{"id":"f7484944-ac27-4242-ae9e-9b54cf91355f","arxiv_id":"2411.18863","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For every finite nonabelian simple group T, the order |T| is bounded above by a function of m(T), the maximum number of Aut(T)-classes of p-elements over all primes p.","lead":"This paper proves that the size of any finite nonabelian simple group is bounded by a function of the maximum, over all primes p, of the number of automorphism classes of elements of p-power order. This generalizes a 1992 bound of Pyber and has consequences for Brauer groups of global fields.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified","rationale":"I read the proof as a whole, focusing on the places where a hidden assumption could break the bound. The reduction to bounding q via a cyclic torus H is sound: Lemma 2.3 correctly transfers Aut(T)-class counts to normalizer indices, and the applications in Propositions 4.2 and 5.3 use the resulting inequalities with valid estimates on |π(H)| and |H_s|. I checked the linear, symplectic, and orthogonal cases of Proposition 4.2, including the handling of scalar matrices in the cases m=d, and found no gap. In Section 5, the exceptional-group argument is entirely conditional on the cited maximal subgroup classifications; the reader's weakest assumption identifies exactly this. Since these classifications are standard and the paper does not purport to reprove them, I do not regard the dependence as an internal flaw or as a reason to change the ACCEPT verdict. The only textual issue I found is a harmless overcount in Lemma 4.1 for PΩ^-(2d,q) with d even, where the displayed count d should be d−1 for nonidentity involution classes; the required total of d unipotent classes still holds once the identity class is included. No load-bearing error surfaced, so the verdict remains unchanged.","tokens_in":20490,"tokens_out":32666,"duration_ms":279390,"concrete_test":"Verify the E7(q) rows of Table 4 against Craven [7, Table 4.1] and Liebeck–Saxl–Seitz [24, Table 5.1]: confirm the torus orders (q±1)/d and that N_T(H)=(E6(q)×H).2 or (2E6(q)×H).2 is maximal in Aut(T). This is the least independently verified step in Proposition 5.3, and Lemma 5.1(2)–(3) depends on it more heavily than the other rows.","verdict_should_be":"UNCHANGED","load_bearing_attack":"No significant objection identified. The central argument is internally coherent: the counting lemmas (2.3, 4.1, 5.1, 5.2) are applied correctly, and the arithmetic bounds in Propositions 4.2 and 5.3 close. The obvious soft spot is the dependence of Section 5 on the cited maximal subgroup classifications (Liebeck–Saxl–Seitz [24], Kleidman [20,21], Malle [26], Suzuki [35], Craven [7,8]). If any Table 4 row misstated a torus normalizer, Proposition 5.3 would fail for that family. However, this is an external literature dependence of a standard kind, not an internal inconsistency, and no row appears suspect. The minor verbal overcount in Lemma 4.1 for PΩ^-(2d,q) with d even (d written where d−1 nonidentity classes are meant) does not affect the conclusion, since the identity supplies the missing class.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves that there exists an increasing function f such that every finite nonabelian simple group T satisfies |T| ≤ f(m(T)), where m(T) is the maximum, over all primes p, of the number of Aut(T)-classes of elements of p-power order in T. The proof combines an elementary counting lemma (Lemma 2.3), which controls the number of Aut(T)-classes meeting a cyclic subgroup in terms of the index of its normalizer, with a case-by-case analysis over the finite simple groups: alternating groups, classical groups of Lie type, and exceptional groups of Lie type. For classical groups the argument uses lower bounds on unipotent class numbers and carefully chosen cyclic tori whose normalizers have small index; for exceptional groups it relies on known maximal-subgroup classifications to identify cyclic subgroups with maximal normalizers. The paper also gives a corollary lower bound on the number of Aut(T)-classes of p-elements in terms of |T| and discusses applications to relative Brauer groups.","tokens_in":15,"tokens_out":35206,"duration_ms":853143,"significance":"If correct, this is a substantial strengthening of previous results of Pyber and of Hethelyi and Kulshammer, replacing bounds on the total number of conjugacy classes by a much weaker statistic, the maximum over primes of the number of Aut(T)-classes of p-power elements. The proof is structured and auditable: the counting arguments are elementary, the constants are explicit, and there are no fitted parameters. The main risk is the external dependence of Section 5: Lemma 5.1 and Table 4 rest on the maximal-subgroup classifications in [7,8,20,21,24,26,35], and a misstated normalizer row would break Proposition 5.3 for the corresponding family. This is a standard kind of reliance in the subject, and I found no internal inconsistency or circularity; the use of the classification of finite simple groups is independent of the theorem being proved.","major_comments":[],"minor_comments":[{"comment":"The sentence beginning 'Since T is pp-bounded by n, it follows from Lemma 2.3 that |C(s)| divides z|Out(T)|n!' misstates the conclusion of Lemma 2.3: what divides z|Out(T)|n! is φ(|Hs|), not |C(s)|. The deduction that s−1 divides 360a(n!) uses φ(|Hs|)=r|C(s)| with r dividing z|Out(T)|, and the current wording is insufficient.","section":"§5, Proposition 5.3, Case a)i"},{"comment":"The estimate '|Hs| ≤ z|Out(T)|n' for odd s drops the factor s/(s−1), which is explicitly included in the corresponding step of Case a)i). The correct bound is |Hs| ≤ 2z|Out(T)|n ≤ 8an; the final qualitative conclusion is unaffected once this factor is restored.","section":"§5, Proposition 5.3, Case a)ii"},{"comment":"In the proof of statement (1), the one-line deduction 'Since s is coprime to z, it follows that Hs = Os(NT(H))' deserves a brief justification when s divides |D|: because D is simple nonabelian, Os(D)=1, and because Z has order coprime to s, the largest normal s-subgroup of NT(H) lies in D×H and equals Hs. Please add this clarification.","section":"§5, Lemma 5.1"},{"comment":"In the displayed chain 'q < (q^3−1)/(q−1)^2 ≤ Prod(3,q)/(q−1)_3 = 3|H|', the final denominator is the 3-part of q−1, not the cube of q−1; using the same subscript notation as elsewhere would avoid ambiguity.","section":"§4, Proposition 4.2, linear exceptional case"},{"comment":"The sentence 'Let H be the cyclic subgroup of T defined in Lemma 5.1' should refer to Lemma 5.2, since this case treats the groups in Table 5.","section":"§5, Proposition 5.3, Case b"},{"comment":"The phrase 'we may assume q > 2 so that we may use all rows of Table 4' should explicitly say that q=2 gives only finitely many exceptional groups and that their orders can be absorbed into the function h; as written the reduction is slightly too quick.","section":"§5, Proposition 5.3, opening reduction"},{"comment":"There are several small typos: 'elements of elements order 3' in Lemma 3.1, 'for for all positive integers m' in Remark 3.2, and 'elements of p-elements' in the definition of mp(T) in the Introduction.","section":"§3 and §1"}],"recommendation":"minor_revision","confidential_remarks":"The main external vulnerability is the maximal-subgroup input in Section 5. I do not recommend rejection on this basis, but a careful editor may wish the authors to double-check the Table 4 rows against the cited theorems, especially the E6 and E7 rows where the cyclic torus can share primes with the other direct factor. The reported issues are all local and fixable; after these corrections the paper should be suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the theorem is new and the proof is detailed and mostly convincing. I'd send it to a referee.\n\nThe new thing is Theorem 1.1: for a finite nonabelian simple group T, |T| is bounded by a function of m(T), the maximum over primes p of the number of Aut(T)-classes of elements of p-power order. This is a real generalization of Pyber's bound, and it uses a statistic far weaker than the total number of conjugacy classes. It is exactly the kind of bound needed for Praeger's covering subgroup conjecture and for relative Brauer groups.\n\nThe proof is a genuine case analysis over alternating, classical, and exceptional groups of Lie type. Lemma 2.3 is the workhorse: it relates φ(|S|) to the number of Aut-classes of elements of order |S|, and it is applied carefully. The arithmetic in Propositions 4.2 and 5.3 is coherent, and the paper explicitly notes the n! growth of f. The literature review is fair; the claim that the single-prime version is new checks out against [15] and [28].\n\nThe soft spot is Section 5's reliance on external maximal subgroup classifications (Liebeck-Saxl-Seitz, Kleidman, Malle, Suzuki, Craven). If a row in Table 4 misstated a torus normalizer, Proposition 5.3 would fail for that family. This is a standard kind of dependence, and nothing in the rows looks suspect, but a referee should verify at least the E7 rows against Craven [7,8] and the E8 row against [24, Table 5.2]. There is also a harmless verbal overcount in Lemma 4.1: for PΩ^-(2d,q) with d even, the count should be d−1 nonidentity classes, but the identity supplies the missing class, so the lemma still holds.\n\nThe stress-test note's assessment matches my reading: no structural flaw, and the dependence on the literature is ordinary.\n\nThis paper is for finite group theorists working on Landau-type bounds, and for number theorists using class-count arguments. It deserves a serious referee.","headline":"New bound on |T| in terms of Aut-classes of p-elements; proof is detailed and likely correct, with the usual heavy reliance on external classifications.","tokens_in":21150,"tokens_out":42229,"would_cite":true,"duration_ms":357210,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20D05","20D06","20E45","20G40"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that the order of any finite nonabelian simple group is bounded above by an increasing function of the largest number, over all primes $p$, of $\\mathrm{Aut}(T)$-classes of elements of $p$-power order.","keywords":["finite simple groups","p-elements","Aut(T)-classes","conjugacy classes","order bounds","groups of Lie type","Singer cycles","maximal subgroups"],"falsifier":"Check the external fact the proof leans on most heavily: that the cyclic subgroup $H$ in the $E_8(q)$ row of Table 4 has normaliser $H.30$ maximal in $\\mathrm{Aut}(T)$. If that normaliser failed to be maximal, the field-size bound in Proposition 5.3 would fail for $E_8(q)$; verifying it against the cited maximal-subgroup table for $E_8(q)$ would confirm the most delicate exceptional case.","tokens_in":20283,"feed_emoji":"📏","tokens_out":14002,"duration_ms":117552,"temperature":0.7,"pith_summary":"This paper proves that the order of a finite nonabelian simple group is controlled by how many automorphism-classes of elements of prime-power order it has. For a simple group $T$, define $m(T)=\\max_p m_p(T)$, where $m_p(T)$ is the number of $\\mathrm{Aut}(T)$-classes of elements whose order is a power of $p$. The paper shows there is an increasing function $f$ with $|T|\\le f(m(T))$ for every finite nonabelian simple group $T$. Because $m(T)$ counts only $p$-power elements for a single best prime, it can be far smaller than the total number of conjugacy classes, so the result strengthens previous bounds expressed in terms of the full class count. A corollary is that every such $T$ has some prime $p$ with at least $g(|T|)$ $\\mathrm{Aut}(T)$-classes of $p$-power elements; the introduction also points to consequences for relative Brauer groups of finite extensions of global fields.","feed_headline":"A simple group's size is capped by its prime-power class count","feed_subtitle":"Fewer prime-power element classes means smaller simple groups: order is bounded by a computable function.","key_machinery":"The load-bearing object is a cyclic subgroup $H$ (often a Singer cycle, meaning a cyclic subgroup acting regularly on the nonzero vectors of the natural module, or a maximal torus) with known normaliser structure in $\\mathrm{Aut}(T)$. For a prime $s$ dividing $|H|$, let $S$ be the Sylow $s$-subgroup of $H$. Since $S$ is cyclic of order $s^b$, it contains $\\varphi(|S|)=s^{b-1}(s-1)$ elements of order $|S|$, and these split into exactly $\\varphi(|S|)/r$ $\\mathrm{Aut}(T)$-classes, where $r=|\\mathrm{N}_{\\mathrm{Aut}(T)}(S):\\mathrm{C}_{\\mathrm{Aut}(T)}(S)|$. The hypothesis $m(T)\\le n$ therefore forces $\\varphi(|S|)\\le rn$ and $\\varphi(|S|)\\mid r(n!)$, and because $r$ is controlled by the known normaliser, each prime divisor $s$ of $|H|$ is constrained. The same inequality bounds the number of such primes, and together with the divisibility identities in Lemma 2.1 this pins down $q$ and $a$. For exceptional groups the normaliser information comes from tables of maximal subgroups; for classical groups the dimension bound comes from counting unipotent classes.","core_discovery":"The central claim is Theorem 1.1: there exists an increasing function $f$ on the natural numbers such that, for every finite nonabelian simple group $T$, the order $|T|$ is at most $f(m(T))$, where $m(T)=\\max_p m_p(T)$ and $m_p(T)$ is the number of $\\mathrm{Aut}(T)$-classes of elements of $p$-power order. The proof splits the simple groups into alternating groups, classical groups of Lie type, and exceptional groups of Lie type, with the sporadic groups handled by the Monster. For alternating groups the argument uses the $\\lfloor m/3\\rfloor$ classes of elements of order $3$ in $\\mathrm{Alt}(m)$. For classical groups it combines a lower bound of $d$ unipotent classes, where $d$ is the dimension of the natural module, with analysis of cyclic tori and their normalisers to force bounds on the field size $q$ and the exponent $a$ in $q=p^a$. For exceptional groups the proof uses cyclic subgroups $H$ whose normaliser is a maximal subgroup of $\\mathrm{Aut}(T)$, listed in Tables 4 and 5, and shows that the prime divisors of $|H|$ force $q$ to be bounded.","pith_inferences":["A sharper $f$ is likely: the paper's proof uses a crude divisor-counting bound to control the number of prime divisors of $|H|$, and the authors note in Remark 4.3 that a better function may be possible; replacing the $n!$-type term by an exponential bound would make the finiteness threshold practical.","The same cyclic-subgroup normaliser strategy could plausibly extend from simple groups to almost simple groups, or from $m(T)$ to the $m_{\\mathrm{exp}}$ parameter for the exceptional groups, which the paper leaves open in Remark 4.4.","A computational census of $m(T)$ for simple groups of increasing order would show how fast the true optimal function grows; the theorem guarantees $m(T)\\to\\infty$ with $|T|$, but not whether the growth is logarithmic, polynomial, or something in between."],"forward_implications":["Corollary 1.2: there is an increasing function $g$ such that every finite nonabelian simple group $T$ has at least $g(|T|)$ $\\mathrm{Aut}(T)$-classes of elements of $p$-power order for some prime $p$ dividing $|T|$.","Because $f$ is increasing, for each fixed $n$ only finitely many nonabelian simple groups have $m(T)\\le n$; the theorem is thus a finiteness statement for simple groups with few $p$-element classes.","The bound generalises earlier results that used the total number of conjugacy classes, since $m(T)$ is never larger than the total class count and is often much smaller.","For simple classical groups the proof actually bounds $|T|$ by a function of $m'(T)=\\max\\{m_p(T),m_{S(T)\\text{-exp}}(T)\\}$, where the second term counts $\\mathrm{Aut}(T)$-classes of elements of maximal $s$-part of the exponent for a set of primes $S(T)$.","The introduction states that Theorem 1.1 is used in follow-up work to prove a case of a conjecture about subgroups meeting every $\\mathrm{Aut}(G)$-class of prime-power elements, with consequences for relative Brauer groups of finite extensions of global fields."],"supporting_citations":[{"why":"Supplies the maximal-rank subgroup classification for exceptional groups, giving the cyclic subgroups and normaliser structures used for the $E_7(q)$ and $E_8(q)$ rows of Table 4.","marker":"[24]"},{"why":"Classifies involutions in Chevalley groups of even characteristic; used in Lemma 4.1 to obtain the dimension lower bound for classical orthogonal groups.","marker":"[1]"},{"why":"Gives the maximal subgroups of $F_4(q)$, $E_6(q)$, and ${}^2E_6(q)$, providing the normaliser data used in Table 4 for these families.","marker":"[8]"},{"why":"Supplies the maximal-subgroup classification for $E_7(q)$ and related almost simple groups, cited for the $E_7(q)$ rows of Table 4.","marker":"[7]"},{"why":"Classifies the maximal subgroups of ${}^3D_4(q)$, justifying the cyclic subgroup and normaliser in the ${}^3D_4(q)$ row of Table 4.","marker":"[20]"},{"why":"Classifies maximal subgroups of $G_2(q)$ with $q$ odd and of ${}^2G_2(q)$, used for those rows and for the $G_2(q)$ case in Table 5.","marker":"[21]"},{"why":"Classifies the maximal subgroups of ${}^2F_4(q)$, used for the ${}^2F_4(q)$ row of Table 4.","marker":"[26]"},{"why":"Determines the doubly transitive groups underlying ${}^2B_2(q)$, cited for the ${}^2B_2(q)$ row of Table 4.","marker":"[35]"},{"why":"Describes the subgroup structure of finite classical groups; used to embed Singer cycles in orthogonal groups for the cyclic subgroups in Proposition 4.2.","marker":"[22]"},{"why":"Gives the normaliser of a Singer cycle in $\\mathrm{GL}(d,q)$, used to compute the index $r$ in the linear and unitary cases of Proposition 4.2.","marker":"[29]"}],"fun_headline_variants":["Simple groups: few p-power classes imply small order","Few p-power classes cap the order of simple groups","Simple group order bounded by prime-power element class count","Bounding simple group order via p-power Aut-classes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof for the exceptional families relies on previously published classifications asserting that certain cyclic subgroups have normalisers that are maximal subgroups of $\\mathrm{Aut}(T)$ with exactly the listed structure; if any one of those classification facts were wrong, the field-size bound for that family would collapse.","fun_headline_variants_meta":{"raw":{"variants":["Simple groups: few p-power classes imply small order","Few p-power classes cap the order of simple groups","Simple group order bounded by prime-power element class count","Bounding simple group order via p-power Aut-classes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000563,"raw_usage":{"total_tokens":2671,"prompt_tokens":945,"completion_tokens":1726,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":561,"completion_tokens_details":{"reasoning_tokens":1662}},"tokens_in":561,"tokens_out":1726,"duration_ms":11367,"temperature":1.0,"reasoning_tokens":1662,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T10:48:56.761575+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Check the external fact the proof leans on most heavily: that the cyclic subgroup $H$ in the $E_8(q)$ row of Table 4 has normaliser $H.30$ maximal in $\\mathrm{Aut}(T)$. If that normaliser failed to be maximal, the field-size bound in Proposition 5.3 would fail for $E_8(q)$; verifying it against the cited maximal-subgroup table for $E_8(q)$ would confirm the most delicate exceptional case.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the maximal-rank subgroup classification for exceptional groups, giving the cyclic subgroups and normaliser structures used for the $E_7(q)$ and $E_8(q)$ rows of Table 4."},{"cited_title":"Aschbacher and G","cited_arxiv_id":null,"evidence_quote":"Classifies involutions in Chevalley groups of even characteristic; used in Lemma 4.1 to obtain the dimension lower bound for classical orthogonal groups."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the maximal subgroups of $F_4(q)$, $E_6(q)$, and ${}^2E_6(q)$, providing the normaliser data used in Table 4 for these families."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Classifies the maximal subgroups of ${}^3D_4(q)$, justifying the cyclic subgroup and normaliser in the ${}^3D_4(q)$ row of Table 4."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Classifies maximal subgroups of $G_2(q)$ with $q$ odd and of ${}^2G_2(q)$, used for those rows and for the $G_2(q)$ case in Table 5."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Classifies the maximal subgroups of ${}^2F_4(q)$, used for the ${}^2F_4(q)$ row of Table 4."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Determines the doubly transitive groups underlying ${}^2B_2(q)$, cited for the ${}^2B_2(q)$ row of Table 4."},{"cited_title":"Kleidman, M","cited_arxiv_id":null,"evidence_quote":"Describes the subgroup structure of finite classical groups; used to embed Singer cycles in orthogonal groups for the cyclic subgroups in Proposition 4.2."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the normaliser of a Singer cycle in $\\mathrm{GL}(d,q)$, used to compute the index $r$ in the linear and unitary cases of Proposition 4.2."}],"review_version":1}