claude-fable-5-2 high: cf-modular-challenges
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.
Both criticisms survive adversarial checking, and they amount to two independent defects plus one proof gap. (1) Coefficient conflation at p=2: H_1(Mod_{g,1}(2);Z) has a Z/4 quotient by a direct homomorphism and exponent 8 by Sato Prop 4.9, so Theorem 8.4 is false as stated at p=2. The earlier claim about H_1(Gamma^{Sp}{2g}(2);F_2) is the same error; that group is sp{2g}(F_2), and Putman's Cor 3.6 is integral. (2) The exterior-cube table is reversed; the correct class is V(w3) + (1 + [g = 1 mod p]) V(w1). The identity delta(epsilon) = g-1 alone refutes only the g = 1 row; the full list needs simplicity of the remaining quotient, supplied either by Baranov-Suprunenko 2.1/2.9 with a character bridge or by a short direct weight argument. (3) Conditions I and II are never checked. What survives, conditionally on the unsupplied Putman Theorem 7.8: for odd p the group is an F_p-representation (new -I argument) with period-p multiplicities; for p=2 only a periodic Jordan-Holder statement for the integral Z[G]-module, which the corrected lists match against Sato's order exactly. Retrospective limit: I cannot remove later knowledge from my weights; no post-cutoff result is used to close any obligation.
MC1
Checked statement: CF Theorem 8.4 at p=2, namely that H_1(Mod_{g,1}(2);Z) is an F_2-representation of Sp_{2g}(F_2) with the printed composition factors and is periodically uniformly representation stable; the preceding sentence that 'it follows from Putman's characterization that H_1(Mod_{g,1}(p);Z) is in fact an F_p-representation'; and the earlier sentence that H_1(Gamma^{Sp}{2g}(2);F_2) is an extension of sp{2g}(F_2) by H.
Returned assessment: claim_false
The prior criticism survives: Theorem 8.4 is false as literally stated at p=2, and I found no convention that rescues it. Non-blind audit; my weights may contain later literature (including a recollection of Putman's published Theorem 7.8), which I do not use as evidence.
(1) Same group. Sato (p.2) takes Sigma_{g,r} of genus g with r boundary components, diffeomorphisms fixing the boundary pointwise, and M_{g,r}[d] the kernel of M_{g,r} -> Sp(2g;Z_d). CF's Mod_{g,1}(p) is the kernel of Mod_{g,1} -> Sp_{2g}Z -> Sp_{2g}F_p. With r=1, d=2 these agree, and both mean integral H_1. CF's definition of S_{g,1} is in an omitted section, but the excerpt uses the one-boundary Johnson and Putman b=1 conventions. The conclusion below does not depend on this point: the direct obstruction applies to any group surjecting onto Gamma^{Sp}_{2g}(2), whether closed, punctured or bounded.
(2) Concrete contradiction from Sato. Prop 4.9 (g>=3) states H_1(M_{g,1}[2];Z) is Z_8^{2g} + Z_4^{C(2g,2)} + Z_2^{C(2g,3)}. An F_2-vector space has exponent at most 2, so no F_2-structure exists for any g>=3. The stable range does not help. Sato's proof is not audited.
(3) Contradiction independent of Sato (the Sol character, which I rechecked by hand). Write M = I+2X in Gamma^{Sp}_{2g}(2) with X = (A B; C D) in the basis a_1..a_g, b_1..b_g. From M^T J M = J one gets X^T J + J X = 0 mod 2, so B and C are symmetric mod 2 and D = A^T mod 2. Put f_i(M) = B_ii + 2 sum_j A_ij B_ij mod 4. The product has X'' = X + X' + 2XX', so B''ii = B_ii + B'ii + 2 sum_j (A_ij B'ij + B_ij A'ij) mod 4. The quadratic term contributes the same cross term a second time, giving 4(...) = 0. So f_i is a homomorphism to Z/4. I + 2E{i,g+i} is symplectic and maps to 1. Since Mod{g,1}(2) -> Gamma^{Sp}{2g}(2) is onto, H_1(Mod{g,1}(2);Z) has a Z/4 quotient for every g>=1. This alone refutes the F_2-vector-space claim; Sato sharpens the exponent to 8.
(4) What Putman's supplied statements say. Cor 3.6 (g>=3, L even) is an exact sequence 0 -> H_1(Sigma_g;Z/2) -> H_1(Sp_{2g}(Z,L);Z) -> sp_{2g}(Z/L) -> 0 of Sp_{2g}(Z)-modules, with integral coefficients. CF line 4157 attributes this extension to H_1(Gamma^{Sp}{2g}(2);F_2). That is false. Sato Cor 2.2 gives A = H_1(Gamma_g[2];Z) = Z_2^{2g^2-g} + Z_4^{2g}, of order 2^{2g^2+3g}, matching Cor 3.6. Then A/2A has dimension 2g^2+g = dim sp{2g}(F_2), not 2g^2+3g. The surjection A -> sp factors through A/2A, so H_1(Gamma^{Sp}{2g}(2);F_2) is isomorphic to sp{2g}(F_2) equivariantly and the kernel H equals 2A. Tensoring with F_2 is only right exact.
(5) Order bookkeeping for Theorem 8.4's p=2 paragraph. CF writes that H_1(Mod_{g,1}(2);Z) is an extension of H_1(Gamma^{Sp}(2);F_2) by wedge^3 H + B_2/F_2. Sato's order is 2^{4g^2+4g+C(2g,3)}. This equals |H_1(Gamma(2);Z)| x |wedge^3 H| x |B_2/F_2| = 2^{(2g^2+3g)+C(2g,3)+(2g^2+g)}. With the true mod-2 congruence homology the exponent would be 2g short. So only the integral reading of Putman's extension is consistent with Sato. CF's two coefficient misstatements cancel in the factor count. This is one root defect (Z versus F_2 for the level-2 symplectic congruence group), not several.
(6) Three objects to keep apart at p=2. (a) The integral group A_g, a finite Z[Sp_{2g}(F_2)]-module (conjugation by Mod_{g,1} factors through Sp_{2g}(F_2)) of exponent 8. It is not an F_2-representation, so Definition 8.1 does not apply. (b) Its Jordan-Holder factors as a finite Z[G]-module, which are simple F_2[G]-modules and well defined; this is what CF's table actually computes. (c) H_1(Mod_{g,1}(2);F_2) = A_g/2A_g, a genuine F_2-representation of dimension 2g+C(2g,2)+C(2g,3) by Sato. Its composition factors are not determined by the supplied statements, and the printed table does not describe it (its total dimension is far larger).
(7) What survives at p=2 under reading (b). This is conditional on Putman 7.8 at L=2 (unsupplied), Cor 3.6, Shvartsman's B_2 list and Hogeweij's sp list as CF quote them, and the MC2 correction. The class is [sp_{2g}F_2] + [H] + [wedge^3 H] + [B_2/F_2]:
- g odd: V(0)^2, V(w1)^5, V(w2)^2, V(w3).
- g even: V(0)^4, V(w1)^4, V(w2)^2, V(w3). CF print V(w1)^4 and V(w1)^5 respectively. The only discrepancy is the MC2 reversal; it is not a new defect. Cross-check with dim V(w1)=2g, dim V(w2)=2g^2-g-1 (odd) or 2g^2-g-2 (even), dim V(w3)=C(2g,3)-4g (odd) or C(2g,3)-2g (even): the corrected lists total exactly 4g^2+4g+C(2g,3) in both parities, which is Sato's log_2 order. The printed lists total 4g^2+2g+C (odd) and 4g^2+6g+C (even). Period-2 multiplicities therefore survive for the Z[G]-composition series, but this is a different statement from Theorem 8.4, and Conditions I and II are unverified.
(8) Odd primes. The inference from 'extension of F_p-modules' to 'F_p-module' is invalid in general (Z/p^2 is a counterexample), so CF's 'it follows' is a gap. NEW REPAIR, separate from CF's proof. Hypotheses: p odd, and an equivariant exact sequence 0 -> S -> A -> Q -> 0 with Q = H_1(Gamma^{Sp}{2g}(p);Z) of exponent p and S the image of the Torelli abelianization, identified equivariantly with wedge^3 H_p. Multiplication by p kills S and lands in S, so it induces an equivariant f: Q -> S. A lift of -I acts trivially on Q, because -I is central in Sp{2g}(Z), and acts by -1 on wedge^3 H_p. Hence f = -f, so 2f = 0, so f = 0 since p is odd, and pA = 0. The argument fails at p=2, as it must. So for odd p the F_p-structure holds conditional on Putman 7.8 with its equivariance. The odd-p multiplicities, corrected as in MC2, are periodic with period p. Conditions I and II are not addressed in CF's proof; see MC2.
Premises used
- Sato, arXiv:0804.4789v1. available_from: 30 April 2008. date_evidence: packet label 'Sato-2008-text: arXiv0804.4789v1, 30 April 2008' and the recorded review policy; the arXiv stamp is not visible on the supplied pages. date_basis: preprint version date as recorded. version: v1. Used: Cor 2.2 (g>=2: H_1(Gamma_g[d];Z) = Z_d^{2g^2+g} for d odd, Z_d^{2g^2-g} + Z_{2d}^{2g} for d even) and Prop 4.9 (g>=3: H_1(M_{g,1}[2];Z) = Z_8^{2g} + Z_4^{C(2g,2)} + Z_2^{C(2g,3)}), with the p.2 definitions. Statements read; proofs not audited. Before the 2012-01-19 cutoff.
- Putman, arXiv:0908.0555v3. available_from: 31 August 2011. date_evidence: packet label and recorded review policy. date_basis: preprint version date as recorded. version: v3. Used: Cor 3.6 (g>=3, L even, exact sequence of Sp_{2g}(Z)-modules with integral H_1) and Lemma 6.9 for context only. Theorem 7.8 is NOT supplied and is used only conditionally, as paraphrased by CF and constrained by the order comparison with Sato.
- Elementary and checked here: block multiplication for the Z/4 character; right-exactness of tensoring with Z/2; Jordan-Holder for finite Z[G]-modules; the conjugation action of Mod_{g,1}/Mod_{g,1}(2) on the abelianization.
- Hogeweij 1982 (Cor 2.7, Lemma 2.10) and Shvartsman 2009, as quoted in CF-modular lines 4163-4181 and 4279-4289. available_from: 1982 and 2009. date_evidence: CF bibliography. date_basis: cited publication years. version: journal articles, not inspected. Used only for the sp_{2g}F_2 and B_2 lists; their dimension totals were checked for consistency.
- MC2's corrected exterior-cube list (see MC2 for its own premises).
Missing inputs
- The exact statement of Putman v3 Theorem 7.8: rank range, hypothesis on L (in particular whether L=2 is covered and for which g), the kernel term for even L, and module-equivariance.
- Sato's odd-level theorem (p.3), which would give the F_p-structure for odd p directly.
- CF's own definition of S_{g,1} and Mod_{g,1} (omitted section).
- For any F_2-coefficient replacement at p=2: which integral layers survive in A/2A, and the stabilization maps. Dimension counts are consistent with A/2A being an extension of sp_{2g}F_2 by wedge^3 H and 2A carrying [H]+[B_2/F_2], but that is an unverified guess.
Scope changes
- Sato's and Putman's proofs were not audited; only their statements were read. Putman Theorem 7.8 was not available.
- The Z/4-character argument was supplied by the Sol assessment and rechecked here; it is not an independent discovery.
- The mod-2 congruence homology sentence was checked only in the arXiv v3 TeX; journal pp.309-310 were not supplied.
- Conjecture 8.5 inherits the integral-coefficient convention problem (H_0 = Z is not an F_p-space). This is noted as downstream of the same defect and not examined further.
Located concern
CF-modular TeX lines 4310-4318 and 4342-4355; journal-modular-p62 (Theorem 8.4 and the sentence before it)
False literal statement at p=2: H_1(Mod_{g,1}(2);Z) is not an F_2-vector space, so it is not a sequence of F_2-representations in the sense of Definition 8.1.
Sato Prop 4.9 gives a Z_8^{2g} summand for g>=3. Independently, f_i(I+2X) = B_ii + 2 sum_j A_ij B_ij mod 4 is a homomorphism Gamma^{Sp}{2g}(2) -> Z/4 taking the value 1 on I+2E{i,g+i}; pulling back along the surjection from Mod_{g,1}(2) gives a Z/4 quotient for all g.
Restrict the F_p-module theorem to odd p. For p=2 state a separate result, either on Jordan-Holder factors of the finite Z[Sp_{2g}(F_2)]-module (corrected list 2,5,2,1 for g odd and 4,4,2,1 for g even, conditional), or on H_1(-;F_2) after actually computing A/2A.
Theorem 8.4 is false as stated at p=2 for every g. Periodicity of composition multiplicities of the integral group survives conditionally in a different category.
Located concern
CF-modular TeX lines 4157-4159, reused at 4342-4345
Coefficient conflation: Putman's integral extension is attributed to F_2-homology. This is the same root defect as the previous issue and should not be counted separately.
Putman Cor 3.6 is for H_1(Sp_{2g}(Z,L);Z). Sato Cor 2.2 gives dim H_1(Gamma_g[2];F_2) = 2g^2+g, whereas the claimed extension has dimension 2g^2+3g. In fact H_1(Gamma^{Sp}{2g}(2);F_2) is sp{2g}(F_2) and the kernel H is 2A.
State the extension for integral H_1 and state H_1(Gamma^{Sp}{2g}(2);F_2) = sp{2g}(F_2) for g>=3. In the Theorem 8.4 proof write H_1(Gamma^{Sp}_{2g}(2);Z).
The period-2 multiplicity pattern for {H_1(Gamma^{Sp}{2g}(2);F_2)} is unaffected, since it comes from V(0) in sp{2g}F_2. The V(w1) from H does not occur in mod-2 homology. In Theorem 8.4 the two misstatements cancel in the factor count.
Located concern
CF-modular TeX lines 4310-4312 (odd p)
Invalid inference for odd p: an extension of F_p-modules need not be an F_p-module.
0 -> Z/p -> Z/p^2 -> Z/p -> 0.
New repair: the central element -I acts by +1 on H_1(Gamma(p);Z) and by -1 on wedge^3 H_p, so the equivariant map induced by multiplication by p vanishes for odd p. Alternatively cite Sato's odd-level computation (statement not supplied).
For odd p the conclusion is true conditional on the equivariant extension; this is a missing argument, not a false statement.
MC2
Checked statement: The exterior-cube table in the proof of Theorem 8.4 (wedge^3 H has factors V(w1),V(w3) for g=1 mod p and V(w1),V(w3),V(w1) otherwise); the odd-p and p=2 lists derived from it; and the prior report's correction (3), (4) with its map-condition repair.
Returned assessment: claim_false
The printed table is reversed, and the prior report's corrected class (3) is right. However, its derivation skips a bridging step and its dimension check is not independent. Conventions: H of dimension 2g over F_p, omega = sum a_i^b_i, delta(x^y^z) = w(x,y)z - w(x,z)y + w(y,z)x, epsilon(x) = x^omega.
(a) The identity. delta(x^a_j^b_j) is x if x lies outside handle j and 0 if x is a_j or b_j. Summing over j gives delta(epsilon(x)) = (g-1)x. For g>=2, delta is onto and epsilon is injective.
(b) What the identity alone proves. If g = 1 mod p, then 0 < epsilon(H) <= ker(delta) < wedge^3 H, so H occurs at least twice. This refutes the printed g = 1 row for arbitrarily large g (for example p=2 with g odd, or p=3 with g=7). If g is not 1 mod p, then wedge^3 H = epsilon(H) + ker(delta) as a direct sum. The identity gives nothing more: it does not bound the multiplicity of V(w1) inside ker(delta) and does not show the remaining quotient is simple. The exact lists need that extra input.
(c) The imported input, with exact hypotheses. Baranov-Suprunenko Theorem 2.1 (attributed to Premet-Suprunenko, stated for p>2) says W_i^n ~ sum_k d_k^{n+1-i+2k} omega^n_{i-2k}. Here G_n = Sp_{2n}(K) with K algebraically closed of characteristic p, d^m_k = 1 exactly when each base-p digit of k is 0 or equals that of m, and terms with negative index are omitted. Cor 2.9 states that 2.1 and 2.4 also hold in characteristic 2; it rests on Theorems 2.3/2.4 on an unsupplied page. For i=3, n=g>=3: k=0 gives omega_3; k=1 gives omega_1 exactly when d_1^g = 1, that is g = 1 mod p; k>=2 is omitted. So [W(w3)] = [V(w3)] + [g = 1 mod p][V(w1)].
Bridge omitted by the prior report: the theorem concerns the Weyl module, not wedge^3 H. One needs [wedge^3 H] = [W(w3)] + [W(w1)] in the Grothendieck group, which holds because formal characters are characteristic-free and determine composition factors, and W(w1) = H is simple since its weights form one Weyl orbit. One also needs passage from Sp_{2g}(F_p-bar) to Sp_{2g}(F_p): w1 and w3 are p-restricted for every p, by the Steinberg restriction statement CF quote from Humphreys. With these, [wedge^3 H] = [V(w3)] + (1 + [g = 1 mod p])[V(w1)] for g>=3. Sanity check at i=2: W(w2) = V(w2) + [p | g] V(0), matching CF's parity statement for wedge^2 H.
(d) NEW REPAIR removing the import (it follows the Sol sketch, with the algebraic-group step made explicit). Work over F_p-bar with g>=3, K = ker(delta), Q = K/(K meet epsilon(H)). Submodules for the algebraic group are sums of weight spaces. The weights of wedge^3 H are +-L_i+-L_j+-L_k with i,j,k distinct (multiplicity 1) and +-L_i (multiplicity g-1, spanned by a_i^a_j^b_j). Let s be the symplectic shear a_j -> a_j+a_k, b_k -> b_k-b_j. Then s(a_i^a_j^b_k) - a_i^a_j^b_k = -a_i^(a_j^b_j - a_k^b_k) - a_i^a_k^b_j, so the three-handle weight vectors generate K. Let M be a nonzero submodule of Q. If M contains a three-handle weight vector, the signed-permutation Weyl group gives all of them and M = Q. Otherwise M contains the class of v = a_i^sum c_j a_j^b_j with sum c_j = 0 that is not a multiple of epsilon(a_i), so c_j differs from c_k for some j,k. Then s(v)-v = (c_j - c_k) a_i^a_k^b_j is a nonzero three-handle vector in M. Hence Q is simple with highest weight w3 (vector a_1^a_2^a_3). Its restriction to Sp_{2g}(F_p) is simple because w3 is restricted. The torus argument is not valid directly over F_2, where the split torus of the finite group is trivial; this is why the algebraic group is used. Combined with (a): for g not 1 mod p, K is V(w3) and there is one V(w1); for g = 1 mod p, K/epsilon(H) is V(w3) and there are two.
(e) The g=5, p=2 evidence. dim H = 10, dim wedge^3 H = 120, dim K = 110, and since 5 = 1 mod 2, dim V(w3) = 100. The corrected row gives 100+20 = 120 and the printed row gives 110. The arithmetic is right. But dim V(w3) = 100 comes from the same theorem (or from (d)), so this is a consistency illustration and not independent evidence, and one finite case proves nothing universal. Independent cross-checks: Sp_6 at p=2 gives dim V(w3) = 14-6 = 8; and under the Z[G]-reading of MC1 the corrected p=2 lists reproduce Sato's order exactly while the printed ones are off by a factor 2^{2g} in opposite directions.
(f) Corrected lists and their status.
- wedge^3 H, g>=3, all p: V(w3) + 2V(w1) if g = 1 mod p; V(w3) + V(w1) otherwise. Established by (d), or by Baranov plus the bridge.
- Odd p, g>=5: [H_1(Mod_{g,1}(p);Z)] = V(2w1) + V(w3) + (1 + [g = 1 mod p]) V(w1). Conditional on Putman 7.8 (unsupplied), Hogeweij (sp simple of weight 2w1 for p odd) and the F_p-structure from MC1(8).
- p=2: Z[G]-composition list as in MC1(7). Conditional, and not a statement about an F_2-representation. Period p in the V(w1) multiplicity survives; the residue class is simply the opposite one.
(g) Map conditions. CF's proof ends with 'periodic and uniformly multiplicity stable' and never addresses Conditions I and II; multiplicity counts cannot establish them. The prior report's odd-p repair is logically sound given its hypotheses. Hypotheses: the extension is natural under the stabilization maps, with outer maps the standard inclusions; g>=3; p odd. Then the four lemma gives injectivity. The G-span of the image contains wedge^3 H_{g+1}, because basis wedges involve at most three handles (equivalently, any three vectors lie in a nondegenerate subspace of dimension at most 6, which is G-conjugate into H_g by Witt). It also maps onto sp(H_{g+1}), because basis elements of Sym^2 involve at most two handles (or by simplicity of sp). So it is everything. Naturality is not contained in any supplied statement, so I and II remain unverified for odd p and unformulated for p=2.
Premises used
- Baranov-Suprunenko, arXiv:math/0003210v1. available_from: 30 March 2000. date_evidence: packet label 'Baranov-2000-text: arXiv math/0003210v1, 30 March 2000' and the recorded review policy. date_basis: preprint version date as recorded. version: v1. Used: Theorem 2.1 (p>2; composition factors of fundamental Weyl modules of Sp_{2n}(K), K algebraically closed of characteristic p) and Cor 2.9 (validity in characteristic 2). Statements read; proofs not audited; p.1 definitions and Theorems 2.3/2.4 not supplied. Before the cutoff.
- Steinberg restriction for p-restricted weights and the highest-weight classification, as stated in CF-modular lines 4085-4133 citing Humphreys. available_from: 2006. date_evidence: CF bibliography. date_basis: cited edition year. version: book, not inspected.
- Characteristic-free formal characters (wedge^3 H has the character of W(w3)+W(w1)) and linear independence of simple characters. Standard; used only for the bridge in (c) and not needed for the direct repair (d).
- Putman v3 Theorem 7.8 as paraphrased by CF (unsupplied; conditional) and Hogeweij Cor 2.7 as quoted by CF, for the odd-p list; dates as in MC1.
- Direct calculations checked here: delta(epsilon(x)) = (g-1)x, the shear computation, and the dimension counts.
Missing inputs
- Baranov-Suprunenko p.1 (precise definitions of W_i^n and omega_i^n, lower bound on n) and p.4 (Theorems 2.3, 2.4) to confirm the characteristic-2 extension independently of the wording of Cor 2.9.
- Putman Theorem 7.8 with naturality under the genus-stabilization maps and equivariance of the kernel identification.
- CF's Definition repstab1 (exact wording of Conditions I and II and the stabilization maps) from the omitted section.
Scope changes
- The direct simplicity argument (d) is a new repair, valid for g>=3 and all primes, over the algebraic closure and then restricted to Sp_{2g}(F_p). It is separate from CF's proof and from the Baranov-Suprunenko citation.
- The Baranov-Suprunenko, Premet-Suprunenko and Adamovich proofs were not audited.
- No general audit of Section 8, of Theorem 8.3, or of Hogeweij's and Shvartsman's lists was attempted beyond dimension consistency.
Located concern
CF-modular TeX lines 4323-4329; journal-modular-p62 first display in the proof of Theorem 8.4
False table: the residue classes are interchanged.
delta(epsilon(x)) = (g-1)x. For g = 1 mod p, epsilon(H) lies in ker(delta) and wedge^3 H / ker(delta) is H, giving two copies of V(w1) where the table lists one (p=2, g=5: 0 < H < K < wedge^3 H with dimensions 10, 110, 120). For g not 1 mod p, wedge^3 H = H + K with K simple by (d), giving one copy where the table lists two (p=3, g=5).
V(w3) + 2V(w1) for g = 1 mod p; V(w3) + V(w1) otherwise (g>=3).
The multiplicity tables at lines 4336-4339 and 4350-4352 are wrong in the V(w1) exponent only. The period-p pattern is unchanged. This is one defect propagated to three displays.
Located concern
prior-modular lines 360-376
Gaps in the criticism, which do not defeat it: the identity (2) alone does not yield the list (3); the Weyl-module theorem is applied to wedge^3 H without the bridge [wedge^3 H] = [W(w3)]+[W(w1)]; and the algebraically-closed-field hypothesis, the transfer to the finite group, and the dependence of p=2 on Cor 2.9 are not stated.
From (2) one gets only multiplicity at least 2 at g = 1 mod p and a direct-sum splitting otherwise. Baranov-2000-p3 states 'we assume K = F = F-bar' and Theorem 2.1 carries 'Let p > 2'.
Add the character bridge and Steinberg restriction, or use the self-contained weight argument (d), which covers all p and g>=3.
The corrected list (3) stands. Its justification in the prior report is incomplete but repairable.
Located concern
prior-modular line 378
The g=5, p=2 dimension check is circular as evidence.
dim L(w3) = 100 is itself obtained from the corrected composition formula (110 - 10); no independent source for it is given.
Present it as a consistency illustration. For an independent check use the two-step filtration from (2), or the order comparison with Sato Prop 4.9 in MC1(7).
None on the correction; it only limits the evidential weight of that paragraph.
Located concern
CF-modular TeX lines 4354-4355; prior-modular lines 388-394
Conditions I and II are not verified in CF's proof, and the prior report's verification assumes naturality of the extension.
CF conclude only 'multiplicity stable'. Equal or periodic lists do not even show the stabilization maps are nonzero.
For odd p: state naturality of Putman's sequence under stabilization, then use the four lemma and the at-most-three-handles and at-most-two-handles orbit-span argument (g>=3). For p=2 first choose the category (Z[G]-filtration or F_2-homology).
Periodic stability in the full sense of Definition 8.1 is unverified even for odd p. Only the corrected periodic multiplicities are established, conditionally.
Outside the audit
- Proofs of Sato Cor 2.2 and Prop 4.9; proofs of Putman Cor 3.6 and Lemma 6.9; statement and proof of Putman Theorem 7.8.
- Proofs of Baranov-Suprunenko Theorem 2.1 and Cor 2.9, and the cited Premet-Suprunenko and Adamovich results.
- Hogeweij, Shvartsman, Humphreys, Hain and Johnson, used only as quoted by CF.
- Composition factors and stabilization maps of H_1(Mod_{g,1}(2);F_2).
- Naturality of the level-p extensions under genus stabilization, hence Conditions I and II for any p.
- Theorem 8.3, Conjectures 8.2 and 8.5, and the SL_n material of Sections 8.1-8.2, beyond remarks needed for the two targets.
Additional supporting archive items are identified in the original report. View the manuscript hashes and repository snapshot.