{"id":"c3e17298-986f-4e34-b4c3-6038f5ea2d35","arxiv_id":"1909.01581","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Surface groups admit faithful actions by germs of one-variable analytic diffeomorphisms, answering a question of Ghys.","lead":"This paper proves that the fundamental group of any closed orientable surface, or non-orientable surface of genus at least 4, can be embedded into the group of analytic germs of diffeomorphisms of the real or complex line fixing the origin. It answers a question of Étienne Ghys and gives three different proofs.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Appendix degree bound in Theorem 9.1 is false as written, so the non-orientable and general-field cases rest on an unproved lemma.","rationale":"I read the paper in good faith and agree with the reader's weakest-assumption analysis. The central theorem is supported by three proofs for orientable surfaces, and the p-adic proof in Section 7 appears to give an independent route for that case. The non-orientable case, however, has no proof independent of Theorem 3.4, and the appendix proof of Theorem 3.4 contains a concrete, checkable degree error: the composition Q_k o P_k has degree (2k+3)(2k+4), not (2k+1)(2k+2). This invalidates the induction as written. I also noticed a separate sign mismatch in the first orientable proof (Section 3.4 writes Phi_{s_N}(t_i) = f_i^N f_0^N, while equation (3.4) gives p o tau^N(t_i) = a_i^N a_0^{-N}, so the claimed equality Phi_{s_N} = rho_0 o p o tau^N does not hold verbatim). I do not treat this as the central concern because the p-adic proof independently establishes the orientable case, and the sign issue is easily repaired by choosing s_0 = lambda_0^{-N}. The appendix error is more load-bearing because it affects the non-orientable theorem and Theorem B, and no alternate proof is given for those parts. Since the defect is localized and plausibly repairable, the existing CONDITIONAL verdict is appropriate; I would not move to ACCEPT or REJECT on the current text.","tokens_in":30745,"tokens_out":20303,"duration_ms":181531,"concrete_test":"Expand Q_k o P_k symbolically for k >= 1 with P_k = z + eta z^2 prod_{j=0}^{2k}(z-z_j), Q_k = z + beta z^2 prod_{j=0}^{2k+1}(z-z_j) and generic z_j, eta, beta. Verify whether the leading term has degree (2k+3)(2k+4) rather than (2k+1)(2k+2). Then re-run the induction of Theorem 9.1 with the corrected degree bound and check whether the final statement still holds with a weaker bound such as (2ell+2)! instead of (2ell)!. If a finite polynomial bound still holds, the fix is formal; if not, Theorem 3.4 lacks a valid proof in the current manuscript.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing point is the Appendix proof of Theorem 9.1, the only support offered for Theorem 3.4 (free groups with prescribed multipliers). Theorem 3.4 is used in the non-orientable case in Section 4 and in Theorem B. In the recursion, P_k has degree 2k+3 because it is z + eta_k z^2 prod_{j=0}^{2k}(z-z_j), and Q_k has degree 2k+4 because it is z + beta_k z^2 prod_{j=0}^{2k+1}(z-z_j). The composition Q_k o P_k therefore has degree (2k+3)(2k+4) for generic nonzero eta_k, beta_k, not the asserted (2k+1)(2k+2). Thus the displayed bound deg(P_{k+1}) <= deg(P_k)(2k+1)(2k+2) <= (2k+2)! in equation (9.9) is false; the correct recurrence gives deg(P_{k+1}) <= (2k)! (2k+3)(2k+4), which exceeds (2k+2)! by the factor (2k+4)/(2k+1). The induction as written therefore does not prove the stated degree bound. The orientable case has an independent p-adic proof in Section 7, but Theorem 4.1 for non-orientable surfaces and Theorem B for general complete valued fields depend directly on Theorem 3.4, and no alternative proof is supplied for them. The result is probably repairable by replacing the degree bound, but the manuscript as submitted does not contain a valid proof of these statements.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs embeddings of closed surface groups into the group of germs of analytic diffeomorphisms in one variable. Theorem A asserts that the fundamental group of any closed orientable surface, and of any closed non-orientable surface of genus at least 4, embeds into Diff(R,0) and hence into Diff(C,0). Three proofs are presented: a first proof via Koenigs linearization and Baumslag's residual-freeness argument, a second proof using a final topology on the group of germs, and a p-adic proof for the orientable case. The appendix proves a free-group realization theorem (Theorem 9.1, from which Theorem 3.4 is derived), and Theorem B extends the embedding statements to complete non-discrete valued fields. The orientable case over C has an independent p-adic proof in Section 7, but the non-orientable case and the general-valued-field statements rely on the appendix theorem.","tokens_in":31082,"tokens_out":8072,"duration_ms":76582,"significance":"If the missing technical points are repaired, the paper answers a question of Ghys and provides the first analytic embeddings for non-orientable surface groups in this setting. The paper contains several useful and transferable ingredients: a clean Baire-category framework (Lemma 3.1), a careful final topology on Diff(k,0) with continuity of Koenigs linearization (Sections 5–6), and an explicit p-adic construction with controlled jet data (Section 7). The p-adic proof of the orientable case appears sound and is independent of the contested appendix. The main weakness is that Theorem 3.4, the only source of free groups with prescribed multipliers, is supported by an appendix whose induction contains a false degree estimate; consequently the non-orientable and general-valued-field claims are not fully proved as submitted.","major_comments":[{"comment":"The degree bound in the induction of Theorem 9.1 is false. With P_k(z)=z+η_k z^2∏_{j=0}^{2k}(z−z_j) and Q_k(z)=z+β_k z^2∏_{j=0}^{2k+1}(z−z_j), one has deg(P_k)=2k+3 and deg(Q_k)=2k+4, so deg(Q_k∘P_k)=(2k+3)(2k+4) for generic nonzero η_k and β_k. The text asserts deg(S_k)≤(2k+1)(2k+2), and Eq. (9.9) concludes deg(P_{k+1})≤deg(P_k)(2k+1)(2k+2)≤(2k+2)!. The correct recurrence gives (2k)!(2k+3)(2k+4), which exceeds (2k+2)! by the factor (2k+4)/(2k+1). The induction therefore does not establish the stated polynomial bound, and Theorem 3.4, which is a corollary of Theorem 9.2, is not proved. Since Theorem 3.4 is used in the first proof of the orientable case, in the non-orientable case of Section 4, in the second proof of Section 6.2, and in Theorem B, these statements currently lack a valid proof. The orientable case over C has an independent p-adic proof in Section 7, but the non-orientable case and the general-field statements do not.","section":"Appendix, Theorem 9.1 and Eq. (9.9)"},{"comment":"Lemma 4.3 is not proved: the text states 'For the proof. The proof of this statement is completely analogous ... we leave it as an exercise to the reader.' This lemma is load-bearing because it provides the separation property in Lemma 3.1 for the even-genus non-orientable embedding, and no alternative proof is supplied. The reference to [9, Proposition 4.13] may contain an analogous statement, but the dependence is not made explicit and the needed statement is not quoted. Without a complete proof of Lemma 4.3, Theorem 4.1 for even genus is unsupported.","section":"§4.1, Lemma 4.3"}],"minor_comments":[{"comment":"The word 'retriction' should be 'restriction' in the last sentence of the proof of Lemma 4.4.","section":"§4.2, Lemma 4.4 proof"},{"comment":"In the proof of Theorem 9.2, the phrase 'over the base field of k, see []' contains an empty citation; the authors should either supply a reference or justify the existence of the algebraically independent sequence (c_i) directly.","section":"§9, Theorem 9.2 proof"},{"comment":"The notation \\bar f is used both for the prescribed power series in Eq. (7.3) and for the unknown germ solving the conjugacy equation; renaming one of these would remove a genuine ambiguity.","section":"§7.3, after Eq. (7.3)"}],"recommendation":"major_revision","confidential_remarks":"The paper is technically rich and likely correct in its orientable-case conclusion, but the appendix proof of Theorem 3.4 and the missing proof of Lemma 4.3 block the non-orientable and general-field claims. I recommend major revision rather than rejection, since the issues appear repairable within the scope of the manuscript."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe orientable half of the main result is in good shape; the non-orientable half and the general valued-field version are not, because of a false degree bound in the appendix.\n\nThe paper's real contribution is constructing embeddings of surface groups into Diff(C,0), answering a question of Ghys. The three proof strategies are genuinely different: a Baumslag/Dehn-twist argument with flows from Koenigs linearization, a new final topology on the group of germs that makes the representation variety tractable, and a p-adic proof that transfers embeddings from compact p-adic groups to complex analytic germs. The p-adic proof (Section 7) is independent of the contested theorem and appears sound; it already proves the orientable surface group case over C.\n\nThe soft spot is Theorem 3.4 / Theorem 9.1 in the appendix. The induction for the degree bound claims deg(Q_k∘P_k) ≤ (2k+1)(2k+2). Direct computation gives deg(Q_k∘P_k) = (2k+3)(2k+4) in the generic case. The recurrence in equation (9.9) therefore does not prove the stated bound, and the conclusion deg(P_{k+1}) ≤ (2k+2)! does not follow. This theorem is used only for the non-orientable case (Section 4) and for Theorem B on complete valued fields; those statements currently lack a valid proof. The result may be perfectly true and the fix might be a matter of replacing the degree bound with something softer, but as written the appendix does not support it.\n\nMinor point: Lemma 4.3 is left as an exercise with a reference; that's acceptable for a paper of this scope, but it is another unproved step in the non-orientable chain.\n\nWho is the paper for? People working on local analytic group actions, foliations, and embedding problems for surface groups. The p-adic and final-topology techniques are worth studying even if the appendix needs repair. The orientable case will likely stand.\n\nRecommendation: send it to peer review, but ask the authors to fix or replace the appendix lemma, or remove or shrink the claims that depend on it.","headline":"The orientable case is proved by a sound p-adic argument; the non-orientable and general-field claims rest on an appendix degree bound that is plainly wrong, so the paper overreaches as written.","tokens_in":31631,"tokens_out":2527,"would_cite":true,"duration_ms":23169,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20F65","37C85","37F50"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every fundamental group of a closed oriented surface, and every closed non-oriented surface of genus at least 4, embeds into the group of real analytic germs fixing the origin, and hence into the complex group.","keywords":["surface groups","germs of analytic diffeomorphisms","Koenigs linearization","free groups","Baire category","valued fields","p-adic diffeomorphisms","fully residually free groups"],"falsifier":"Expand $Q_k\\circ P_k$ in the appendix's recursion: if the error term has degree $(2k+3)(2k+4)$ rather than the stated $(2k+1)(2k+2)$, the induction in Theorem 9.1 fails as written, leaving Theorem 3.4—and with it the first two proofs of Theorem A—without a complete proof until the bound is corrected or another construction is supplied.","tokens_in":30554,"feed_emoji":"🌀","tokens_out":13811,"duration_ms":116228,"temperature":0.7,"pith_summary":"The paper proves that the fundamental group of any closed orientable surface, and of any closed non-orientable surface of genus at least 4, embeds faithfully into the group of germs of real analytic diffeomorphisms of the line fixing the origin, and therefore also into the complex group. This answers a question of Ghys and gives the first analytic, not merely formal, realizations of surface groups as local symmetries. The authors offer three proofs: one based on Baumslag's residual freeness and Dehn twists, one on a new final topology on the space of germs, and one via compact p-adic groups and algebraic independence. The common engine is Koenigs linearization, which converts hyperbolic germs into flows of homotheties and produces large families of representations on which a Baire category argument forces faithfulness.","feed_headline":"Surface groups embed into germs of analytic diffeomorphisms","feed_subtitle":"Koenigs linearization and Baire category make every closed surface group act faithfully on germs at zero.","key_machinery":"The central mechanism is Koenigs linearization: a germ $f$ with $|f'(0)|\\ne1$ is uniquely conjugate to the homothety $z\\mapsto\\lambda z$, and the conjugacy extends $f$ to a multiplicative flow $s\\mapsto h\\circ m_s\\circ h^{-1}$ ($s\\in\\mathbb{R}^*_+$ or $k^*$) whose coefficients are polynomial in $s$ and $s^{-1}$. In the genus-2 presentation $\\Gamma_2=\\langle a_0,a_1,a_2,t_1,t_2 \\mid a_0a_1a_2=1,\\ a_0t_1^{-1}a_1t_1t_2^{-1}a_2t_2=1\\rangle$, the paper fixes $a_0,a_1,a_2$ as hyperbolic generators and sets $t_i=\\varphi_i^{s_i}\\varphi_0^{s_0}$; the relation holds automatically, and every $A_k(\\Phi_s(g))$ is polynomial in the $s_i^{\\pm1}$, so the set of parameters killing a word is either everything or a closed set with empty interior. Baumslag's asymptotic injectivity, applied to the Dehn twist along $a_0,a_1,a_2$, gives at least one parameter for which $g$ survives, and the Baire category lemma yields a residual faithful set. A second proof endows $\\mathrm{Diff}(k,0)$ with a final topology making it a topological group and an irreducible component of the representation variety; a third proceeds through compact p-adic groups and transfers to $\\mathbb{C}$ by algebraic independence.","core_discovery":"The central claim is that surface groups act faithfully by convergent analytic germs, not just by formal power series. The paper constructs, for the genus-2 group, a family of homomorphisms $\\Phi_s$ from $\\Gamma_2$ to $\\mathrm{Diff}(\\mathbb{R},0)$ parameterized by $s\\in(\\mathbb{R}^*_+)^3$, by fixing three hyperbolic generators $f_0,f_1,f_2$ of a free subgroup and letting two meridian generators move along the Koenigs multiplicative flows. Coefficient functions $A_k(\\Phi_s(g))$ depend polynomially on the $s_i^{\\pm1}$, so the family is Baire and irreducible; Baumslag's lemma plus a Dehn twist supplies parameters that separate every non-trivial word. A generic parameter is therefore faithful, and the same scheme, with suitable twists, handles all non-orientable genera $\\ge4$. The paper's Theorem B extends this to every complete non-discrete valued field, and a third proof obtains formal embeddings over $\\mathbb{Z}_p$ and converts them into convergent complex germs via an algebraically free sequence of small coefficients.","pith_inferences":["The appendix's inductive degree bound $(2k+1)(2k+2)$ for the error term of $Q_k\\circ P_k$ appears to be off by two; direct expansion gives $(2k+3)(2k+4)$, so the proof of Theorem 9.1 as written needs a correction before Theorem 3.4 can be considered proved by this text.","A corrected bound would likely still close the induction, but if Theorem 3.4 fails for some prescribed pair of multipliers, the first two proofs of Theorem A would need a different source of free subgroups with prescribed derivatives; the p-adic proof's dependence on it is looser.","The Baire-family strategy seems adaptable to other fully residually free groups, such as limit groups, provided one has a Dehn-twist-like automorphism and a Koenigs-style flow; the paper does not claim this.","Whether $\\Gamma_2$ embeds into the analytic diffeomorphisms of the circle fixing a point remains open; the paper's methods give germs at a fixed point, and a positive answer would produce via suspension a compact 3-manifold foliation transverse to a fibration over the genus-2 surface."],"forward_implications":["Theorem B embeds surface groups into $\\mathrm{Diff}(k,0)$ for every complete non-discrete valued field $k$, so surface groups act by germs over $\\mathbb{Q}_p$ and, after conjugation, by p-adic analytic homeomorphisms on thin annuli.","The image can be forced to contain any prescribed free subgroup generated by two hyperbolic germs, so surface groups arise as overgroups of dense free groups in $\\mathrm{Diff}(k,0)$.","Theorem C realizes every faithful formal representation of a surface group over $\\mathbb{R}$ by $C^\\infty$ germs with the same Taylor expansion, placing the analytic embeddings inside the smooth theory.","The non-orientable genus-3 group is excluded because it is not fully residually free, so the paper's Baumslag-based arguments do not apply to it.","The construction produces embeddings with controlled linear part—one generator can have transcendental derivative while the others are tangent to the identity—which matters for the rigidity of conjugacy classes."],"supporting_citations":[{"why":"Baumslag's lemma that Γ2 is fully residually free; supplies the asymptotically injective homomorphisms used for separation.","marker":"[1]"},{"why":"Berthier–Cerveau–Lins Neto prove free subgroups for generic pairs of derivatives; the appendix extends this to all pairs.","marker":"[2]"},{"why":"Breuillard–Gelander–Souto–Storm compact-group argument and Dehn twist; the template for the Baire-family construction and Proposition 1.2.","marker":"[3]"},{"why":"Glass's account of the Wilson trick, producing free groups tangent to the identity with integer coefficients.","marker":"[10]"},{"why":"Herman–Yoccoz non-Archimedean Koenigs linearization, needed for valued fields in Theorem B.","marker":"[11]"},{"why":"Leslie's final topology on analytic germs, adapted here to make Diff(k,0) a topological group.","marker":"[15]"},{"why":"Mattei–Rebelo–Reis dynamic construction of free subgroups of Diff(C,0); the appendix adapts its proof.","marker":"[17]"},{"why":"Milnor's Koenigs linearization theorem for complex germs; the core linearization device.","marker":"[18]"},{"why":"Szegedy's theorem that almost every pair in the Nottingham group generates a free group, used in finite characteristic and for p-adic compact groups.","marker":"[22]"},{"why":"White's theorem that z↦z+1 and z↦z^p generate a free group; base of the explicit p-adic free group.","marker":"[25]"}],"fun_headline_variants":["Surface groups embed faithfully in analytic diffeo germs","All surface groups act faithfully on analytic germs","Koenigs flows give faithful surface group actions on analytic germs","Surface groups act faithfully by convergent analytic germs"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that for every pair of non-zero multipliers one can find two germs with those derivatives that generate a free group (Theorem 3.4); the appendix's proof of that lemma relies on an induction whose stated degree bound $(2k+1)(2k+2)$ is contradicted by direct computation, which gives $(2k+3)(2k+4)$, so the induction as written does not close.","fun_headline_variants_meta":{"raw":{"variants":["Surface groups embed faithfully in analytic diffeo germs","All surface groups act faithfully on analytic germs","Koenigs flows give faithful surface group actions on analytic germs","Surface groups act faithfully by convergent analytic germs"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000286,"raw_usage":{"total_tokens":1596,"prompt_tokens":772,"completion_tokens":824,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":388,"completion_tokens_details":{"reasoning_tokens":763}},"tokens_in":388,"tokens_out":824,"duration_ms":7081,"temperature":1.0,"reasoning_tokens":763,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T05:15:45.734626+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Expand $Q_k\\circ P_k$ in the appendix's recursion: if the error term has degree $(2k+3)(2k+4)$ rather than the stated $(2k+1)(2k+2)$, the induction in Theorem 9.1 fails as written, leaving Theorem 3.4—and with it the first two proofs of Theorem A—without a complete proof until the bound is corrected or another construction is supplied.","supporting_citations":[{"cited_title":"On generalised free products","cited_arxiv_id":null,"evidence_quote":"Baumslag's lemma that Γ2 is fully residually free; supplies the asymptotically injective homomorphisms used for separation."},{"cited_title":"Berthier, D","cited_arxiv_id":null,"evidence_quote":"Berthier–Cerveau–Lins Neto prove free subgroups for generic pairs of derivatives; the appendix extends this to all pairs."},{"cited_title":"Dense embeddings of surface groups","cited_arxiv_id":null,"evidence_quote":"Breuillard–Gelander–Souto–Storm compact-group argument and Dehn twist; the template for the Baire-family construction and Proposition 1.2."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Glass's account of the Wilson trick, producing free groups tangent to the identity with integer coefficients."},{"cited_title":"Herman and J.-C","cited_arxiv_id":null,"evidence_quote":"Herman–Yoccoz non-Archimedean Koenigs linearization, needed for valued fields in Theorem B."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Leslie's final topology on analytic germs, adapted here to make Diff(k,0) a topological group."},{"cited_title":"Mattei, J","cited_arxiv_id":null,"evidence_quote":"Mattei–Rebelo–Reis dynamic construction of free subgroups of Diff(C,0); the appendix adapts its proof."},{"cited_title":"Dynamics in one complex variable , volume 160 of Annals of Mathematics Studies","cited_arxiv_id":null,"evidence_quote":"Milnor's Koenigs linearization theorem for complex germs; the core linearization device."},{"cited_title":"Almost all finitely generated subgroups of the Nottingham group are free","cited_arxiv_id":null,"evidence_quote":"Szegedy's theorem that almost every pair in the Nottingham group generates a free group, used in finite characteristic and for p-adic compact groups."},{"cited_title":"The group generated by x↦→ x + 1 and x↦→ xp is free","cited_arxiv_id":null,"evidence_quote":"White's theorem that z↦z+1 and z↦z^p generate a free group; base of the explicit p-adic free group."}],"review_version":1}