Margulis for the commensurator). These are established theorems and the paper cites them precisely, so I do not treat this as a correctness objection.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper gives a short algebraic proof of Greenberg's theorem for finitely generated groups: every such group A occurs as the automorphism group of a Riemann surface, and the surface can be taken compact when A is finite. The construction chooses a hyperbolic triangle group Δ(l,m,n), uses Dirichlet's theorem and Macbeath's theorem to map Δ onto PSL2(Fq) for a suitable prime power q, and takes N to be the preimage in Δ of the stabilizer of a point. The subgroup N is a torsion-free surface group of genus g; for a given A, a normal subgroup M⊲N is chosen with N/M≅A. The main technical step is proving that the normalizer of M in PSL2(R) coincides with its normalizer in Δ. This is achieved by choosing Δ to be maximal and non-arithmetic, so that Margulis' theorem identifies the commensurator of Δ with Δ itself, and then showing that any element normalizing M lies in that commensurator.","tokens_in":5191,"tokens_out":51637,"duration_ms":495999,"significance":"If the proof is correct, the paper provides a notably short and explicit route to a restricted version of a classical theorem, avoiding the N-equivalence and N-maximality machinery of Greenberg's original proof. The construction is transparent, yields explicit genus bounds, and gives concrete examples of usable triangle groups. The proof is not elementary, but the paper is honest about this and precisely cites the external inputs: Dirichlet, Macbeath, Singerman, Takeuchi, and Margulis. The value of the paper is mainly methodological and expository: it shows how standard tools about triangle groups, finite quotients, and commensurability combine to realize automorphism groups of Riemann surfaces.","major_comments":[],"minor_comments":[{"comment":"The symbol g is used both for the genus in Eq. (1) and for an arbitrary element of N(M) in the final paragraph; please use a different letter (for instance γ) for the group element to avoid confusion.","section":"Section 2, final paragraph"},{"comment":"The triple (l,m,n) is not fixed to be maximal and non-arithmetic until after N and M have already been constructed; please move this choice earlier or state explicitly that from that point onward the triple is chosen with these properties and q is then selected so that g ≥ d.","section":"Section 2, transition before the final step"},{"comment":"The notation for the commensurator appears to be lost in the typeset text (it reads 'its commensurator Δ in PSL2(R)'); use a distinct symbol such as \\tilde{\\Delta}, and add one sentence explaining that a finite-index subgroup N of Δ has the same commensurator as Δ.","section":"Section 2, final paragraph"},{"comment":"There are several small presentation issues: 'nornaliser' should be 'normaliser'; 'generators Ai of Δ' should be 'generators Ai of N'; and the subgroup M and the hypermap M are denoted by the same symbol, which is confusing.","section":"Section 2, typos and notation"},{"comment":"The claim that N has no elliptic elements would be clearer if it explicitly invoked the standard fact that every elliptic element of a triangle group is conjugate to a power of X, Y, or Z, so the semiregularity of the images rules out elliptic elements in N.","section":"Section 2, paragraph on elliptic elements"},{"comment":"The statement that N and N^g have finite index in N(M) because they are cocompact is correct but terse; one sentence indicating that a cocompact Fuchsian subgroup of a Fuchsian group necessarily has finite index would improve readability.","section":"Section 2, final paragraph"}],"recommendation":"minor_revision","confidential_remarks":"This is a short and correct proof of a known theorem, with the value lying in the concise combination of external results. The reliance on Margulis, Singerman, Takeuchi, and Macbeath is precisely cited and openly acknowledged in Remark 1. The only concerns are local matters of notation and exposition, not correctness."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe main thing you should know: this is a new proof of a non-new theorem. Greenberg proved the statement for countable groups; Jones proves the finitely generated case, explicitly, using triangle groups, maximal non-arithmetic triangle groups, and Margulis. The proof is about three pages, and it is honest about being a restricted version. If you want a route to Greenberg for finite groups that does not follow Greenberg's original maximality machinery, this is the best I know.\n\nWhat's good: the construction is concrete. Pick a hyperbolic triangle group Δ(l,m,n) with Δ maximal non-arithmetic. Use Dirichlet plus Macbeath/Garion to get a Fuchsian kernel N of genus g ≥ d, then map N onto A, take kernel M. The delicate step is showing Aut(H/M) = N(M)/M equals A, i.e., that the normalizer of M in PSL2(R) does not exceed Δ. The argument via maximality of Δ and Margulis' commensurator theorem is sound. I checked the normalizer equality; the logic holds. The proof does not smuggle in Greenberg's theorem—all tools are independent classifications and commensurability facts.\n\nSoft spots: the title is slightly overbroad, because the paper proves only finitely generated groups, not all countable groups. The abstract and Remark 1 make that clear, but the title invites quibbles. The proof leans on heavy external results—Singerman, Takeuchi, Margulis, Macbeath—so \"short\" means short in execution, not elementary. For a finite A the surface is compact; for infinite finitely generated A you get a non-compact surface, as expected. There are a few typos (e.g., 'nornaliser', 'B1 /nelementK') but no mathematical errors I could find. The remarks about variants are a nice bonus, especially the super-exponential genus growth via symmetric groups.\n\nWho this is for: people working on automorphism groups of Riemann surfaces, dessins, and Fuchsian groups. It is a useful reference and could be assigned to students as a modern route to a classical theorem. The theorem's age and the heavy machinery cap the significance, but the proof is rigorous and reusable. I would not desk reject it; it deserves a serious referee, and I expect a good referee would accept with minor revisions.\n\nRecommendation: send it to review. It is not a breakthrough, but it is a correct, well-cited, clearly written proof that improves accessibility of a known result.","headline":"A clean, explicit proof of a known theorem; useful, honest about its limits, and sound enough to referee.","tokens_in":5659,"tokens_out":13109,"would_cite":false,"duration_ms":128416,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20H10","11F06","11G32","14H57","20B25","30F10"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every finitely generated group is the full symmetry group of a Riemann surface.","keywords":["Riemann surface","automorphism group","triangle group","Fuchsian group","commensurator","finite group","surface-kernel epimorphism","dessin d'enfant"],"falsifier":"Compute the automorphism group of the surface $\\mathbb{H}/M$ for a small group, say $A=C_2$, using the paper's explicit choices $(l,m,n)=(2,3,13)$ and $q=311$: if the conformal automorphism group is strictly larger than $C_2$, the normalizer equality that carries the proof is false for that instance. To disprove the theorem itself, one would need a finitely generated $A$ for which every admissible choice of triple, prime power, and epimorphism leaves the automorphism group larger than $A$.","tokens_in":4877,"feed_emoji":"🔺","tokens_out":23709,"duration_ms":206675,"temperature":0.7,"pith_summary":"This paper proves that every finitely generated group $A$ is isomorphic to the full automorphism group of a Riemann surface, and that the surface can be chosen compact whenever $A$ is finite. The result is a restricted form of a long-standing theorem, but the proof is new: instead of constructing maximal Fuchsian groups with prescribed signatures, it builds the surface from a hyperbolic triangle group and a finite quotient $\\mathrm{PSL}_2(\\mathbb{F}_q)$. A reader should care because the construction is explicit, algebraic, and arithmetically parameterized, so the theorem becomes something one can see and test rather than a distant existence result.","feed_headline":"Every finitely generated group is the full symmetry group of a surface","feed_subtitle":"A short algebraic proof builds the surface from triangle groups and finite fields.","key_machinery":"The load-bearing object is a hyperbolic triangle group $\\Delta(l,m,n)$, the group generated by $X,Y,Z$ with $X^l=Y^m=Z^n=XYZ=1$ and $l^{-1}+m^{-1}+n^{-1}<1$, acting by isometries on the hyperbolic plane. It is chosen maximal and non-arithmetic so that its commensurator in $\\mathrm{PSL}_2(\\mathbb{R})$ is itself: every element of the ambient group that maps a finite-index subgroup of $\\Delta$ to another subgroup with finite-index intersection is already inside $\\Delta$. That fact turns the normalizer of the kernel $M$ into a group computable inside $\\Delta$, and the chain $\\operatorname{Aut}(S)\\cong N_{\\mathrm{PSL}_2(\\mathbb{R})}(M)/M=N_{\\Delta}(M)/M\\cong A$ carries the whole argument.","core_discovery":"At the center of the paper is the claim that for any finitely generated group $A$ there is a Riemann surface $S$ whose automorphism group is exactly $A$, and if $A$ is finite the surface may be taken compact. The construction fixes a hyperbolic triangle group $\\Delta(l,m,n)$ that is maximal and non-arithmetic, chooses a prime power $q$ for which $\\Delta$ maps onto $\\mathrm{PSL}_2(\\mathbb{F}_q)$ with the three standard generators retaining orders $l,m,n$, and takes the inverse image $N$ of the stabilizer of $\\infty$; this $N$ is a surface group of genus large enough to map onto $A$. The kernel $M$ of that epimorphism yields $S=\\mathbb{H}/M$, and maximality plus non-arithmeticity imply that the normalizer of $M$ in $\\mathrm{PSL}_2(\\mathbb{R})$ equals its normalizer in $\\Delta$, giving $\\operatorname{Aut}(S)\\cong N/M\\cong A$.","pith_inferences":["The paper leaves open whether arithmetic or non-maximal triangle groups can realize groups exactly; one plausible reading is that the maximal non-arithmetic condition is exactly the feature that prevents extra automorphisms, so any relaxation must replace the commensurability step with a new idea.","Because every step is algebraic and parameterized by $q$, a small computational search could turn the existence argument into an explicit generator of curves or dessins with a prescribed finite automorphism group.","For a fixed group $A$, varying the triple, the prime power, and the epimorphism should produce many non-isomorphic surfaces realizing $A$; counting such surfaces is a natural next question."],"forward_implications":["Every finite group occurs as the full automorphism group of a compact Riemann surface, which the paper notes is then defined over a number field.","Every infinite finitely generated group is realized as the exact symmetry group of a non-compact Riemann surface.","The proof is an arithmetic recipe: for a group of rank $d$, choose a maximal non-arithmetic triangle group and a prime power $q$ so that the resulting surface has genus at least $d$, then form the kernel of a map onto $A$.","Replacing the natural action of $\\mathrm{PSL}_2(\\mathbb{F}_q)$ by other primitive actions makes the genus grow cubically or super-exponentially, so the construction reaches groups of very large rank."],"supporting_citations":[{"why":"Supplies the classification of maximal Fuchsian triangle groups; the proof needs the chosen triple to be maximal.","marker":"[20]"},{"why":"Supplies the classification of arithmetic triangle groups; the proof needs the chosen triple to be non-arithmetic.","marker":"[22]"},{"why":"Supplies the theorem that a non-arithmetic Fuchsian group is its own commensurator, forcing the final normalizer equality.","marker":"[18, §IX.7]"},{"why":"Supplies generators of $\\mathrm{PSL}_2(\\mathbb{F}_q)$ with prescribed element orders, giving the surface-kernel epimorphism from $\\Delta$.","marker":"[17]"},{"why":"Also supplies surface-kernel epimorphisms from triangle groups to $\\mathrm{PSL}_2(\\mathbb{F}_q)$ with the needed element orders.","marker":"[4, Corollary C]"},{"why":"Shows $N(M)$ is a Fuchsian group when $M$ is a non-cyclic Fuchsian group.","marker":"[14, Theorem 5.7.5]"},{"why":"Identifies the automorphism group of $\\mathbb{H}/M$ with $N(M)/M$.","marker":"[14, Theorem 5.9.4]"},{"why":"Provides the proof strategy of realizing groups as automorphism groups of hypermaps by kernels of epimorphisms from triangle groups.","marker":"[12, Theorem 3(a)]"}],"fun_headline_variants":["Every finitely generated group is a surface's full symmetry group","All finitely generated groups are surface symmetry groups","Short proof: every finitely generated group is a surface automorphism group","Surfaces realize every finitely generated symmetry group"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof needs the chosen triangle group to have no commensurability symmetries outside itself, a fact obtained by combining maximality with non-arithmeticity; if that guarantee failed, the final step equating the surface's automorphism group with $N/M$ would collapse.","fun_headline_variants_meta":{"raw":{"variants":["Every finitely generated group is a surface's full symmetry group","All finitely generated groups are surface symmetry groups","Short proof: every finitely generated group is a surface automorphism group","Surfaces realize every finitely generated symmetry group"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001047,"raw_usage":{"total_tokens":4311,"prompt_tokens":770,"completion_tokens":3541,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":386,"completion_tokens_details":{"reasoning_tokens":3474}},"tokens_in":386,"tokens_out":3541,"duration_ms":23790,"temperature":1.0,"reasoning_tokens":3474,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:38:13.478224+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the automorphism group of the surface $\\mathbb{H}/M$ for a small group, say $A=C_2$, using the paper's explicit choices $(l,m,n)=(2,3,13)$ and $q=311$: if the conformal automorphism group is strictly larger than $C_2$, the normalizer equality that carries the proof is false for that instance. To disprove the theorem itself, one would need a finitely generated $A$ for which every admissible choice of triple, prime power, and epimorphism leaves the automorphism group larger than $A$.","supporting_citations":[{"cited_title":"Singerman, Finitely maximal Fuchsian groups, J","cited_arxiv_id":null,"evidence_quote":"Supplies the classification of maximal Fuchsian triangle groups; the proof needs the chosen triple to be maximal."},{"cited_title":"Takeuchi, Arithmetic triangle groups, J","cited_arxiv_id":null,"evidence_quote":"Supplies the classification of arithmetic triangle groups; the proof needs the chosen triple to be non-arithmetic."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies generators of $\\mathrm{PSL}_2(\\mathbb{F}_q)$ with prescribed element orders, giving the surface-kernel epimorphism from $\\Delta$."}],"review_version":1}