gpt-6.2-sol high: cf-lie
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.
Two defects affect literal theorem statements: Definition 5.2 does not ensure coefficient compatibility, and Definition 5.5 imposes all-rank orbit generation, which fails at the birth of a new Schur type even for the canonical free and abelian Lie examples. These are distinct from the repairable chain-complex inferences in 5.3, the failure of the §3 cancellation proof in its stated generality, and the distinction between multiplicities and maps in the symplectic applications. With compatible coefficients and the finite-dimensional uniform §3 inputs, the intended graded strong-stability argument can be repaired. Theorem 5.4 is a multiplicity-only result and needs less. In §6 the low-degree evidence is not proof of the general conjectures. Historical cutoff used: 2010-08-07, conservative arXiv-v1 fallback from the supplied submission-history record, not verified journal receipt; the audited author text is v3 dated 2013-06-10. Bibliography dates are manuscript evidence, not independent verification of theorem contents or of early preprint versions. No external search, computation tool, cited-paper inspection, or journal comparison was made. Prior knowledge cannot be filtered by publication date, so it was not used as evidence of availability.
L1
Checked statement: Theorems 5.3–5.4: the equivalence between degreewise strong stability and strong stability of graded Lie algebra homology, and the one-way symplectic uniform multiplicity implication.
Returned assessment: claim_false
The literal coefficient definition omits compatibility of the transition maps with the Lie-module action. With the printed hypotheses, (1) does not even force eventual constancy of the dimensions in (5). This is distinct from the status of the intended theorem with compatible coefficient maps. In that setting, the fixed grading gives a bounded complex of finite-dimensional algebraic representations. The reverse arguments are calculations of multiplicities in split exact sequences. With trivial coefficients or with adjoint coefficients, the unknown Lie piece is recovered directly, with no tensor cancellation. For general coefficients, let a be the least grading with an eventually nonzero coefficient piece. Below a the pieces are eventually zero. In total degree a+k the only still-unknown chain summand is L_n^k⊗M_n^a in C_1. The two-ended calculation in the manuscript recovers its stable multiplicities; only finite-dimensional uniform tensor cancellation is then needed. The forward argument works with a correction at the quotient step: use the multiplicity-one restriction of the matching type and the stabilized kernel and boundary multiplicities to prove injectivity before applying a criterion that assumes it. Theorem 5.4 concludes uniform multiplicity stability only. Its reverse calculation neither needs type preservation nor uses any Lie-algebra transition map, and it does not use nontrivial tensor cancellation.
Premises used
- The representation-theoretic reading used for the intended theorem is finite-dimensional algebraic representations in each grade, as in §2's highest-weight discussion. Bare finite-dimensional linear actions of the abstract groups would not suffice for the semisimplicity or highest-weight steps. The counterexample below uses only algebraic trivial representations.
- Internal §3 input actually used: Theorem 3.1(1)–(2), finite-dimensional uniform multiplicity versions for tensors and exterior powers; the strong versions for SL and GL in the forward direction only. No nonuniform version, shifted-sequence result, GL-to-Sp restriction result, or symmetric-group result is needed.
- Internal §3 input actually used: Theorem 3.3 (\label{thm:LRinvert}), finite-dimensional, uniform SL/GL cancellation in (4) ⇒ (1), with the tensor factor eventually nonzero. If the tensor product is eventually zero the other factor is eventually zero, so this case needs no division. For (2) ⇒ (1) and for 5.4 the coefficient factor is the tensor unit.
- FH import: available_from=1991; date_evidence=\bibitem{FH}; date_basis=dated book entry; version=Fulton–Harris, A First Course, GTM 129. Needed as quoted: polynomial Littlewood–Richardson decomposition (Eq. 6.7), fixed polynomial plethysm (Ex. 6.17a), restriction by adjoining a trivial coordinate (Ex. 6.12, via \eqref{eq:SLres}), and the stable, fixed-label symplectic tensor formula (Eq. 25.27). Use these for algebraic representations in characteristic zero, with rank sufficiently large for the fixed labels; not as a nonuniform assertion or as a formula at arbitrary symplectic rank. The book was not inspected.
- Kab import: available_from=1997; date_evidence=\bibitem{Kab}; date_basis=dated journal entry; version=J. Algebra 195, 233–240, Theorem 7 as quoted in §3. The use through 3.1(2) in the symplectic argument is only fixed exterior degree on fixed irreducible symplectic types, then finite sums, with the quoted stable range n≥ℓ(μ)|λ|. Source and precise version of the formula remain conditional.
Missing inputs
- A statement requiring degree-preserving transition maps of coefficients that are module maps over the Lie-algebra inclusions, or an explicit convention supplying that requirement.
- The cited FH and Kabanov pages, to verify exact formula hypotheses and stable ranges. Bibliography establishes dates of the cited editions, not their contents.
Scope changes
- The negative verdict concerns the literal 5.3(1) ⇒ (5). The intended compatible-coefficient theorem and the multiplicity-only Theorem 5.4 have been evaluated separately.
- The tensor-cancellation repair is new and restricted to finite-dimensional uniform sequences; it does not validate the general 3.3 statement or its original proof.
- For the literal modular, mixed-tensor, and reversed-map notions, no homological equivalence is inferred from 5.3–5.4.
Located concern
Native TeX lines 2427–2755, Definition 5.2 [Admissible coefficients] and \label{eq:wedgeLnsquare}.
The printed admissibility conditions do not imply the chain square.
For every n take the one-dimensional abelian Lie algebra Qz in grade 1 with identity transitions and trivial G_n-action. Take M_n^0=Qu and M_n^1=Qv with trivial G_n-actions and identity transitions, all other pieces zero. Let zu=v for even n, zu=0 for odd n, and zv=0 always. At each rank this is a graded equivariant Lie-module action. Each coefficient grade is finite-dimensional and strongly stable and M_n^0 is nonzero. The Lie sequence satisfies all consistency and type-preservation conditions and (1). But H_0(L_n;M_n)^1 is zero at even n and Q at odd n. Thus (5) fails even as a multiplicity statement; this does not just exhibit a missing proof of naturality.
Require f_n(xm)=i_n(x)f_n(m), where i_n is the graded Lie inclusion and f_n the graded equivariant coefficient transition. Then the differentials commute with the transitions term by term.
Theorem 5.3 is false on the printed coefficient hypotheses. The example is excluded by the natural compatibility requirement; it does not refute the intended compatible-coefficient theorem.
Located concern
Native TeX lines 2427–2755, \label{eq:gradedLhom} and the proofs of (4) ⇒ (1), (3) ⇒ (1), (1) ⇒ (5).
Chain indexing and the inference of injectivity on homology need correction.
In C_2 take j<j′ summands and the diagonal exterior summands, only with nonnegative coefficient grade. In adjoint coefficients grade 1 is H_0=L^1, not H_1. For ∂i:C_i→C{i-1}, H_i=ker∂i/im∂{i+1}, not the shifted quotient printed in the forward proof. A split equality [C_i]=[ker∂_i]+[im∂_i] is a same-rank character identity, not a natural splitting. Finally, IV′ on a homology image does not by itself justify applying Proposition 2.11: that proposition assumes injectivity. The proof also calls the zero Lie algebra an admissible adjoint coefficient module, although the definition excludes it.
New local repair: use IV on adjacent chain objects and the multiplicity-one matching restriction in Remark \ref{remark:strong}. The multiplicities of kernels and images are nondecreasing, and their sums are constant per type once the chain terms stabilize. Write Z_n=B_n⊕D_n equivariantly in the fixed slice. The orbit spans of the images of B_n and D_n are disjoint: apply the multiplicity-preservation assertion to their direct sum at each type. The first span is all of B_{n+1} by equality of multiplicities. Hence the induced map on Z_n/B_n is injective; the span of D_n in the quotient has precisely the matching types and fills the quotient. This proves I, II and IV without using the injectivity criterion circularly. Work per fixed slice, not with a common threshold for the entire graded object. Treat the zero Lie algebra separately for the adjoint implication.
These proof repairs suffice for the forward chain step once coefficient compatibility is imposed. The index errors do not furnish a counterexample to the intended theorem.
Located concern
§3, \label{thm:LRinvert}, proof of the projection and final compactness paragraph; consumed in native TeX lines 2427–2755 in (4) ⇒ (1).
The supplied tensor-cancellation proof does not justify the projection as a ring homomorphism, and the full nonuniform theorem is false.
For SL, multiply the standard representation by its (n−1)st exterior power. In rank n+1 the column of height n is sent to zero by the proposed projection, whereas in rank n that exterior power is trivial. For Sp, at the step Sp_4→Sp_2 the projection kills the two-row type (1,1), while its tensor square contains the trivial type since it is self-dual. For the nonuniform statement take SL_n with W_n=Sym^n Q^n, and V_n=Q at even n and Q^n at odd n, ignoring irrelevant small ranks. W_n and V_n⊗W_n are nonzero and pointwise multiplicity stable: the latter has only polynomial diagrams of sizes n or n+1 and at most two rows, with no determinant-column collapse for n≥3. V_n is not multiplicity stable.
New repair for the finite-dimensional uniform case only: choose a fixed constituent μ of the stabilized nonzero W. For each constituent λ of V_n the tensor of highest-weight vectors forces λ+μ in the product. In SL and Sp this bounds λ in a finite set because product support is fixed. In GL write a type as a partition with last coordinate zero plus an integer determinant exponent, using the manuscript's repeat-the-last-entry convention. The Cartan constituent bounds the partition and restricts the exponent to a finite set as well. On those finite supports all multiplication coefficients are the fixed-label stable coefficients of 3.1(1). Transfer the two character equations to a common sufficiently large rank and cancel the same nonzero character in the domain of weight Laurent polynomials; injectivity of the character follows from highest weights. This uses neither the proposed projection nor compactness.
The imported full theorem is not available literally. The finite-dimensional uniform case needed for 5.3 has a separate argument conditional on stable tensor coefficients. Neither symplectic 5.4 nor the trivial-coefficient direction of 5.3 needs this cancellation.
Located concern
§2, \label{prop:SLstrong}, used in §3's strong closure argument, then in native TeX lines 2427–2755.
The highest-weight proof is phrased too broadly: one cannot in general rechoose a vector in a specified copy to be a larger torus eigenvector, or assert uniqueness among all weights restricting to a given weight.
The needed argument concerns larger-group highest lines only. Decompose the image of the given smaller highest vector into larger torus weights after observing it is killed by all positive roots. Each nonzero component is a larger-group highest vector. For the finite stabilized label support, and rank larger than all normalized partition lengths, two distinct stabilized highest labels cannot have the same smaller highest weight. For GL this follows from the long constant tail in the weight signature; for SL a zero-tail representative removes the determinant ambiguity.
Use that component argument without replacing the original vector. In the other direction the matching smaller type has multiplicity one and contains the larger highest vector; P_{n+1} fixes that vector and is normalized by G_n, so it fixes its entire smaller-group span.
The eventual finite-support version needed for strong closure survives. This is not an all-rank polynomial assertion; the attempted simple version has a separate counterexample in L2.
L2
Checked statement: Definition 5.5, Theorem 5.6, and Corollaries 5.7–5.9: simple polynomial representation stability at every rank where a type exists, including transition maps for free Lie algebras and for trivial- and adjoint-coefficient homology of their nilpotent truncations.
Returned assessment: claim_false
Simple stability is not just strong stability with an earlier threshold. It concerns polynomial Schur labels, including the rank equal to the number of rows; negative pseudo-partitions are excluded, both map conditions hold from rank 1, and SIV is imposed at every rank. A type of length r can first occur at r. For a fixed grade of a finite polynomial construction this leaves finitely many labels, so simple stability gives eventual strong stability, not the converse. Theorem 5.6 inherits the literal coefficient counterexample in L1, which is already simply stable in every input grade and also as a total representation. Independently, the claimed all-rank SIV characterization is false even for finite polynomial representations with injective and orbit-spanning maps. Neither issue disproves 5.7–5.9 for their actual canonical maps. Those corollaries admit a separate argument in which the rank-independent objects are coefficient modules for finite symmetric groups and homology is taken before evaluation in rank n.
Premises used
- Definitions 5.1, 5.2, 5.5 and the GL label convention in §2. SIII's lower cutoff means that a Schur functor vanishes, not that it is the nonzero trivial representation, below its row length.
- FH import: available_from=1991; date_evidence=\bibitem{FH}; date_basis=dated book entry; version=Fulton–Harris GTM 129, §§6 and 15.3 and Eq. 6.7/Ex. 6.17a as quoted. The original argument uses polynomial Schur decomposition and Schur-functor identities over Q; the new argument uses only the manuscript's identification of V^{⊗d}⊗_{Q S_d} V_λ with the Schur representation of shape λ, irreducible exactly in the row-length range. It does not use the nonuniform conclusions of §3.
- Tirao import for antecedent and examples: available_from=2002; date_evidence=\bibitem{Ti}; date_basis=dated journal entry; version=J. Lie Theory 12, 309–323, Theorem 2.9 as cited for trivial-coefficient free k-step nilpotent Lie homology, plus the unnumbered small-degree computations quoted. This is an attribution and conditional comparison, not needed to prove the three principal examples by the new argument.
- Hain import for the manuscript's alternative free-group route: available_from=1991; date_evidence=\bibitem{Ha2}; date_basis=dated proceedings entry; version=Proc. Sympos. Pure Math. 53, Corollary 5.10 as cited, for the free group's continuous cohomology comparison. No comparison theorem is needed if one computes free Lie homology directly.
- Bakhturin import for the displayed character formula only: available_from=1987; date_evidence=\bibitem{Ba}; date_basis=dated book entry; version=Identical relations in Lie algebras, §3.4.5 Proposition 3, for the degree-m free Lie component in characteristic zero. CT1 import for the quoted 2-step computations only: available_from=2002; date_evidence=\bibitem{CT1}; date_basis=dated journal entry; version=Q. J. Math. 53, Theorems 3.1 and 4.4 as cited, ordinary and adjoint homology of free 2-step nilpotent Lie algebras. Neither formula is needed for the rank-stability repair.
Missing inputs
- Clarification whether the hypothesis 'type-preserving' in 5.6 is meant to be eventual per grade, simultaneous across grades, or an all-rank condition with polynomial labels rather than §2 pseudo-partitions.
- A valid all-rank converse for general compatible coefficient sequences, if the authors want the full equivalence of 5.6 rather than a polynomial-functor result for the applications.
- The cited sources for the exact Tirao, Bakhturin, and Cagliero–Tirao formulas and their conventions; these formulas were not independently established in this audit.
Scope changes
- The rank-1 counterexamples test the actual natural maps and positive k, m, and homological degrees. They do not refute eventual strong stability.
- The slot-module argument is a new verification of the surviving polynomial multiplicity and transition statements, not validation of the manuscript's proof or of its all-rank definition.
- No claim is made here that the omitted all-rank compatible-coefficient converse follows merely by changing Condition II.
Located concern
Native TeX lines 2756–3221, \label{thm:simplehomLie}, first proof paragraph; compare native TeX lines 2427–2755, Definition 5.2.
The literal all-coefficient theorem fails, and the proposed all-rank equivalence SIV ⇔ IV′ fails independently of coefficient compatibility.
The L1 example has simple Lie and coefficient pieces but alternating dimensions in homology. For the independent assertion, take E_n=Q^n and V_n=E_n⊕Λ²E_n. Use the canonical maps except at 1→2, where e_1 maps to (e_1,e_1∧e_2). This map is GL_1-equivariant. Its orbit spans both distinct irreducibles at rank 2, and at later steps the canonical images span; the maps are injective and SIII holds. P_2 fixes both displayed components, and all the later P_{n+1} fix the canonical images. But at 1→2 the line's GL_2-span is E_2⊕Λ²E_2 rather than E_2. The ambiguity is a highest weight with an extra last entry that was invisible on restriction.
Require compatible coefficient maps. Do not use IV′ alone at all ranks: it needs further control, for example preservation of homogeneous polynomial degree along with the highest-vector condition. A fixed smaller weight and the same polynomial degree then force the additional coordinate to be zero. Even with this repair to the criterion, provide a separate all-rank inverse argument, not an appeal to the projection in 3.3.
Theorem 5.6 is false literally. The all-rank auxiliary assertion used in its proof cannot be salvaged by correcting just the coefficient definition. This audit does not assert that the full intended compatible-coefficient equivalence is false.
Located concern
Native TeX lines 2756–3221, Definition 5.5, \label{thm:simplehomLie}, and \label{cor:freeLie}; compare the GL pseudo-partition convention in §2.
The application does not distinguish SIV from literal IV at the first rank a polynomial type occurs.
The canonical inclusion sends Λ²Q² into Λ²Q³ and generates the latter. But under the §2 GL label convention the type of Λ²Q² has signature (1,1), continued as (1,1,1) at rank 3, i.e. the determinant of GL_3, not Λ²Q³. Likewise at rank 1 the standard representation has this distinction. This is not a failure of the canonical maps to satisfy SIV. It is a failure of interchanging the two requirements at the boundary. Eventual IV on a fixed polynomial grade avoids the boundary, but does not check it.
Use polynomial-label transition conditions in a simple-stability theorem, spell out their quantifiers, and prove the boundary assertions separately. In the examples read the free nilpotent algebra as the quotient by the ideal generated by L_{k+1}, equivalently by ⊕{d>k}L_d, not by the single vector subspace L{k+1}.
The all-rank examples cannot be accepted just by the proof interface as printed. The truncation notation is harmless under its evident ideal interpretation.
Located concern
Native TeX lines 2756–3221, \label{cor:freeLie}, \label{corollary:nilp}, and \label{cor:adjnil}.
The principal examples need an argument independent of the defective general simple-stability proof.
Their maps come from linear inclusions before applying any construction. This is stronger information than numerical multiplicities and than the right-column unipotent fixing the image.
New argument for these examples: in total degree d form the Q S_d-module A of formal operations with d distinct input slots. For the free Lie piece use multilinear bracketed words modulo antisymmetry and Jacobi. Substituting tensors identifies the resulting functor with V^{⊗d}⊗_{Q S_d}A: in the reverse direction number the leaves of each word, and changing the numbering is exactly the tensor-over-S_d relation. The same description gives the truncation by keeping only word pieces of degree ≤k. In a Chevalley–Eilenberg slice use exterior products of these bracketed words, with either no coefficient word or one extra adjoint word. Swapping chain words imposes the exterior sign, and all relations are multilinear; numbering the input slots proves the same evaluation identification. The differential on the slot modules joins words by the bracket, including the adjoint action term, so substitution commutes with the differential. Over Q the slot modules split by averaging over S_d, and evaluation is exact; the evaluated homology is therefore evaluation of the homology of the slot complex. Decompose that module into Specht modules. Each summand evaluates to its Schur functor and every transition is the canonical Schur map tensored with the identity on multiplicities. The maps on each Schur summand are injective by functoriality and a linear retraction; a nonzero image in the irreducible matching summand generates it. The same argument on any line in the multiplicity space gives SIV. A type with too many rows vanishes at the smaller rank; nevertheless it must be generated at the next step for Condition II. Address this last point specifically: when a shape has exactly n+1 rows its summand does not occur at rank n, and the canonical Schur maps are block-diagonal. Thus the canonical maps cannot generate that newly appearing summand. This proves the polynomial multiplicity and SIV statements, but it shows why one must not infer the all-rank II asserted by Definition 5.5. The grading sums are finite at fixed homological degree (at most ki for trivial coefficients and k(i+1) for adjoint coefficients).
For the actual canonical maps the corollaries are not literally true with all-rank Condition II. In 5.7 use m=2: L_2(Q)=0 and L_2(Q²)=Λ²Q²≠0. In 5.8 take k=1 and i=2: the Lie algebra is abelian and the same 0→Λ²Q² occurs. In 5.9 take k=1 and i=1: homology is E_n⊗E_n with trivial adjoint action, and the rank-1 image generates Sym²Q² but not Λ²Q². All three have the stated polynomial multiplicity pattern and type-preserving Schur maps; the false condition is orbit generation at the birth rank. A corrected definition would impose II only for types already defined at the source, or only once the finite support has appeared.
L3
Checked statement: Remaining Lie applications, Malcev comparisons, and the Torelli and IA conjectures: verify what follows on multiplicities and natural maps, distinguish computations from conjectures, and separate retrospective references.
Returned assessment: repairable_gap
The surface-group calculation supports uniform multiplicity stability, subject to the quoted comparisons, but not forward stability for the maps provided. The symplectic exterior-power lemma survives an erroneous coefficient in its proof. Ordinary Heisenberg homology gives actual eventual uniform representation stability with the stated inclusions. For the example with coefficients the proof as printed gives multiplicities conditional on cited computations, not the map conditions; a direct map check for positive-degree adjoint homology is possible given the quoted decomposition. The pure-braid Malcev statement and the Torelli/IA proposals remain conjectures as presented historically. Degree-one calculations are evidence with explicit rank and coefficient restrictions; in the Torelli Malcev degree-two case the displayed decomposition verifies the multiplicity part, not by itself injectivity on the lower central quotients. Several ancillary assertions need correction independently of the conjectures.
Premises used
- Surface/free comparison imports: Labute, available_from=1970, date_evidence=\bibitem{La}, date_basis=dated journal entry, version=J. Algebra 14, 16–23, one-relator lower central series result as specialized to the closed orientable genus g≥2 surface and its symplectic relation. Nomizu, available_from=1954, date_evidence=\bibitem{No}, date_basis=dated journal entry, version=Ann. of Math. 59, 531–538, nilmanifold with a lattice in a simply connected nilpotent Lie group; use on torsion-free nilpotent quotients/Malcev models, not an unqualified application to every quotient with torsion. Hain Ha2, available_from=1991, date_evidence=\bibitem{Ha2}, date_basis=dated proceedings entry, version=Proc. Sympos. Pure Math. 53, Cor. 5.10 for free groups, Prop. 5.11 for surface groups and Ex. 5.12 for pure braids, as cited. The separate comparison with the associated graded Lie algebra must be checked in each case.
- Hain surface stability attribution: available_from=1997; date_evidence=\bibitem{Ha}; date_basis=dated journal entry; version=JAMS 10, Corollary 8.5 as cited. Whether the result includes any structure-map conclusion is not established by the citation in the supplied text.
- Symplectic exterior powers: FH, available_from=1991, date_evidence=\bibitem{FH}, date_basis=dated book entry, version=GTM 129 §17 as cited, primitive/contraction decomposition in the range exterior degree ≤ symplectic rank. CT2, available_from=2004, date_evidence=\bibitem{CT2}, date_basis=dated journal entry, version=Adv. Math. 181, Theorems 4.2, 4.13 and Corollary 4.15 as cited. Needed for the original coefficient example: the fixed-i,k large-rank decomposition with adjoint or exterior adjoint coefficients, not merely the dimensions of vector spaces or cohomology of an unspecified cotangent bundle. The translation from the cited results, including degree conventions and any naturality, was not verified.
- Pure-braid graded presentation: available_from=1985; date_evidence=\bibitem{Ko}; date_basis=dated journal entry; version=Kohno, Invent. Math. 82, 57–75, infinitesimal pure braid algebra presentation in characteristic zero as cited. Theorem 4.1 supplies ordinary rational braid cohomology stability, not a theorem transferring it to graded Lie algebras for S_n. Its cited Hemmer input has available_from=2011 for the listed journal version, date_evidence=\bibitem{He}, date_basis=dated journal entry, version=JCTA 118, 1136–1139; availability of a version by the cutoff is unresolved. The published 2011 version is not counted as cutoff evidence.
- Torelli imports: Johnson Jo2, available_from=1985, date_evidence=\bibitem{Jo2}, date_basis=dated journal entry, version=Topology 24, 127–144, rational abelianization of the one-boundary Torelli group for g≥3, using the Johnson map and its naturality for stabilization. Hain Ha, available_from=1997, date_evidence=\bibitem{Ha}, date_basis=dated journal entry, version=JAMS 10 as cited, infinitesimal presentations in the closed case. Habegger–Sorger HS, available_from=2000 as listed, date_evidence=\bibitem{HS}, date_basis=preprint year in manuscript, version=unidentified 2000 preprint, boundary extension for g≥6 and Theorem 2.2 as cited for degree two; exact preprint content and naturality unresolved.
- Rational extension context: Margulis Ma, available_from=1991, date_evidence=\bibitem{Ma}, date_basis=dated book entry, version=Discrete Subgroups, VIII.B as cited, virtual extension in the higher-rank symplectic arithmetic setting; BMS, available_from=1967, date_evidence=\bibitem{BMS}, date_basis=dated journal entry, version=IHES 33, Theorem 16.2 as cited, congruence subgroup result in the stated higher ranks; Borel Bo2, available_from=1960, date_evidence=\bibitem{Bo2}, date_basis=dated journal entry, version=Ann. of Math. 72, Prop. 3.2 as cited, density for restrictions of algebraic representations. None by itself supplies the conjectured action on the full homology or uniqueness of a GL_n extension from GL_n(Z).
- IA abelianization import: available_from=2006 as listed, date_evidence=\bibitem{Ka}, date_basis=preprint year in manuscript, version=unidentified Kawazumi preprint cited for the Andreadakis/Farb/Kawazumi/Cohen–Pakianathan computation. Needed: the natural first Johnson identification over Q for IA_n, n≥3 for the two-type formula. The map argument here is conditional on that identification.
- Stable cohomology context, not proofs of the conjectures: Morita Mo1, available_from=1999, date_evidence=\bibitem{Mo1}, date_basis=dated proceedings entry, version=Geom. Topol. Monogr. 2, Conjecture 3.4 for Torelli invariants; Madsen–Weiss MW, available_from=2007, date_evidence=\bibitem{MW}, date_basis=dated journal entry, version=Ann. of Math. 165 for stable rational mapping class cohomology, not Torelli; Galatius Ga, available_from=2011 for the journal version, date_evidence=\bibitem{Ga} also lists arXiv:math/0610216, date_basis=dated journal entry and preprint identifier, version=2011 article or an as-yet-unverified cutoff preprint, stable positive-degree rational cohomology of Aut(F_n), not IA_n. The separate claim about restriction of odd MMM classes has no precise source at this point in the manuscript.
- Finiteness evidence: Mess Me, available_from=1992, date_evidence=\bibitem{Me}, date_basis=dated journal entry, version=Topology 31 Cor. 1 and Prop. 5 as cited (the latter attributes the example to Johnson–Millson); BBM1 and BBM2, available_from=2007 and 2010 for the listed journal versions, date_evidence=\bibitem{BBM1}, \bibitem{BBM2}, date_basis=dated journal entries, versions=Invent. Math. 170 and JAMS 23 respectively, for the indicated top-degree Torelli/IA classes. The module identifications and any passage from closed surfaces or Out(F_n) to the exact groups here are conditional.
Missing inputs
- Cited primary text for the natural and grading-compatible Malcev comparisons and for Hain's exact statement on surface Lie algebras; no forward surface Lie map is supplied.
- Cited CT2 statements with coefficient and degree conventions, and either a map statement or a proof of I and II for the general exterior-coefficient example.
- The exact HS and Kawazumi preprint versions, the Johnson comparison maps, and the precise module assertions in Mess and Bestvina–Bux–Margalit.
- A pre-cutoff record for the Hemmer result if the historical proof of Theorem 4.1 is used, and version evidence for any early Galatius preprint invoked under the cutoff.
- No current-literature status is inferred for the §6 conjectures.
Scope changes
- Corollary 5.10 is supported only in uniform multiplicity form by the argument audited here. No forward or reversed map theorem is substituted silently.
- For Example 5.14 distinguish the conditional adjoint conclusion with an additional map check from the exterior-coefficient conclusion whose map conditions remain open in this audit.
- Retrospective only: the Conjecture 5.15 footnote refers to CEF1, listed as April 2012, and §6 mentions Boldsen–Dollerup, listed as a 2012 article with arXiv:1101.5767, for the surjectivity part in degree two. Neither is used as cutoff evidence, and neither is an independently verified current-literature report.
Located concern
Native TeX lines 2825–3221, 'Continuous cohomology and pseudo-nilpotent groups' and \label{cor:hain}.
The displayed maps run in the opposite direction from forward representation stability; cohomology multiplicities do not remedy this.
Collapsing a handle gives gr(π_{g+1})→gr(π_g), not a Lie inclusion. In the Labute presentation the obvious inclusion on the old generators sends the old defining relation to −[a_{g+1},b_{g+1}], nonzero in degree two modulo the single total relation. The dual cohomology argument can recover multiplicities grade by grade without any Lie transition maps. The phrase 'for all j' at the end of the corollary has no bound variable j. Also, the full automorphism group of a closed orientable surface group includes orientation-reversing automorphisms, which negate the intersection form, so the symplectic assertion needs a restriction to orientation-preserving automorphisms or a separate choice of symplectic action.
State the deduced conclusion as uniform multiplicity stability of each gr(π_g)^i. Given the equivariant graded comparison, dualize each finite-dimensional graded piece and apply the character calculation behind 5.4: H_0 occurs in grade 0, H_1 in grade 1, and the surface H_2 class is in grade 2 (also seen from the single quadratic relation). If reversed-map stability is intended, prove both map conditions of Definition 2.5 separately; do not use the forward theorem as proof of them.
The multiplicity consequence is conditional and repairable. Forward representation stability with the stated maps is not a well-formed conclusion.
Located concern
Native TeX lines 2825–3221, \label{lem:wedgeH}, \label{remark:substitute}, \label{example:Heis}, \label{example:adjHeis}.
An exterior-power coefficient is wrong, and the example with coefficients conflates uniform multiplicities and full stability.
Write t=a_{n+1}∧b_{n+1}. Since t²=0 and it commutes with ω_{n+1}, the expansion is (ω_{n+1}−t)^j=ω_{n+1}^j−jω_{n+1}^{j−1}t, not the printed expression for j>1. The contraction calculation only needs C^j(v)≠0 and C^{j+1}(v)=0; the numerical coefficient depends on normalization. The kernel argument in the remark uses equivariance of f_n; the image multiplicities follow from the quotient by the kernel, not merely 'the same' argument on the codomain. For the adjoint formula, H_0(H;H)=H/[H,H] is a single standard module, so the displayed two-term positive-degree formula must not be used at i=0. The exterior-coefficient range 'for i≤n' is insufficient if read independently of k: at i=0,k=4 the coefficient is zero at n=1, whereas at n=2 the coinvariants retain the nonzero horizontal Λ⁴Q⁴ quotient.
Correct the expansion and keep its factor j in the next line. The same wedge-image bound and the nonzero difference under interchange of symplectic pairs prove the lemma over Q. For ordinary homology write the chains as horizontal plus central-wedge terms: the only differential is contraction from the horizontal part to the central part. In the stable range contraction from Λ^{i+1}H onto Λ^{i−1}H is surjective by the symplectic exterior decomposition, leaving the primitive horizontal Λ^i type. Its map is injective and its nonzero image spans the matching irreducible. For adjoint i≥1, conditional on the quoted two-type computation, both types are already represented by horizontal cycles in the minimum total degree i+1: use (a_1∧⋯∧a_i)⊗a_1 and the alternating image of a_1∧⋯∧a_{i+1}. There are no boundaries in that total degree, and both highest vectors persist under inclusion. This gives I and II there; handle H_0 directly. For general exterior coefficients specify the correct stable range and supply a corresponding map argument, or limit the example to uniform multiplicity stability.
Ordinary Heisenberg stability survives. The coefficient example is not disproved by the map omission, but its full exterior-coefficient conclusion is not established by the displayed multiplicity evidence.
Located concern
Native TeX lines 2825–3569, §5.4, presentation preceding \label{conjecture:malcevpn} and argument immediately after it.
The printed presentation lacks the symmetry of the infinitesimal generators, and the comparison and hypothetical S_n implication have different statuses.
The generators are indexed by all ordered distinct pairs. The relations as printed are quadratic only, leaving degree one of dimension n(n−1), not the n(n−1)/2 obtained from braid cohomology in \eqref{eq:h1pn} and its basis. Pseudo-nilpotence compares group cohomology to continuous cohomology; it does not alone identify cohomology of the associated graded. And no S_n analogue with the necessary hypotheses is proved in this section.
Impose X_{ij}=X_{ji}, specify the comparison with the associated graded and its naturality, and retain the displayed statement as a conjecture in the historical main text. With the corrected presentation, adding or deleting a strand does define a split consistent system on the graded Lie algebras: deleting kills terms with that new index. A degree-i bracket word uses at most 2i indices, which verifies orbit generation once n≥2i; this elementary check is distinct from uniform multiplicity stability.
The presentation is false literally. The main text provides evidence and a conditional analogy, not a proof of the conjecture. The footnote's later proof is retrospective only.
Located concern
Native TeX lines 2825–3569, §6, \label{conjecture:torelli}, \label{conjecture:malcev:torelli}, \label{eq:ia1}, \label{conjecture:ian}.
Rational extension, finite-dimensional-part preservation, and stability of actual maps must not be conflated.
The natural homology maps initially intertwine the integral groups. For IA the paragraph before \eqref{eq:ia1} calls them GL_n(Q)-maps before an extension is available; moreover even algebraic extension from GL_n(Z) is not unique, since twisting by det^{2a} leaves the integral restriction unchanged. The finite-orbit part is locally finite, not automatically finite-dimensional, and its dual cannot simply be identified with the finite-orbit part of full cohomology. The equivalence claimed in Remark 1 after 6.1 needs proof, especially since the manuscript explicitly distinguishes reversed stability from stability on duals. Finally a group inclusion need not give injections on its lower central quotients. The cited degree-two decomposition therefore does not check I for the Torelli Malcev conjecture.
Use integral equivariance until algebraic extensions and their compatibilities are established; specify a coherent GL choice. Treat the cohomological proposal separately or supply a duality argument on the indicated parts and maps. For the one-boundary Torelli H_1, the natural Johnson identification with Λ³H_Q gives injective exterior maps and orbit generation by moving the symplectic pairs supporting a monomial into the old genus. For IA H_1 the Johnson map uses the exterior map and extension of covectors by zero; the natural tensor basis has support on at most three indices, giving I and II. Its stabilized labels are V(1,1;1) and V(1) from rank 3, so the result is uniform mixed stability, not ordinary uniform stability. For the Malcev degree-two claim, add a map proof or state uniform multiplicity stability only. Generation by grade one gives a separate route to II: bracket words of grade i in wedge-cube generators involve at most 3i symplectic pairs, so they can be moved into the old genus in that range, conditional on the stated presentation and compatible algebraic actions.
The conditional degree-one evidence is substantive. It does not establish the conjectures for higher homology. The two pieces of Malcev low-degree evidence differ in how much of the map conditions they verify.
Located concern
Native TeX lines 2825–3569, §6, unlabelled Morita conjecture, \label{conjecture:invpart}, \label{conjecture:stablyfinite}, \label{conjecture:finitegen} and following evidence.
Several surrounding literal assertions require qualifications; none of these conjectures is a theorem of the manuscript.
The MMM class e_i conventionally lies in degree 2i, rather than the printed degree i. Aut(F_n) vanishing is a positive-degree statement; the IA invariant vanishing conjecture must also exclude degree zero, where the rational unit survives. An ordinary permutation module on an infinite set over Q has a nonzero proper augmentation kernel, so it cannot itself be an infinite-dimensional irreducible permutation module as said in the discussion of the Mess examples. Local finiteness in 6.6 is not finite dimension of the whole space. Also a cyclic permutation representation may be infinite-dimensional despite being finitely generated as a Q[group]-module.
Specify positive degree and the MMM grading, and verify the exact Mess representation before using 'irreducible permutation'. Spell out Q[Sp_{2n}(Z)] and Q[GL_n(Z)] in 6.7. Distinguish the finiteness conjectures: if both held in a stable degree, a finite set of module generators with finite-dimensional orbits would imply finite dimension there; neither conjecture says this by itself.
These corrections concern the formulation and the evidence, not a claimed proof or disproof of the positive-degree conjectures.
Outside the audit
- Standalone proof families in §§2–4 and §§7–8 except for interfaces expressly discussed; in particular, no full historical audit of braid cohomology or of the modular-stability claims.
- Exact quoted computations in Bakhturin, Tirao, Cagliero–Tirao, Hain, Habegger–Sorger, Johnson, Kawazumi, Mess, and Bestvina–Bux–Margalit beyond the source-only and elementary checks recorded.
- Verification of later proofs or current conjecture status, comparison with the arXiv-v1 text, and publisher-text or oblatum verification.
Additional supporting archive items are identified in the original report. View the manuscript hashes and repository snapshot.