claude-fable-5-2 high: cf-braids
Reviewer record, not the publication verdict. The full response includes unaccepted proposals and source limitations. The curated report states what was adopted. Initial packets used a conservative 2010 cutoff; focused packets used the verified 2012 journal received date. The focused Fable answer needed a metadata-only derivative; its original failure and partial source access remain recorded.
Completed visible assessment, preserved as a readable check record. It is a model assessment, not a formal certificate. This document contains no private reasoning stream.
No false theorem was found in the assigned scope. Theorem 4.1 holds with its stated range n>=4i. Conditions I and II are proved directly, and Condition III' follows from the Orlik-Solomon/Lehrer-Solomon decomposition together with Theorem 4.2. A hand computation with the Pieri/Young rule reproduces all displayed decompositions of H^1 and H^2. The range is slightly loose for i>=2 (n>=4i-1 suffices by the same method), consistent with the manuscript's own 'close to sharp'.
The main proof-interface weakness is that the multiplicity part of Theorem 4.2 is wholly imported from Hemmer. The supplied source dates that paper only to 2011, after the 2010-08-07 fallback cutoff, so its chronological admissibility is unresolved. The statement is nevertheless elementary, and I supply a separate self-contained repair (staged induction plus the Pieri rule) that also gives the exact threshold.
Corollary 4.4 is correct as a dimension count, but it needs the extra hypothesis n>=|lambda|+lambda_1 for V(lambda)_n to exist, and it proves abstract isomorphism rather than stability of induced maps. Theorem 4.6 is correct. Its definition of X'_n has a typographical error (C^{2n} for C^n), its structure maps are unspecified and its injectivity step is only sketched; all are repairable in a line. Corollary 4.7 additionally needs self-duality of W_n-irreducibles. Conjecture 4.8 is a conjecture at the cutoff; the footnoted later proof is retrospective only.
Outside the assigned lines, the Introduction's phrase 'S_n acts freely on C^n' is inaccurate (the action is free on X_n), with no consequence.
Disclosure: I cannot remove later knowledge (FI-module theory, Wilson's work) from my training. None of it is used to justify any step above. No web search, software or external paper was consulted.
B1
Checked statement: Theorem 4.1 (label thm:pure): for each fixed i>=0 the consistent sequence of S_n-representations H^i(P_n;Q), with maps induced by forgetting the last strand, is uniformly representation stable (Conditions I, II, III' of Section 2.3) and stabilizes once n>=4i.
Returned assessment: no_issue_found
The theorem and its range hold, conditional on the classical inputs listed below and on the multiplicity part of Theorem 4.2, which I re-derived by hand (see B2). Checked step by step.
Consistency: forgetting the last point commutes with S_n, so the maps are S_n-equivariant.
Condition I: Arnol'd's basis for P_n maps to a subset of the basis for P_{n+1}, so the map is injective for all n.
Condition II: a degree-i monomial uses at most 2i indices. For n>=2i (so 2i<n+1) a permutation in S_{n+1} moves it to a monomial avoiding index n+1, which lies in the image of H^i(P_n). The permuted monomial need not be a basis element, but that is not needed.
Condition III': the decomposition of H^*(P_n;Q) over set partitions S is the Brieskorn/Orlik-Solomon decomposition by flats of the braid arrangement. The localisation at the flat of S is the product of braid arrangements on the blocks, so H^S is the image of the top cohomology of P_S and lies in degree n-|S|. S_n permutes the summands transitively within each type mu, which gives eq:mupiece. The count 'some part equals 1 once n>=2i' is right, since n+1-i parts all >=2 would force n+1<=2i. The map H^S(P_n) -> H^{S<n+1>}(P_{n+1}) is an isomorphism because the projection to P_S x P_1 factors through P_n. The S_{m_1} factor permuting singletons fixes every w_{jk} with j,k in a common non-singleton block, so it acts trivially. Hence for n>=2i, H^i(P_n) is the direct sum, over the finitely many partitions nu with no part 1 and |nu|-l(nu)=i, of Ind from H_nu x S_{n-k} to S_n of (top cohomology of P_{S_nu}) tensor trivial, with k=|nu|=i+l(nu)<=2i. Theorem 4.2 with range n>=2k then gives constancy for n>=4i.
Undefined-label clause of III': every constituent V(eta) has |eta|<=k<=2i and eta_1<=2i, so V(eta)_{4i} is defined.
Hand cross-check using only the Pieri/Young rule: i=1 gives V(0)+V(1)+V(2) exactly from n=4. For i=2 the two pieces are nu=(3), with Ind = V_(2,1), and nu=(2,2), with Ind = V_(3,1) by a character check on a transposition. These reproduce every line of eq:stabilizationinaction for n=4,5,6 and eq:braid2stab for n>=7. The dimension at n=4 is 11, as it should be.
Premises used
- Arnol'd [Ar]: H^*(P_n;Q) is generated by the classes w_{jk} modulo the three-term relations, with the stated monomial basis. available_from: 1969. date_evidence: manuscript bibliography entry (Math. Notes 5 no. 2, 1969). date_basis: journal year as printed in CF-v3-TeX; not independently verified. version: journal article. hypotheses used: rational coefficients, all n>=1.
- Orlik-Solomon [OS], cited as Proposition 2.10, together with Lehrer-Solomon [LS]: S_n-equivariant decomposition of the cohomology of an arrangement complement over the intersection lattice, each summand being the top-degree part for the localised arrangement. available_from: 1980 (OS), 1986 (LS). date_evidence: bibliography entries. date_basis: journal years as printed. version: journal. The proposition number is unverified because the article is not supplied; only the braid-arrangement case over Q is used.
- Theorem 4.2 (label thm:hemmer), used only for H=Stab(S_nu)<S_k, V the finite-dimensional top cohomology of P_{S_nu}, k<=2i, and only for multiplicity stability with range n>=2k. Chronology of [He] is unresolved; see B2, where the needed statement is re-derived from the Pieri/Young rule.
- Definitions of consistent sequence, Conditions I, II, III' and padded partitions lambda[n] (Sections 2.1 and 2.3 of the manuscript).
Missing inputs
- Text of [OS] and [LS] to confirm the cited proposition and equation numbers.
- arXiv v1 text (7 Aug 2010), to confirm whether the n>=4i range and the word 'uniformly' were already present; the reviewed text is v3.
- A dated record showing that Hemmer's result was available on or before 2010-08-07.
Scope changes
- The multiplicity range was checked through a self-contained Pieri/Young-rule argument rather than by reading [He]. This is a new verification, kept separate from the manuscript's proof.
- The historical v1 text was not available. The audit is of the v3 wording against a 2010-08-07 external-results cutoff.
Located concern
proof of thm:pure, definition of S_mu and the displayed decomposition of H^i(P_{n+1};Q), within lines 1892-2211
Minor index typos, non-blocking. The last block of S_mu is written with mu_{n-1} where mu_{l(mu)-1} is meant, and the sum over nu |- n+1 carries the condition 'l(mu)=n+1-i' instead of l(nu)=n+1-i. The bijection with partitions of i is with partitions of i having at most n-i parts, so the index set is n-independent only for n>=2i.
Direct reading. For n<2i a partition of i with more than n-i parts has no preimage.
Correct the indices and say 'for n>=2i'.
None.
Located concern
paragraph after eq:braid2stab ('this is false for H^2 and indeed is false for H^i for every i>=1'), and the Introduction's 'this coherence never happens'
The failure of type-preservation is demonstrated only for i=1. For i>=2 it is asserted without proof.
The i=1 computation is correct. The image of the invariant vector is (sum of all edges) minus (star at vertex n+1). Its S_{n+1}-translates span an (n+1)-dimensional space, because (1/2)J - I is invertible on Q^{n+1} for n+1>2, and that space is V(0)+V(1). No argument is given for i>=2, where V(0) does not occur at all.
Either restrict the remark to i=1 or supply an argument for general i.
None on Theorem 4.1; this is an unproved side remark.
Located concern
thm:pure statement and final paragraph of its proof; Introduction sentence 'the stable range ... is close to being sharp'
Loose range (not an error). The bound n>=4i is valid but not sharp for i>=2.
The Pieri/Young rule shows that a constituent V_lambda of Ind to S_k stabilises exactly at n>=k+lambda_1. For nu=(2^i) the stabiliser acts on the top class through the sign of the block permutation, so for i>=2 the trivial representation of S_{2i} does not occur and lambda_1<=2i-1. All other nu have k<=2i-1. Hence n>=4i-1 suffices for i>=2, and for i=2 the exact threshold is 7, as in eq:braid2stab. For i=1 the bound 4 is sharp. This refinement is my own derivation, not part of the manuscript.
None needed; optionally state the refinement.
None; the stated theorem stands.
Located concern
Introduction (eq:stabilizationinaction, 'using work of Lehrer-Solomon and the computer program Magma, we computed') versus Section 4 ('thanks to computations by Hemmer')
Inconsistent attribution of the H^2 computation.
The two passages credit different sources for the same formula. The formula itself was reproduced by hand above.
Harmonise the attribution.
None.
B2
Checked statement: Theorem 4.2 (label thm:hemmer, attributed to Hemmer): for fixed k, H<S_k and an H-representation V, the sequence Ind from H x S_{n-k} to S_n of V tensor trivial is uniformly representation stable and stabilizes once n>=2k. Item 4.3 (an unlabelled Remark, not a corollary). Corollary 4.4 (label corollary:twistedstability): H_i(B_n;V(lambda)n) is isomorphic to H_i(B{n+1};V(lambda)_{n+1}) once n>=4i.
Returned assessment: repairable_gap
Numbering. With the section-based shared counter, 4.1 = thm:pure, 4.2 = thm:hemmer, 4.3 = a Remark citing [LS] and [CEF2], 4.4 = corollary:twistedstability, 4.5 = a Remark. So 'Corollaries 4.3-4.4' contains only one corollary. The Remark 4.3 makes no claim needing proof; it points to [CEF2], listed as 'in preparation' and unavailable.
Theorem 4.2, Conditions I and II. The manuscript's argument is correct. By Mackey, the S_n-span of the canonical copy of V inside the induction to S_{n+1} is induced from the intersection of S_n with H x S_{n+1-k}, which is H x S_{n-k}. So the natural equivariant map is injective, and its S_{n+1}-span is everything.
Theorem 4.2, multiplicities. The manuscript gives no proof and defers entirely to [He]. The bibliography dates [He] to 2011, after the 2010-08-07 cutoff, and the source documents no earlier preprint, so chronological admissibility is unresolved. The statement can be closed without [He] by the following new, self-contained argument. Induce in stages through S_k x S_{n-k} and decompose Ind from H to S_k of V into irreducibles V_lambda with lambda |- k. By the Pieri/Young rule, Ind of V_lambda tensor trivial is the sum of V_nu over nu |- n with nu/lambda a horizontal strip. Writing nu = eta[n], the multiplicity of V(eta)_n is 1 exactly when lambda_1>=eta_1>=lambda_2>=eta_2>=... and n>=|eta|+lambda_1, and 0 otherwise. Since |eta|<=k with equality at eta=lambda, the decomposition is constant exactly from n=k+lambda_1<=2k on. The bound 2k is attained if and only if V contains the trivial H-representation. The clause of III' on undefined labels holds at N=2k.
Corollary 4.4. The proof is correct as a dimension count. Transfer over Q identifies H_i(B_n;V) with the S_n-invariants of H_i(P_n;Q) tensor V, with diagonal action, since P_n acts trivially on V. By self-duality of S_n-representations and Schur's lemma, the dimension is the multiplicity of V(lambda)_n in H_i(P_n;Q), which equals that in the dual H^i(P_n;Q). Theorem 4.1 gives constancy for n>=4i. Only Condition III' and the range of Theorem 4.1 are used; Conditions I and II are not. Two defects of statement remain, listed below.
Premises used
- Hemmer [He], 'Stable decompositions for some symmetric group characters arising in braid group cohomology'. Exact statement used: multiplicity stability of Ind from H x S_{n-k} to S_n of V tensor trivial, with range n>=2k. available_from: unresolved. date_evidence: the bibliography gives J. Combin. Theory Ser. A 118 (2011), which is after the cutoff; the v3 Introduction says the authors conjectured it to Hemmer 'who then proved it', which suggests but does not document availability by 2010-08-07. date_basis: manuscript bibliography only. version: unknown preprint, if any. I may recall a 2010 preprint from training data but cannot verify it here and do not rely on it.
- Pieri/Young rule for inducing V_lambda tensor trivial from S_k x S_{n-k} to S_n (classical; contained in Fulton-Harris [FH], 1991, per the manuscript's bibliography; exact item number not verified). available_from: no later than 1991. date_basis: book year as printed. Used only in the new verification.
- Transfer isomorphism for a finite-index normal subgroup with rational coefficients (classical; no citation in the manuscript).
- Self-duality of S_n-representations (stated in Section 2.1 of the manuscript) and Schur's lemma.
- Theorem 4.1 (thm:pure): only uniform multiplicity stability of H^i(P_n;Q) for n>=4i.
Missing inputs
- A dated preprint or record of [He] on or before 2010-08-07, with the exact statement and range it contained at that date.
- Text of [CEF2], cited in Remark 4.3 as in preparation.
- Journal-version comparison (pending by instruction).
Scope changes
- Target reinterpreted as Theorem 4.2, Remark 4.3 and Corollary 4.4.
- Hemmer's own proof was not inspected (not supplied); an independent derivation is substituted and labelled as a new repair.
Located concern
thm:hemmer, sentence 'For the proof of uniform multiplicity stability, see Hemmer [He]'
The multiplicity part and the range n>=2k are entirely imported, and the imported source's availability by the cutoff is undocumented.
The bibliography lists [He] as a 2011 journal article; no preprint identifier or date is given in the source.
New self-contained repair, not the original proof: the staged-induction and Pieri/Young-rule argument in the 'argument' field, which uses only classical pre-1991 facts and yields the exact threshold n>=k+lambda_1.
The statement of Theorem 4.2 is correct. The historical proof obligation is unresolved as to chronology but is closed mathematically by the repair.
Located concern
corollary:twistedstability, 'For any partition lambda ... once n>=4i there is an isomorphism' (eq:twisted)
Quantifier/definedness gap. V(lambda)_n is defined only for n>=|lambda|+lambda_1 (Section 2.1), and this can exceed 4i.
Take i=0 and lambda=(1): the statement quantifies over n=0 and n=1, where V(1)_n is not defined. Take i=1 and lambda=(3): 4i=4 but V(3)_n needs n>=6.
State the isomorphism for n>=max(4i, |lambda|+lambda_1). Uniform stability with N=4i then gives it at once.
Statement needs a one-line hypothesis; conclusion otherwise stands.
Located concern
corollary:twistedstability, phrase 'satisfies classical homological stability', its proof's first sentence 'There are no natural maps ...', and the Remark numbered 4.5
Type-of-stability mismatch. The Introduction defines homological stability as the induced maps being isomorphisms; the corollary proves only equality of dimensions, that is, an abstract isomorphism.
The proof says so explicitly ('we show that their dimension is eventually constant'). The cohomology maps of Theorem 4.1 come from P_{n+1}->P_n, which does not descend to the braid groups.
Reword as 'the groups are abstractly isomorphic (have equal dimension)', or construct stabilisation maps and prove they are isomorphisms.
The isomorphism in eq:twisted is established; the stronger reading with maps is not proved here.
Located concern
packet description 'Corollaries 4.3-4.4'
Item 4.3 is a Remark, not a corollary.
Shared counter: thm:pure 4.1, thm:hemmer 4.2, Remark 4.3, corollary:twistedstability 4.4.
Refer to 'Remark 4.3 and Corollary 4.4'.
None.
B3
Checked statement: Theorem 4.6 (label thm:genpure): for each fixed i>=0 the sequence of W_n-representations H^i(WP_n;Q) is uniformly representation stable (no range stated). Corollary 4.7 (label cor:gen): for each double partition lambda and fixed i, H_i(B_{1,n};V(lambda)n) is isomorphic to H_i(B{1,n+1};V(lambda)_{n+1}) for n sufficiently large depending on i. Conjecture 4.8 (label conjecture:string): H^i(PSigma_n;Q) is uniformly representation stable as W_n-representations; this is a conjecture, not a theorem of the paper.
Returned assessment: repairable_gap
Theorem 4.6. The mathematical argument is sound; the defects are in the literal definition and the proof interface.
Flats and orbits: flats of the type-B arrangement are given by a zero block of size l>=0 and blocks of coordinates equal up to sign. The W_n-orbits are indexed by (lambda, l) with lambda having no part 1, and codim = |lambda|-l(lambda)+l. The example ((3,3),2) has m=8, codim 6, as stated.
Stabiliser: it is G_J x W_{n-m}. The group written 'S_3 wr W_2' is (S_3 x Z/2) wr S_2, which is the correct stabiliser of the two size-3 blocks. W_{n-m} fixes every hyperplane containing J, hence acts trivially on H^J.
Decomposition: dependent monomials vanish, so H^J sits in degree codim(J), and H^k is the sum over orbit representatives of induced representations. For fixed k one has |lambda|<=2(k-l) and m<=2k, so the index set is finite and n-independent for n>=2k.
Induction: staged induction through W_m x W_{n-m} and the hyperoctahedral Littlewood-Richardson rule give Ind of V_(lambda+,lambda-) tensor trivial as the sum of V_(nu,lambda-) over horizontal strips nu/lambda+ of size n-m. Adding a box to the first row is a bijection between consecutive decompositions as soon as n-m>=lambda+_1, because at most lambda+_1 boxes of a horizontal strip lie below the first row. The manuscript's sufficient condition n-m>|lambda+| is valid.
Range (my own extraction, not in the manuscript): every piece stabilises once n>=m+lambda+1<=2m<=4i, and the undefined-label clause holds at N=4i. For i=1 the threshold 4 is attained by the constituent V((2),0) coming from the hyperplanes z_a=+-z_b.
Conditions I and II: surjectivity is correct for n>=2i. Injectivity is only sketched ('naturality of these relations') but is true.
Corollary 4.7. 'Same argument as Corollary 4.4' works. W_n acts freely on X'n, so WP_n is normal of index |W_n| in WB_n = B{1,n}. Transfer plus Schur gives dimension = multiplicity of the dual of V(lambda)_n in H_i(WP_n;Q). This needs self-duality of W_n-irreducibles, which the manuscript states only for S_n. Because the statement says 'sufficiently large n', the definedness problem of Corollary 4.4 does not arise. As in 4.4, only an abstract isomorphism is proved, and coefficients are rational throughout.
Conjecture 4.8. It is stated as a conjecture and motivated by analogy only; no proof is offered or claimed. The footnote reports a later proof by Wilson [Wi1] (2012), which is after the cutoff and is retrospective context only. The surrounding checks pass: under alpha_ij -> w_ij the relation alpha_ij alpha_ji maps to w_ij^2=0 and the three-term relation maps to Arnol'd's relation, so the surjection onto H^*(P_n;Q) is consistent; and the remark that the W_n-action does not descend is correct.
Premises used
- Brieskorn [Bri] and Orlik-Solomon [OS, Theorem 5.2 and Proposition 2.10 as cited]: generation of H^*(X'_n;Q) by the logarithmic classes w_H, and decomposition over the intersection lattice, permuted by W_n. available_from: 1973 and 1980. date_evidence: bibliography entries. date_basis: years as printed. version: published. Item numbers unverified; only the type-B arrangement over Q is used.
- Lehrer-Solomon [LS, Equation 2.2 as cited]: dependent monomials vanish. available_from: 1986. date_basis: bibliography. Equation number unverified.
- Geck-Pfeiffer [GP, Lemma 6.1.3 as cited]: induction from W_m x W_{n-m} to W_n multiplies by Littlewood-Richardson coefficients componentwise. available_from: 2000. date_basis: bibliography year. Lemma number unverified; the manuscript cites the same lemma number for the W_n branching rule in Remark remark:strong.
- Pieri rule (classical, [FH] 1991) for the coefficient C^{nu}_{lambda+,(n-m)}.
- Douglass [Do] (1992) and Vershinin [Ve] (1999): exposition and the identification WB_n = B_{1,n}; used as background only.
- Jensen-McCammond-Meier [JMM] (2006, per bibliography): presentation of H^*(PSigma_n;Q); context for the conjecture only.
- Transfer over Q, and self-duality of irreducible W_n-representations (standard; not stated in the manuscript).
- Theorem 4.2 is NOT used in the proof of Theorem 4.6. The proof 'mimics' it but is self-contained given [GP] and Pieri.
Missing inputs
- Texts of [OS], [LS], [GP], [Do] to confirm cited item numbers and the exact form of the W_n-action on the Orlik-Solomon summands.
- arXiv v1 text of the proof of Theorem 4.6. The v3 proof incorporates a simplification credited to J. Wilson after the cutoff, so the original v1 argument was not audited.
- Text of [JMM] to confirm the sign and index conventions of the relations and of the W_n-action on the alpha_ij.
Scope changes
- An explicit range n>=4i for Theorem 4.6 and Corollary 4.7 was derived from the supplied proof; the manuscript states none.
- Conjecture 4.8 was assessed only for status and internal consistency of the surrounding statements, not for truth.
Located concern
section:gbg, displayed definition of X'_n ('z in C^{2n}')
Literal definition error: the ambient space should be C^n.
W_n acts by permuting and negating n coordinates; the proof takes hyperplanes z_i=+-z_j, z_i=0 for 1<=i,j<=n and computes codimension 'in C^n'. With C^{2n} and all 2n coordinates constrained one would get X'_{2n}.
Replace C^{2n} by C^n and add 'for i not equal to j' to the first two conditions.
None after correction.
Located concern
thm:genpure statement and first two paragraphs of its proof
Structure maps are never specified, and injectivity is justified only by an informal appeal to 'naturality of these relations'.
Representation stability is defined only for a consistent sequence with given maps. The proof speaks of 'the image of H^(X'_n) in H^(X'_{n+1})' without naming the map.
Take the maps induced by forgetting the last coordinate; they are W_n-equivariant. Injectivity: the projection X'_{n+1}->X'n has a continuous section, for example adjoining the point 1+|z_1|+...+|z_n|, so the pullback on cohomology is injective. Alternatively, the hyperplanes not involving z{n+1} form a localisation, whose Orlik-Solomon algebra injects.
Gap closed by a one-line argument; theorem stands.
Located concern
final paragraph of proof of thm:genpure ('where the last sum is over those partitions mu ...', 'must have multiple boxes added to the first row')
Wording slips, and no explicit range is stated although one is promised as extractable.
'mu' should be 'nu'. What is needed, and true for n-m>=lambda+_1, is at least one added box in the first row, not 'multiple'. The proof's own bound gives stabilisation for n>=m+lambda+_1<=4i.
Correct the wording; optionally state the range n>=4i (my derivation from the supplied proof; sharp for i=1).
None.
Located concern
cor:gen and the preceding sentence 'by the same argument as Corollary 4.4'
The transferred argument uses self-duality of W_n-irreducibles, which the manuscript records only for S_n. As in Corollary 4.4, 'classical homological stability' means only abstract isomorphism.
Section 2.1 states self-duality for S_n only. Without it the dimension count yields the multiplicity of the dual representation.
Add that every element of W_n is conjugate to its inverse (equivalently, the irreducibles are realised over Q), so all representations are self-dual; and state the conclusion as equality of dimensions.
Corollary holds after adding the standard fact.
Located concern
conjecture:string, its footnote, and the displayed relations for H^*(PSigma_n;Q)
Status and typo. The statement is a conjecture as of the cutoff; the footnote's later proof is post-cutoff. The second displayed relation lacks '=0'.
[Wi1] is dated 2012 (arXiv 2011) in the bibliography, after 2010-08-07. The three-term expression is displayed without an equality.
Keep the conjectural status in any citation of this paper; add '=0'.
No theorem is claimed. The conjecture must not be counted as a result of this manuscript.
Located concern
proof of thm:genpure, sentence 'We are grateful to Jenny Wilson for pointing out this simplification', and Acknowledgements
Version/chronology: the v3 proof contains a post-cutoff simplification, so it is not the original v1 argument.
The acknowledgement appears in the v3 text; v1 is not supplied. TeX comments were not treated as assertions.
Compare with v1 when a dated copy is supplied. The simplification itself uses only [OS] and [LS], which predate the cutoff.
None on the truth of the theorem; the historical proof in v1 remains unaudited.
Outside the audit
- Hemmer's own proof [He] and the exact content and date of any pre-cutoff preprint.
- Cited item numbers in [OS], [LS], [GP], [FH], [Do], [JMM] (articles not supplied).
- The assertion that type-preservation fails for H^i(P_n;Q) for every i>=2.
- The Magma and Hemmer computations for n=7,8,9 as computations; the formulas were only re-derived by hand from the Pieri rule.
- The arXiv v1 wording of Theorems 4.1 and 4.6 and their proofs, and comparison with the published journal version (pending).
- Truth of Conjecture 4.8 and the correctness of the stated W_n-action on H^*(PSigma_n;Q) beyond consistency checks.
- All sections other than Section 4, except definitions used; in particular the applications of Theorem 4.1 claimed in the Introduction via [CEF2] and [C], and Section 5.4.
- Integral or positive-characteristic analogues: all results here are over Q only.
Additional supporting archive items are identified in the original report. View the manuscript hashes and repository snapshot.