{"id":"71a9a55a-87d0-422a-8e0b-c1d96a040cad","arxiv_id":"2411.18495","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A proof that Coxeter elements have trace 0, 1, or -1 in every irreducible representation of a Weyl group, plus a Hecke algebra analogue.","lead":"For any Weyl group, the trace of a Coxeter element on any irreducible representation is 0, 1, or -1. This paper gives a proof of that statement, attributed to I.G. Macdonald, and proves a version for Iwahori-Hecke algebras.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Exceptional-type cases are delegated to unpublished table extraction and D_n relies on an unstated induction; both are verifiable proof gaps rather than demonstrated errors.","rationale":"The counting lemma 0.2 is sound and the classical A_n and B_n arguments are self-contained. The weakest point is precisely the delegation of exceptional Weyl groups to existing character tables and cell data, exactly as the reader's weakest_assumption states, together with the unstated induction behind 1.2(a) used in the D_n case. I agree that these are addressable proof-completeness gaps rather than evidence of falsehood; the paper is very likely correct. A direct computational check would settle whether the table extraction lands correctly. The reader's CONDITIONAL verdict already reflects this, so my stress-test does not move the verdict.","tokens_in":8176,"tokens_out":8834,"duration_ms":87282,"concrete_test":"Compute tr(w,E) for one Coxeter element w in each of E6, E7, E8, F4, G2 using CHEVIE/GAP character tables, and compare the nonzero values with the coefficients in 2.10–2.14 and with the cell counts in 2.6; simultaneously verify for D_n (n=4,...,8) by explicit branching that the M_i and 'M_i modules are distinct and have traces (-1)^i, and that all other irreps have trace 0. Agreement confirms the gap is expository; any mismatch falsifies 0.1(a).","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim 0.1(a) is established by 0.2 once h irreps with trace ±1 are exhibited. For exceptional Weyl groups the paper does not exhibit such an h-tuple: §2.6(c) asserts that only cells of the form c_i and the exceptional cells contribute, and that in the exceptional types this 'follows by examining the existing tables'. The aggregate sums E_{c_i}(W) in 2.10–2.14 are not derived, and the reader is left to reconstruct the individual traces and their completeness. A single misread table entry or an error in the cell parametrization from [L84] would invalidate the corresponding case of 0.1(a). Similarly, the D_n case relies on 1.2(a), whose induction is only sketched; without it the 'M_i modules are not shown to have trace (-1)^i. Both are proof-completeness gaps, not demonstrated errors: the theorem is very likely true.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves Macdonald's conjecture that for an irreducible complex representation E of a Weyl group W, the trace of a Coxeter element w on E belongs to {0,1,-1}. The proof uses a counting argument: if one can exhibit h distinct irreducible representations on which w has trace ±1, where h is the order of w and also the order of its centralizer, then the orthogonality relations force all other traces to vanish. The classical types A, B, and D are treated by explicit constructions from exterior powers and induction. The exceptional types are handled by asserting that certain cell sums, listed in §2.10–2.14, contain all contributing irreducibles. Section 3 proves an Iwahori-Hecke algebra analogue: tr(T_w,E(v)) equals tr(w,E) v^{m_E} for some nonnegative integer m_E, hence is either 0 or a monomial with coefficient ±1. Section 4 shows that the analogous statement fails for noncrystallographic Coxeter groups, with explicit counterexamples for H3, H4, and dihedral groups.","tokens_in":8289,"tokens_out":10509,"duration_ms":95870,"significance":"If the proof is completed, the paper settles a fifty-year-old unpublished conjecture of I.G. Macdonald in a clean and conceptual way. The counting reduction in §0.2 is elegant and rigorous, and the classical-type constructions are explicit and verifiable. The Hecke algebra analogue in §3 is a natural strengthening and the noncrystallographic counterexamples in §4 provide useful boundary information. The paper also draws a interesting connection to two-sided cells and special representations. The main gaps are completeness issues: the induction in §1.2 is only sketched and the exceptional-type verification in §2.6(c) is delegated to unnamed table inspections. These gaps are fillable, and the mathematical claims are very likely correct, but the written proof is not yet fully self-contained at these load-bearing points.","major_comments":[{"comment":"The proof of assertion (a) in §1.2 is a single sentence: 'This is proved by induction on n, using the results in 1.1 for S_{n−1} and the known results about restricting an E ∈ Irr(S_n) to S_{n−1}.' The first part, tr(w', Λ'_i) = (−1)^i, is load-bearing because §1.4 uses it to compute tr(w, M'_i) for type D_n. The induction step is not written out: one must show that the (n−1)-cycle acts on the specified Specht modules with the stated signs, and the branching argument is not immediate from §1.1 alone. The vanishing claim for all other irreducibles in §1.2(a) is also stated without proof. Please supply the full induction argument, including the branching rules, or give a precise reference that contains this exact statement.","section":"§1.2(a), used in §1.4"},{"comment":"For the exceptional Weyl groups, Proposition 2.6(c) asserts that no cells other than the c_i and the exceptional cells contribute to E(W), and the lists in §2.10–2.14 describe the corresponding cell sums. The proof says only that in the exceptional case 'this follows by examining the existing tables'. The paper does not tabulate the individual traces tr(w,E) or the cell u(E) for each irreducible used, and it does not identify the specific tables and labeling conventions that would allow a reader to verify the computation. Because the counting argument in §0.2 requires exactly h irreducible representations with trace ±1, the completeness of these lists is essential to the proof of 0.1(a) for E6, E7, E8, F4, and G2. This is a proof-completeness gap. Please provide a verifiable derivation: for instance, a full table of (E, u(E), tr(w,E)) for each exceptional group, or a reproducible computation using Chevie, GAP, or another package, together with a statement of the exact reference used for the character tables.","section":"§2.6(c) and §2.10–2.14"}],"minor_comments":[{"comment":"There is a typo: 'Coxeter elememt' should be 'Coxeter element'.","section":"Abstract and §0.1"},{"comment":"The phrase 'free C-vector spacee' contains a typo; it should be 'vector space'.","section":"§2.5"},{"comment":"In the line 'If Γ(c_i) = S_2 and W is simply laced then p_i = 0. (We use that 1/2 − 1/2 = 1.)', the arithmetic statement is false: 1/2 − 1/2 = 0, not 1. The intended identity is presumably '1/2 − 1/2 = 0', which matches the asserted value p_i = 0.","section":"§2.16"},{"comment":"In the description of the partition for Λ^i, the notation '1 ≤ 1 ≤ 1 ≤ · · · ≤ 1 ≤ n − i' is ambiguous because the number of leading 1's is not specified. Writing the partition as (n−i, 1^i) would remove the ambiguity.","section":"§1.1"},{"comment":"The assertion that tr(T_y,E(v)) ∈ C[v] for all y ∈ W is used without a reference. Please cite the precise result, for example the relevant statement in [KL] or [L84, Chapter 3].","section":"§3.1"}],"recommendation":"major_revision","confidential_remarks":"The two major gaps are proof-completeness issues rather than demonstrated errors. The classical-type arguments and the counting reduction in §0.2 are sound, and the exceptional-type claims are almost certainly correct, but the text as written does not give the reader enough to verify the exceptional cases or the §1.2 induction. These are fillable, and I expect the authors can supply the needed details without changing the main claims. I do not recommend rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a clean, believable proof of Macdonald's old assertion, plus a Hecke algebra analogue. The main content is the counting lemma 0.2, which is elegant and correct, and the explicit trace computations for classical types. As a proof of 0.1(a) it is essentially complete modulo two gaps: the D_n case leans on 1.2(a), whose induction is only sketched, and the exceptional types are delegated to 'examining existing tables' without showing the extraction. Neither gap looks fatal—the theorem is very likely true, and both are verifiable—but as written they are real dents in completeness.\n\nWhat's actually new: a first written proof of Macdonald's statement, and the Hecke algebra version 3.1(a), which is a genuinely useful strengthening. The argument around 3.2 using T_w^h = T_{w0}^2 plus [L84,5.12.2] is slick and gives the monomial form. Section 2's cell-by-cell decomposition is nice and ties the trace statement to cell theory; the explicit lists 2.7-2.14 are a service even if they come from tables. The noncrystallographic counterexample section is a good sanity check.\n\nSoft spots in proportion:\n- 1.2(a) is asserted to follow by induction from 1.1 and restriction to S_{n-1}. That is plausible, but no details are given. The D_n construction of 'M_i depends on it, and the trace formula for 'M_i inherits the gap. This is the weakest spot in the classical part.\n- For exceptional types, Proposition 2.6(c) and the sums 2.10-2.14 are taken from existing tables. That is standard practice for finite type checks, and the tables are independent of the theorem, so not circular. But the paper does not show the individual traces or the completeness check, so a skeptical reader has to redo the table work to certify the case. That is an addressable presentation gap, not a mathematical error.\n- The Hecke argument assumes [L84,5.12.2] about the involution on E(v); that's an established result, and the trace specialization at v=1 correctly recovers 0.1(a). No circularity.\n\nOverall: I trust the result. It deserves a serious referee; the proof is mostly there and the gaps are fillable. A referee should ask for more detail on 1.2(a) and a more explicit description of the exceptional-type table extraction, then accept.","headline":"A believable first written proof of Macdonald's trace claim plus a Hecke analogue; two verifiable proof gaps (D_n induction, exceptional table extraction) keep it from being fully polished.","tokens_in":8857,"tokens_out":1637,"would_cite":true,"duration_ms":14361,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20F55","20C15","20C08"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that for any irreducible Weyl group, the trace of a Coxeter element on any irreducible complex representation is one of 0, 1, or -1, and extends this to a statement about Iwahori-Hecke algebras.","keywords":["Coxeter element","Weyl group","irreducible representation","trace","Iwahori-Hecke algebra","two-sided cells","exterior powers","Macdonald conjecture"],"falsifier":"Independently compute $\\operatorname{tr}(w,E)$ for every irreducible representation of $E_8$ (or $F_4$) using a direct construction of the representations and a Coxeter element, and check that every value lies in $\\{0,1,-1\\}$; the paper's 2.12 and 2.13 lists give the expected nonzero traces, so any coefficient other than $0$, $+1$, or $-1$ would disprove the claim.","tokens_in":7914,"feed_emoji":"🧮","tokens_out":6035,"duration_ms":51552,"temperature":0.7,"pith_summary":"This paper proves a claim about Coxeter elements in Weyl groups that I. G. Macdonald stated in the early 1970s but left unpublished: for any irreducible Weyl group $W$, any Coxeter element $w$ (a product of the simple reflections in some order), and any irreducible complex representation $E$ of $W$, the trace $\\operatorname{tr}(w,E)$ is one of the three numbers $0$, $1$, $-1$. The proof covers the classical families $A$, $B$, $C$, $D$ by direct counting arguments, and the exceptional Weyl groups $E_6$, $E_7$, $E_8$, $F_4$, $G_2$ by checking the known character tables. The paper also proves a Hecke-algebra refinement: in the Iwahori-Hecke algebra, the trace of the basis element $T_w$ on $E(v)$ is either $0$ or a single monomial $\\pm v^{m_E}$, with $m_E$ a nonnegative integer. The noncrystallographic finite Coxeter groups ($H_3$, $H_4$, dihedral groups) do not satisfy the analogue, so the phenomenon is specific to Weyl groups.","feed_headline":"For Weyl groups, Coxeter-element traces are 0, 1, or -1","feed_subtitle":"A long-standing claim of Macdonald now has a proof for classical and exceptional Weyl groups, with a Hecke analogue.","key_machinery":"The load-bearing object is the Coxeter element $w$, the product of all simple reflections of $W$ in a fixed order; its order $h$ is the Coxeter number. The counting identity that carries the argument is the centralizer fact: the centralizer of $w$ in $W$ is cyclic of order $h$, so by the orthogonality of characters the squares of $\\operatorname{tr}(w,E)$ over all irreducibles $E$ sum to $h$. If one can exhibit $h$ irreducibles with trace $\\pm1$, all other traces must vanish. The paper exhibits those irreducibles via the exterior powers $\\Lambda^i$ of the reflection representation, whose Coxeter traces are $\\pm1$, together with further representations in types $B$ and $D$ built from symmetric-group representations (the symbols in [L84]). For the Hecke version, the key identity is $T_w^h = T_{w_0}^2$, valid for a suitable choice of $w$, together with the known involution on $E(v)$ that acts like a scalar times $v^{\\nu-a_E+a_{E^!}}$ on $T_{w_0}$.","core_discovery":"The central assertion, stated as 0.1(a), is that if $W$ is an irreducible Weyl group, $w$ is a Coxeter element, and $E$ is an irreducible complex representation of $W$, then $\\operatorname{tr}(w,E)$ lies in $\\{0,1,-1\\}$. The argument splits according to type. For $S_n$, $B_n$, and $D_n$, the paper exhibits exactly $h$ irreducible representations with trace $\\pm1$, where $h$ is the order of $w$, and then uses the fact that the centralizer of $w$ in $W$ is cyclic of order $h$: since the sum of the squares of all traces $\\operatorname{tr}(w,E)$ equals $h$, any representation not among the exhibited ones must have trace $0$. For the exceptional types, the conclusion is read off the known character tables. The paper also proves the Iwahori-Hecke version 3.1(a): $\\operatorname{tr}(T_w, E(v)) = \\operatorname{tr}(w,E)\\, v^{m_E}$ for some $m_E \\in \\mathbb{N}$, so the Hecke trace is zero or a signed monomial; the proof uses the known involution on $E(v)$ and the identity $T_w^h = T_{w_0}^2$, which forces eigenvalues of $T_w$ to be $h$-th roots of unity times a fixed power of $v$.","pith_inferences":["The centralizer-counting method is robust enough that one could test it on other elements of Weyl groups, or on other finite reflection groups, by looking for a set of $h$ irreducibles with trace $\\pm1$; the paper's negative results for $H_3$, $H_4$, and dihedral groups show where that search fails.","Because the trace is either $0$ or a sign, the Coxeter element acts as a kind of membership test: it distinguishes the cells that carry the 'Coxeter bit' in the cell decomposition from those that do not, a role reminiscent of unipotent representation theory.","The Hecke-algebra monomiality may have consequences for the zeros and poles of Kazhdan-Lusztig polynomials attached to Coxeter elements, though the paper does not develop this direction.","One could try to make the exceptional-type verification independent of the published tables by computing the traces directly from a computer construction of the representations; that would remove the reliance on table completeness."],"forward_implications":["For every irreducible Weyl group, the $h$ characters with nonzero Coxeter trace are precisely the ones listed in Section 1; all others have trace $0$, giving an extremely sparse character-value distribution.","The nonzero traces align with the two-sided cell decomposition: contributions occur only on the cells $c_i$ attached to the exterior powers $\\Lambda^i$ and on the exceptional cells, with the number of contributing irreducibles equal to $1, 2, 4, 6, 10$, or $2$ depending on the cell.","In the Iwahori-Hecke algebra, the element $T_w$ acts on each irreducible $E(v)$ with trace that is either zero or a single signed monomial $\\pm v^{m_E}$, so there is no cancellation in that trace.","For a nonzero trace, the exponent $m_E$ is given explicitly by Proposition 3.3: if $E \\in X_{c_i}$ then $m_E = 2r - 2i$, where $r$ is the rank of $W$.","The noncrystallographic analogue fails, so the $\\{0,1,-1\\}$ bound singles out Weyl groups among finite Coxeter groups."],"supporting_citations":[{"why":"Supplies the parametrization of irreducible representations of Weyl groups by symbols, the two-sided cell data, the nonabelian Fourier transform, and the involution on $E(v)$ used in Section 3.","marker":"[L84]"},{"why":"Introduces two-sided cells and the Kazhdan-Lusztig theory that organizes the nonzero-trace representations in Section 2.","marker":"[KL]"},{"why":"Provides the pattern of unipotent representations of finite Chevalley groups associated to Coxeter orbits, which the cell sums in 2.7-2.14 mirror via the nonabelian Fourier transform from [L84,4.14].","marker":"[L76]"},{"why":"Defines the exceptional two-sided cells $c_{7/2}$ and $c_{9/2}$ used to describe the exceptional contributions in 2.4 and in the noncrystallographic analogue.","marker":"[L17]"}],"fun_headline_variants":["Proof: Coxeter element traces in Weyl groups are 0, 1, or -1","Weyl group Coxeter traces: 0, 1, -1 proven","Coxeter element trace theorem: 0, 1, -1 for Weyl groups","Macdonald's Coxeter trace claim proven with Hecke analogue"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the known character tables and the two-sided cell parametrization for the exceptional Weyl groups are complete and correct, and that the unstated induction behind 1.2(a) for type $D_n$ is valid; if any of these fails, the corresponding case of the theorem collapses.","fun_headline_variants_meta":{"raw":{"variants":["Proof: Coxeter element traces in Weyl groups are 0, 1, or -1","Weyl group Coxeter traces: 0, 1, -1 proven","Coxeter element trace theorem: 0, 1, -1 for Weyl groups","Macdonald's Coxeter trace claim proven with Hecke analogue"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001318,"raw_usage":{"total_tokens":5345,"prompt_tokens":901,"completion_tokens":4444,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":517,"completion_tokens_details":{"reasoning_tokens":4353}},"tokens_in":517,"tokens_out":4444,"duration_ms":30971,"temperature":1.0,"reasoning_tokens":4353,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T11:09:06.465625+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Independently compute $\\operatorname{tr}(w,E)$ for every irreducible representation of $E_8$ (or $F_4$) using a direct construction of the representations and a Coxeter element, and check that every value lies in $\\{0,1,-1\\}$; the paper's 2.12 and 2.13 lists give the expected nonzero traces, so any coefficient other than $0$, $+1$, or $-1$ would disprove the claim.","supporting_citations":[],"review_version":1}