claude-fable-5-2 high: bw-density
Reviewer record, not the publication verdict. The response includes unaccepted proposals and source limits. Missing antecedent proofs are not automatically defects in the paper. The initial request for Lemmas 8.4–8.5 was our inventory error; the journal ends at Lemma 8.3. The curated report records final dispositions.
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 principal theorem and no obstruction to Theorem 5.1 or to the ergodicity and Kolmogorov conclusions was found. One auxiliary statement is false as literally written: Proposition 2.4 (p. 462) omits positivity of the base measure and nestedness, and a single-leaf sequence is a counterexample; adding the hypothesis fixes it and all uses satisfy it. Two proof steps are not valid as printed but are repairable with the stated new arguments: the fake-holonomy sentence in Lemma 6.2 (p. 478), where a genuine stable leaf is treated as lying in a fake cs-leaf, and the lower bound in the F_n/G_n half of Lemma 8.2 (p. 485), which needs a cone argument rather than the triangle inequality. The p. 474 remark that only the first bunching inequality is used for stable saturation is inaccurate, since Prop. 5.4(4) uses the second via Prop. 3.1(vii); this is harmless because both are hypotheses. Remaining items are interface or typographical: Lemma 1.1's last display; 'X inside U' in Prop. 2.5; Lemma 4.4 applied to a function on N rather than M; W^s(p,1) versus W^s(x, sigma_n) domains; C^1 versus 'smooth' transversals. The conclusions are conditional on inputs not supplied or not dated before the 11 Oct 2005 cutoff: absolute continuity of W^s, W^u for non-constant rates ([AV08] is post-cutoff); Prop. 3.1(vii) via [PSW97, PSW00] in the C^2 branch and via the unsupplied [BW06b] in the C^{1+delta} branch; and [BP74, Prop. 5.1] beyond its constant-rate setting for the Kolmogorov clause. The target's Lemmas 8.4-8.5 do not exist in the supplied sources. This is a scoped audit, not a complete line-by-line verification; no web search, code execution or other sources were used, and later knowledge in model weights cannot be excluded but was not used to close any obligation.
D1
Checked statement: Journal Sections 1.3-2.3 (printed pp. 458-464): inequality (9) and Lemma 1.1; definitions of density point, nesting, Lebesgue density sequence, regular and internested sequences; Lemma 2.1 and Corollary 2.2; the two absolute-continuity notions (11), (12); bases, fibers and c-uniformity; Propositions 2.3, 2.4, 2.5, Corollary 2.6 and Proposition 2.7.
Returned assessment: repairable_gap
CHECKED DIRECTLY. (a) Eq. (9) is attainable: after rescaling so that R-balls are small, leaf and ambient distances agree up to a factor near 1, and the strict inequality (2) plus continuity of nu on compact M leaves a uniform margin. f(W^s_loc(p)) inside W^s_loc(f(p)) follows from (9) with q'=p and connectedness. (b) Lemma 1.1: the induction is correct once the hypothesis carries the membership q'k in W^s_loc(q_k), as the proof does. The last display should read nu{k+1}(p), not nu(p_{k+1}) (my reading of p. 458, matching the author text). (c) Lemma 2.1: for n>=k, m(Z_{n+1}) >= m(Y_{n+1+k}) >= delta^{2k+1} m(Y_{n-k}) >= delta^{2k+1} m(Z_n). For n<k one needs m(Z_0) finite, which holds for volume on compact M and for the leaf measures used later. The density equivalence follows from m(X^c:Z_n) <= delta^{-2k} m(X^c:Y_{n-k}) and its symmetric counterpart. The argument is measure-agnostic, which matters because Theorem 5.1 uses it for hat-m_cu. (d) Internestedness is an equivalence relation; Corollary 2.2 follows. (e) Prop. 2.3: correct. (f) Prop. 2.7: (13), (14) and the division are correct. Used implicitly: m(Y_n)>0 (so tau_{Y_n} is nonempty and m_n is finite and positive), and tau_{Y_n} is the disjoint union of its intersections with tau_X and tau_{X*}, which holds because X is a union of leaves. (g) Prop. 2.5: the chain is correct and never uses X inside U. (h) Cor. 2.6 follows from 2.5. (i) Independence of density points from the Riemannian metric, and balls B(x, rho^n) as a density sequence: correct. FINDINGS. Prop. 2.4 is false as literally stated (positivity and nesting are not derived); the rest are precision or interface points listed below. None affects the later use of these results. NOT CHECKED. The imported absolute-continuity theorems for W^s and W^u. The p. 459 remark that every Lebesgue density sequence is regular; I found no short proof, and the paper does not use it because regularity is built into the bi-essentially-saturated definition. The converse half of the p. 460 claim that (11) and (12) are equivalent; the author text asserts the opposite, citing Brin's Remark 3.9, and only the direction (11) implies (12) is used.
Premises used
- Compactness of M; continuity of nu, hat-nu; strict inequalities (1)-(4)
- Definitions on pp. 459-461 exactly as printed
- Import (conditional): transverse absolute continuity with bounded Jacobians of W^s, W^u. Constant-rate case: Anosov 1967, Pugh-Shub 1972, Brin-Pesin 1974; available_from 1967-1974; date_evidence journal reference list p. 488; date_basis publication year; version journal articles; exact theorem numbers not given in the paper.
- Import (conditional): transverse absolute continuity implies the Fubini form (12). The author text cites Brin-Stuck, Introduction to Dynamical Systems, 2002; available_from 2002; date_basis publication year; the journal version gives no citation ('not difficult to verify').
Missing inputs
- A dated, pre-11-Oct-2005 statement of absolute continuity with bounded Jacobians for W^s, W^u when nu, hat-nu, gamma, hat-gamma are non-constant functions and f is only C^{1+delta}. The paper says this 'follows from Pesin theory' without a theorem. The direct proof it cites is [AV08], listed on p. 488 as a 2008 preprint, which is after the cutoff and cannot close the obligation.
- Explicit statement that the transversals in (11), (12) may be C^1, since the fake leaves used later as transversals are only C^1 (Prop. 3.1(i)).
Scope changes
- The target names 'Propositions 2.1-2.5'. In the journal these are Lemma 2.1, Corollary 2.2 and Propositions 2.3-2.5; I also audited Corollary 2.6 and Proposition 2.7 (p. 463) because the proof of 2.5 on p. 464 depends on 2.7.
- The imported absolute-continuity theorems were not audited; they are not supplied.
Located concern
BW journal p. 462, Proposition 2.4 and its proof
False auxiliary statement as literally quantified. The proof derives only m(Y_{n+1}) >= (cC)^{-2} delta^2 m(Y_n). 'Regular' (p. 459) also requires a nested sequence and m(Y_n)>0 for all n. Neither is assumed or derived.
Let F be any foliation satisfying the hypotheses, with codimension at least 1. Fix q0 in tau and let Y_n = F(q0, 2^{-n}) inside the single local leaf through q0. Then tau_{Y_n} = {q0}, the fibers are 1-uniform, (i) reads 0 >= delta*0, and (ii) holds for some delta>0. But m(Y_n)=0 for all n, so Y_n is not regular.
Add the hypotheses that Y_n is nested and measurable and that m_tau(tau_{Y_0})>0. Then (i) gives m_tau(tau_{Y_n})>0 for all n and Prop. 2.3 gives m(Y_n)>0.
None downstream. In both uses (hat-J^cu_n(x) on p. 481, G_n(x) on p. 486) the sets are nested with positive-measure bases.
Located concern
BW journal p. 458, last display of the proof of Lemma 1.1
Typographical error: the final equality gives nu(p_{k+1}) d(q,q') where nu_{k+1}(p) d(q,q') is meant. Also the membership q'{k+1} in W^s_loc(q{k+1}) comes from f(W^s_loc) inside W^s_loc, not from the distance inequality (9) itself.
nu(p_k) nu_k(p) = nu_{k+1}(p) by the cocycle definition of Section 1.1; nu(p_{k+1}) is a single factor.
Replace by nu_{k+1}(p) and state the inductive hypothesis with local-leaf membership.
None.
Located concern
BW journal p. 462, Proposition 2.5 ('essentially F-saturated set X contained in U'); p. 463 Cor. 2.6; use on p. 484
Mis-scoped hypothesis. An essentially F-saturated set (F-saturated means a union of entire leaves, p. 458) is essentially never contained in a small foliation box. The applications take X essentially W^u-saturated in M, not inside U.
On p. 484 Prop. 2.5 is applied to X essentially W^u-saturated in M, with C_n(x), D_n(x) shrinking to x.
Delete 'X contained in U'. The printed proof on p. 464 never uses it and proves the statement for any essentially F-saturated measurable X in M.
None; statement/proof interface only.
Located concern
BW journal p. 461, first paragraph; reference [AV08] on p. 488
Imported absolute continuity is stated for 'a partially hyperbolic diffeomorphism' with no regularity hypothesis, although it needs at least C^{1+delta}. For non-constant rate functions the only precise source is a 2008 preprint, later than the 11 Oct 2005 cutoff.
p. 461: the proofs of [BP74], [PS72] 'were written under the assumption that the functions ... are constant'; the general case 'follows from Pesin theory' with no theorem cited; the direct proof is [AV08].
State the C^{1+delta} hypothesis and supply a dated pre-cutoff theorem, or a proof, for the pointwise-rate case; otherwise treat it as a conditional input.
Everything using (11) or (12) is conditional on this import in the pointwise-rate setting. In the constant-rate setting the classical references apply (not re-verified here).
Located concern
BW journal pp. 460-462: definitions (11), (12) and 'foliation box U' / 'smooth transversal' in Props. 2.3-2.7; p. 460 Lemma 2.1
Interface conventions left implicit. (i) The definitions give constants only for boxes of diameter less than R_0, while Props. 2.3-2.7 take an arbitrary box. (ii) 'Smooth' transversals are later the C^1 leaves fake W^cs(x) and fake W^cu(x). (iii) Lemma 2.1 and Prop. 2.4 are stated for m on M but applied on p. 477 and p. 481 inside fake W^cu(x') with hat-m_cu and to the fake foliation hat-W^u.
p. 477: 'Working inside of hat-W^cu(x'), we will apply Lemma 2.1'. p. 481: Prop. 2.4 is applied to hat-W^u as a subfoliation of hat-W^cu(x).
State the results for boxes of diameter less than R_0, for C^1 transversals, and for an arbitrary finite measure with the Fubini property. The printed proofs go through unchanged.
None once stated; all applications are at arbitrarily small scale near x.
Located concern
BW journal p. 459 ('It is easily shown that a Lebesgue density sequence Y_n must be regular'); p. 460 ('these two properties are equivalent')
Two unproved side remarks that I could not verify quickly. The second contradicts the 2006 author text, which says the converse fails.
For the first, the natural test sets built from a non-regular subsequence give no contradiction without a multi-scale argument. For the second, I have only a sketch (saturate a set S in tau_1 by leaf plaques of fixed radius and apply (12) to both transversals), and I did not check measurability or box-shape details.
Supply proofs or delete. Neither remark is used.
None.
D2
Checked statement: Propositions 5.3, 5.4, 5.5 (p. 476) and their proofs: Section 6 (Lemmas 6.1, 6.2, proof of 5.3), Section 7 ((16), Lemmas 7.1, 7.2, the four conclusions of 5.4), Section 8 (sets B_n through G_n, Lemmas 8.1, 8.2, 8.3, the seven-step chain). Includes the choice of sigma, tau and the julienne definitions on pp. 474-475.
Returned assessment: repairable_gap
RESULT. Propositions 5.3, 5.4, 5.5 hold as stated, conditional on Prop. 3.1, Lemmas 4.2-4.4 and absolute continuity of W^s, W^u. Two printed steps are not valid as written and need the repairs below (Lemma 6.2's holonomy sentence; the lower bound in the F_n/G_n half of Lemma 8.2). CHECKED. (1) p. 474: sigma, tau exist because nu/gamma < min(hat-gamma, 1), which uses only nu < gammahat-gamma and (1). Also tau < sigmagamma < gammahat-gamma < 1, and tauhat-nu < sigma uses hat-nu < hat-gamma from (1). (2) Juliennes are nested, since f^{-1} contracts both hat-W^u and W^u plaques. (3) Lemma 6.1: x, x' lie in W^s(p) = hat-W^s(p), so both center plaques lie in hat-W^cs(p). Prop. 3.1(vii), cs-part, gives a C^1 hence uniformly L-Lipschitz hat-W^s holonomy between them. The condition 'sigma_{-k1} > L' must be read at p_n, i.e. sigma_{n-k1}(p)/sigma_n(p) > L; this holds uniformly once (sup sigma)^{-k1} > L. (4) Lemma 6.2: the conclusions d(y'_n, z'n) = O(tau_n) and d(q', z') = O(tau_n/gamma_n) = O(sigma_n) are correct, but see issue 1 for the justification. The pull-back factor used, gamma_n(p)^{-1}, is the correct one. (5) Proof of 5.3: h^s(y) is exactly the y' of Lemma 6.2. Q lies in hat-B^c{n-k}(x') by the triangle inequality in one hat-W^c leaf, and k is uniform in x, x'. (6) Section 7: (16) follows from uniform C^1 size of hat-W^u plaques. Lemma 7.1 follows from the Lemma 4.1/4.4 argument because f^j(T_n) lies in B(p_j, C kappa^j), which I re-derived from sigma < hat-gamma and tau < 1; see issue 4. Lemma 7.2 holds with C_3 = C_2. Conclusions 1-3 of 5.4 are correct. Conclusion 4 follows from Prop. 2.4 applied to hat-W^u inside hat-W^cu(x) with base hat-B^c_n(x), using Prop. 3.1(vii), cu-part; see issue 3. (7) Section 8, in order. B_n is a density sequence and regular. D^cs_n(x) internests with hat-W^cs(x, sigma_n), and C_n with B_n, by uniform local product structure. C_n and D_n have common base D^cs_n(x) with W^u-fibers W^u(q, sigma_n) and J^u_n(q), c-uniform by Prop. 5.4(2), so Prop. 2.5 applies with genuine W^u. Lemma 8.1 is correct; the claim that hat-q_n is the intersection point at time n is justified by constructing that point at time n and pulling it back, using backward contraction of hat-W^u and two-sided local invariance of hat-W^cs. E_n/F_n internest by Prop. 5.3 with uniform k. F_n/G_n: see issue 2. Lemma 8.3 follows from Prop. 2.4 with genuine W^s over the C^1 transversal hat-W^cu(x). The last equivalence follows from Prop. 2.7 with tau_X = X cap hat-W^cu(x), since X is W^s-saturated. Regularity passes from G_n to F_n, E_n, D_n by Lemma 2.1. NOT CHECKED. Prop. 3.1 and Lemmas 4.1-4.4 (journal pp. 465-470 not supplied as images; read only in the author text). Measurability of C_n through G_n was not verified line by line. Uniform local product structure constants for mixed genuine/fake pairs (W^u with hat-W^cs, W^s with hat-W^cu) are taken from the p. 467-type standing assumption in the author text.
Premises used
- Prop. 3.1(i)-(v) (C^1 f): cone condition, two-sided local invariance on r_1-plaques, growth bounds, coherence, hat-W^s(p) = W^s(p,r). Author text only; journal pp. 465-467 not supplied.
- Prop. 3.1(vi) (C^{1+delta}): uniform Holder continuity of the fake distributions; used in Lemma 7.1.
- Prop. 3.1(vii), cs-part (hat-W^s is C^1 inside hat-W^cs): used in Lemma 6.1. C^2 branch needs nu < gamma*hat-gamma via [PSW97, PSW00]; C^{1+delta} branch needs the strong bunching inequality via [BW06b].
- Prop. 3.1(vii), cu-part (hat-W^u is C^1 inside hat-W^cu): used in Prop. 5.4 conclusion 4. Needs hat-nu < gamma*hat-gamma (C^2) or its strong version (C^{1+delta}).
- Lemmas 4.2-4.4 (author text): f^j(T_n) inside B(p_j, C kappa^j); Holder cocycle distortion.
- Lemma 1.1 for contraction along genuine W^s and W^u.
- Lemma 2.1, Props. 2.4, 2.5, 2.7 as repaired in D1.
- Absolute continuity with bounded Jacobians of genuine W^s and W^u (import; see D1).
Missing inputs
- Journal pages 465-470 (Prop. 3.1(iii)(b)-(vii), Section 4). In the author text, Prop. 3.1(iii)(b) prints the cu bound as gamma_{-n}(p)^{-1} d(q,q'); the one-step bound in its own proof gives gamma_{-n}(p) d(q,q'). Lemma 6.2 uses the correct form. I could not confirm how the journal prints this.
- [PSW97, PSW00] and [BW06b] statements giving Prop. 3.1(vii) for the fiberwise map F_r on TM; not supplied.
Scope changes
- The target names Lemmas 8.1-8.5. The supplied journal pages and the author text contain only Lemmas 8.1, 8.2, 8.3 in Section 8; no Lemma 8.4 or 8.5 exists there. I audited 8.1-8.3 and the unnumbered steps of the chain on pp. 481-486.
- Prop. 3.1 and Lemmas 4.1-4.4 are prerequisites outside the assigned pages. They were read in the author text only and used conditionally, apart from the consistency checks noted.
Located concern
BW journal p. 478, proof of Lemma 6.2, sentence 'Note that q_n and z'_n are respectively the images of y_n and y'_n under hat-W^u-holonomy between hat-W^cs_loc(y_n) and hat-W^cs(q_n)'
Unjustified, and in general incorrect, auxiliary claim. y'_n lies on the genuine stable leaf of y_n. For y off the orbit of p there is no reason for a genuine stable leaf to lie in the fake leaf hat-W^cs(y_n); Prop. 3.1(v) gives coincidence only along the orbit of p, and p. 473 stresses that genuine and fake saturation are unrelated. Two smaller points: the bound d(y_n, y'_n) = O(nu_n) is credited to Prop. 3.1(iii.a), which concerns fake leaves; and 'it is not hard to see' that z'_n is the stated intersection point is not automatic, because hat-W^u pieces expand under forward iteration.
The fake cs-leaf through y_n is determined by the forward dynamics of the modified map F_r. The genuine stable leaf is determined by the f-orbit of y_n, which leaves the neighbourhood N after finitely many steps. Their tangent spaces need only agree to within the epsilon-cone.
NEW REPAIR (not in the paper). Define w as the unique point of hat-W^u(y'_n) cap hat-W^cs(q_n). It exists because q_n lies in hat-W^cs(q_n) and d(y'_n, q_n) <= d(y'_n, y_n) + d(y_n, q_n) = O(nu_n + tau_n) = O(tau_n), with d(y_n, y'_n) <= nu_n d(y, y') from Lemma 1.1. Uniform transversality of the epsilon-cones gives d(y'_n, w) <= C d(y'_n, q_n) = O(tau_n). The leaf hat-W^cs(q_n) equals hat-W^cs(q'n) and, by coherence and y'n in hat-W^cu(q'n), w lies in hat-W^c(q'n). Pull w back n steps: hat-W^u contracts, and the hat-W^c distance to q'{n-j} is at most tau_n / (gamma(p{n-j})...gamma(p{n-1})) <= tau{n-j}, since tau < sigma*gamma and sigma < 1. So the backward orbit stays in the r_1-plaques, local invariance applies, and f^{-n}(w) = z' by uniqueness. Hence z'_n = w and the printed estimates follow.
Lemma 6.2 and Proposition 5.3 stand; only the justification changes.
Located concern
BW journal p. 485, proof of Lemma 8.2, second paragraph (F_n and G_n)
Two defects. (i) q' = h^s(q) is asserted to lie in hat-J^cu_n(x'); Prop. 5.3 gives only hat-J^cu_{n-k}(x'). (ii) The lower bound sigma_{n+k} <= d(q, q') is drawn from 'diameters are O(sigma_n) and d(x,x') = sigma_n'. That does not follow: the triangle inequality gives d(q,q') >= sigma_n - d(q,x) - d(q',x'), and d(q,x) alone can be about sigma_n because the julienne contains the center plaque of radius sigma_n.
The three metric facts quoted are consistent with q = q' in an abstract metric space. A directional argument is needed.
NEW REPAIR (not in the paper). Work in exponential coordinates at p with projections pi_s, pi_cu for E^s(p) + E^cu(p). Points on a common fake cu-leaf at distance rho have pi_s-separation at most kapparho; points on a common genuine local stable leaf have pi_cu-separation at most kapparho, where kappa = O(epsilon + oscillation of the splitting on B(p,4)) is small. Write q'-q = (x'-x) + (q'-x') - (q-x), with q-x and q'-x' along fake cu-leaves and x'-x, q'-q along genuine stable leaves. Since d(q,x) <= C_0 sigma_n with C_0 about 2 independent of k, one gets d(q',x') <= (1+O(kappa))(d(q,x) + kappa(d(q,q') + sigma_n)) and then d(q,q') >= sigma_n(1 - O(kappa)(1 + C_0)) >= sigma_n/2 for kappa small. The upper bound d(q,q') = O(sigma_n) follows similarly. The fiber of F_n over q is the homeomorphic image of W^s(x, sigma_n) under fake cu-holonomy with x mapped to q, so its frontier is the image of the frontier and the fiber contains W^s(q, sigma_n/2), which contains W^s(q, sigma_{n+k}). Hence G_{n+k} is inside F_n and F_{n+k} is inside G_n for suitable k.
Lemma 8.2 and Proposition 5.5 stand. The repair uses only smallness of epsilon, which the paper already assumes ('epsilon much less than the angle between any two of the subspaces').
Located concern
BW journal p. 474 ('In what follows, we will use only the first of these inequalities') versus p. 481 ('Proposition 3.1, part 7 implies that hat-W^u C^1 subfoliates hat-W^cu')
Inaccurate dependency statement. Regularity of hat-J^cu_n(x) (Prop. 5.4, conclusion 4) rests on absolute continuity of hat-W^u inside hat-W^cu(x), obtained from Prop. 3.1(vii), cu-part. By the proof of Prop. 3.1 that part uses the second inequality hat-nu < gamma*hat-gamma (or its strong version). So W^s-saturation of density points, as proved, uses both inequalities.
Author text, proof of Prop. 3.1: 'The center bunching hypothesis hat-nu < gamma hat-gamma implies ... E^u is a C^1 bundle over ... the leaves of the center-unstable foliation.' Conclusion 4 of 5.4 feeds Lemma 8.3 and the proof of Theorem 5.1.
Delete the remark, or prove absolute continuity of hat-W^u inside hat-W^cu-leaves without the second inequality. I did not attempt the latter.
None for Theorem 5.1, whose hypotheses include both inequalities. It matters only for a one-sided bunching variant.
Located concern
BW journal p. 480, proof of Lemma 7.1 ('follow from Lemma 4.4 with alpha = Jac(Tf | hat-E^u)')
Hypothesis mismatch. Lemma 4.4 (author text) is stated for a positive Holder function alpha on M. Jac(Tf | hat-E^u) is not a function on M: hat-E^u depends on the base point p_j, so it lives on the disjoint union N.
p. 473 defines hat-W^u on N as hat-W^u_{p_j} on B(p_j, r); these are different foliations for different j.
State Lemmas 4.1/4.4 for a family alpha^{(j)} on B(p_j, r) with a common Holder constant, which Prop. 3.1(vi) and f in C^{1+delta} provide. The proof is verbatim, since the sum over j of d(y_j, y'_j)^theta is at most the sum of (2C kappa^j)^theta.
None after restating.
Located concern
BW journal p. 476 (Props. 5.3, 5.4 quantified over x, x' in W^s(p,1) and q in S_n); p. 482 (D^cs_n(x), E_n(x)); pp. 484-485 (uses)
Domain mismatch. Section 8 uses hat-J^cu_n(x') and J^u_n(q) for x' in W^s(x, sigma_n) and q in D^cs_n(x). For x near the edge of W^s(p,1) these lie over W^s(p, 1+sigma_n), outside the set S_n for which Props. 5.3 and 5.4 are stated. The p. 484 fiber-uniformity claim and Lemma 8.2 also need k and c uniform in x', which the statement of 5.3 does not assert, though its proof gives it.
S_n is the union of hat-B^c_n(x) over x in W^s(p,1) (p. 474). D^cs_n(x) is the union over x' in hat-W^s(x, sigma_n) (p. 482).
Define S_n, T_n and state Sections 4-7 with W^s(p,2) in place of W^s(p,1), and state uniformity of k in Prop. 5.3. All estimates are unchanged.
None.
Located concern
BW journal p. 477 (last line of p. 477: 'h^s(J^cu_{n-k}(x))'); p. 478 ('q' in hat-B^c_{n-k1}(p')'); p. 481 (ratio display for bases); p. 484 (unions over q of J^u_n(x)); p. 485 ((21): 'Proposition 3.1 and (17) imply')
Misprints and one miscitation: missing hat on J^cu; p' for x'; hat-m_c omitted on the right of the base-ratio display; J^u_n(x) for J^u_n(q) in the displays of D_n, E_n; (21) concerns a genuine W^u leaf and needs Lemma 1.1, not Prop. 3.1.
Direct comparison with the definitions on pp. 474-475 and 482.
Correct as indicated.
None.
D3
Checked statement: Deduction of Theorem 5.1 from Props. 5.3-5.5 (pp. 476-477); Corollary 5.2 (pp. 471-472); ergodicity and Kolmogorov property in Theorem 0.1 (p. 472), and by the same deduction Theorem 0.3 and Corollaries 0.2, 0.4; reduction to stable saturation and passage to f^{-1} (pp. 472-473).
Returned assessment: repairable_gap
The logical deduction is sound; I found no obstruction to the principal conclusions. What remains open are citation-scope and chronology questions about imported inputs. THEOREM 5.1. Reduction: any two points of a W^s leaf are joined by a finite chain with consecutive points in some W^s(p_i, 1), so the local statement suffices. A^s is W^s-saturated and, since m(A^s delta A^u) = 0, essentially W^u-saturated, so Prop. 5.5 applies to X = A^s at both x and x'. Bounded Jacobians of h^s between the C^1 transversals hat-W^cu(x), hat-W^cu(x') preserve 'density tends to 1'. h^s(A^s cap hat-W^cu(x)) = A^s cap hat-W^cu(x') by genuine W^s-saturation. Lemma 2.1 in (hat-W^cu(x'), hat-m_cu), with Prop. 5.3 and Prop. 5.4(4), gives the julienne density at x', and Prop. 5.5 returns Lebesgue density. For W^u apply the result to f^{-1}: the rate functions become nu' = hat-nu(f^{-1}.), gamma' = hat-gamma(f^{-1}.), hat-gamma' = gamma(f^{-1}.), so both bunching inequalities, in either the (5) or the strong form, are symmetric under the swap. As written, each direction uses both inequalities (see D2 issue 3). Volume preservation is not used. COROLLARY 5.2. Correct: the density-point set is Borel, bisaturated and differs from A by a null set. ERGODICITY. hat-phi_s is constant on W^s leaves and hat-phi_u on W^u leaves by uniform continuity of phi. Birkhoff for f and f^{-1} with mu equivalent to m gives m(A^s(a) delta A^u(a)) = 0. Cor. 5.2 gives a bisaturated Borel set equal to A(a) mod 0; it is a union of accessibility classes, so essential accessibility gives measure 0 or 1. Continuous functions are dense in L^1(mu). The sum from i=0 to n divided by n is a harmless normalisation slip. KOLMOGOROV. K is equivalent to a trivial Pinsker algebra (Rokhlin-Sinai). The paper then needs every Pinsker set to be bi-essentially saturated, quoted from [BP74, Prop. 5.1]; given that, Cor. 5.2 and essential accessibility finish. REGULARITY BRANCHES. C^2 with (5): Prop. 3.1(vii) via [PSW97, PSW00]. C^{1+delta} with strong bunching: Prop. 3.1(vii) via [BW06b]. Everything else (absolute continuity of W^s, W^u; Holder continuity for Lemma 7.1; Hopf argument) needs only C^{1+delta}. GENUINE VERSUS FAKE. Genuine W^s is used in h^s, F_n, G_n, Prop. 2.7 and saturation of A^s. Genuine W^u is used in C_n, D_n and Prop. 2.5. Fake foliations are used for juliennes, E_n and Lemmas 6.1, 6.2, 8.1, with exchange points D_n/E_n and E_n/F_n. LOCAL FAILURES VERSUS OBSTRUCTIONS. The defects in D1 and D2 are local and repaired; none obstructs Theorem 5.1, 0.1 or 0.3. Corollaries 0.2/0.4: with one-dimensional center, gammahat-gamma can be taken uniformly close to 1; for 0.4 first fix theta in (0, delta) with nu/gamma < mu^theta and hat-nu/hat-gamma < hat-mu^theta, then tighten gammahat-gamma above nu^theta and hat-nu^theta. DISCLOSURE. I cannot remove later knowledge from model weights; no post-cutoff result was used to close any obligation here.
Premises used
- Props. 5.3, 5.4, 5.5 as audited in D2; Lemma 2.1 as audited in D1.
- Transverse absolute continuity with bounded Jacobians of genuine W^s between C^1 transversals (import; see D1 for dating).
- Lebesgue density theorem. available_from early 20th century; date_basis classical textbook result; version any standard text.
- Birkhoff ergodic theorem. available_from 1931; date_basis publication year; version classical.
- Rokhlin-Sinai: K-property equivalent to trivial Pinsker sigma-algebra for an invertible measure-preserving map of a Lebesgue space. available_from 1961; date_basis publication year, recalled and not supplied.
- [BP74, Prop. 5.1]: Pinsker sets are bi-essentially saturated. available_from 1974 (Izv. Akad. Nauk SSSR 38; English translation Math. USSR Izv. 8); date_evidence reference list p. 488; hypotheses there, per the paper's own footnote: constant rate functions; statement not supplied.
- [BPSW01, Section 2] for the Hopf-argument outline. available_from 2001; date_evidence reference list p. 488.
- [PSW97], [PSW00] for the C^2 branch of Prop. 3.1(vii). available_from 1997 and 2000; date_evidence reference list p. 489; exact theorem not specified in the paper.
- Essential accessibility; mu invariant and equivalent to volume.
Missing inputs
- Exact statement and hypotheses of [BP74, Prop. 5.1] (regularity, constant versus pointwise rates, class of invariant measure). Not supplied; its application to pointwise partial hyperbolicity and, for Theorem 0.3, to C^{1+delta} maps is conditional.
- [BW06b] (C^{1+delta} holonomy note). Not supplied in this packet. The journal lists it as a 2006 preprint while the task description calls it a 2005 note, so availability relative to the 11 Oct 2005 cutoff is unresolved. It is the authors' own companion note, and the whole C^{1+delta} branch (Theorem 0.3, Cor. 0.4, the parenthetical case of Theorem 5.1) is conditional on it.
- Pre-cutoff source for absolute continuity of W^s, W^u with non-constant rate functions; [AV08] is post-cutoff (see D1).
- Journal pp. 465-470 (Prop. 3.1 and Section 4) as page images.
Scope changes
- Theorems 0.1, 0.3 and Corollaries 0.2, 0.4 are stated on journal pages not supplied as images; their statements were taken from the author text and checked for consistency with pp. 471-474 of the journal.
- The external Hopf, Pinsker and absolute-continuity inputs were not audited; they are recorded as conditional.
Located concern
BW journal p. 472, last paragraph of the proof of Theorem 0.1 ('According to [BP74, Proposition 5.1], if f is partially hyperbolic, then any set in P is bi-essentially saturated')
Citation-scope gap for the Kolmogorov conclusion. The cited proposition is invoked for the paper's pointwise definition of partial hyperbolicity, for a measure merely equivalent to volume, and in Theorem 0.3 for C^{1+delta} maps. By the paper's own footnote, [BP74] works with constant rate functions.
Author text footnote 1: 'in [BP] the functions nu, hat-nu, gamma and hat-gamma ... are assumed to be constant.' No argument is given for extending Prop. 5.1.
Either check that the [BP74] proof needs only uniformly contracting or expanding absolutely continuous foliations with invariant-measurable-hull structure, or cite a pre-cutoff general theorem that the Pinsker algebra is contained mod 0 in the algebras of sets saturated by stable and by unstable manifolds for C^{1+alpha} diffeomorphisms. I believe such results existed by the mid-1980s (Ledrappier-Strelcyn / Ledrappier-Young circle), but I have not verified exact statements or hypotheses, so this stays conditional.
Affects only the Kolmogorov clause of Theorems 0.1, 0.3 and Corollaries 0.2, 0.4 outside the constant-rate C^2 setting. Ergodicity is unaffected. This is a missing justification, not a counterexample.
Located concern
BW journal p. 474 ('we will use only the first of these inequalities. The second inequality is used to prove W^u saturation') with p. 477 ('Proposition 5.4 implies that hat-J^cu_n(x') is regular') and p. 481
The stated split of the bunching hypothesis between the two saturation directions does not match the proof. Stable saturation uses Prop. 5.4(4), hence the cu-part of Prop. 3.1(vii), hence the second inequality.
See D2 issue 3.
State that both inequalities are used in each direction, or give a proof of regularity of hat-J^cu_n avoiding the cu-part of 3.1(vii).
None for Theorems 5.1, 0.1, 0.3 as stated, since both inequalities are assumed. It blocks only a one-sided reading.
Located concern
BW journal p. 477 ('Transverse absolute continuity of h^s with bounded Jacobians implies ...'); p. 461; references [AV08], [BW06b] on p. 488
Chronology and availability of imported inputs. (i) The step applies (11) to genuine W^s between the C^1 fake leaves hat-W^cu(x), hat-W^cu(x'); for non-constant rates the only precise source is post-cutoff. (ii) The C^{1+delta} branch rests entirely on [BW06b], which is not supplied and whose date relative to the cutoff is unresolved.
p. 488: [AV08] 'Preprint, 2008'; [BW06b] 'Preprint, 2006'. Recorded cutoff: 11 October 2005.
Provide dated versions or self-contained proofs. For (i), in the constant-rate case the 1967-1974 references suffice, subject to checking that they allow C^1 transversals.
C^2, constant-rate case: conditional only on classical imports. Pointwise-rate case and the entire C^{1+delta} branch: conditional or unresolved on the stated inputs. No step of the deduction itself fails.
Outside the audit
- Proposition 3.1 (construction and properties (i)-(vii) of the fake foliations) and its proof: journal pp. 465-467 not supplied as images; read only in the author text.
- Lemmas 4.1-4.4: journal pp. 468-470 not supplied as images; only the containment of f^j(T_n) in B(p_j, C kappa^j) and the distortion mechanism were re-derived.
- The supporting holonomy note [BW06b] (C^{1+delta} branch): not supplied.
- External theorems: absolute continuity of W^s, W^u (Anosov, Pugh-Shub, Brin-Pesin, Pesin theory, Abdenur-Viana); [PSW97, PSW00]; [HPS77]; [BP74, Prop. 5.1]; Rokhlin-Sinai; [PS00, Th. 4.4].
- The remark on p. 459 that every Lebesgue density sequence is regular, and the converse direction of the equivalence of (11) and (12) asserted on p. 460.
- Borel measurability of the sets C_n, D_n, E_n, F_n, G_n and of the bases and fibers, beyond plausibility.
- Section 9 (open questions, examples) and the introduction's historical claims.
Additional supporting archive items are identified in the original report. View the manuscript hashes and repository snapshot.