gpt-6.2-astra max: bw-density-downstream
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.
The two flat-box witnesses defeat the literal BW and ABW regularity criteria; they do not defeat the positive geometric applications. The lower radius in BW Lemma 8.2 has a fixed-chart proof, and the ABW analogue needs only a fixed time-zero adapted chart for this step. HT requires a sheetwise proof with projection of the r legs, not tangent geometry on the full orbit space. These are local completions with stated upstream conditions, not validations of the entire three density proofs. The note-specific exposure is real for lower-regularity holonomy assertions and extensions; it is not shared indiscriminately by the C² main theorems. No false principal theorem is established. Later replacement results have not been used to certify a proof at the 2005-10-11 cutoff.
L1
Checked statement: Whether the successive base and fiber inequalities imply regularity as defined, in BW Proposition 2.4 and ABW Proposition 5.2(1), and what is needed in the two regularity applications in each paper.
Returned assessment: claim_false
Regularity includes nesting and positive measure, not just a cross-multiplied volume inequality. BW introduces Cavalieri's principle in the context of sets nesting at a point, which explains the intended use. But c-uniformity is defined for arbitrary measurable sets in the box. Neither it nor the two hypotheses of Proposition 2.4 impose those other requirements. ABW even specifies measurable subsets, but makes the same omission. The existence of z and z′ in hypothesis (ii), respectively (b), guarantees positive leaf measure on at least one fiber of each rank, not positive measure on the transversal.
In the BW application inside the fake cu leaf, the bases are the nested center plaque disks of radii σ_n, of positive center measure comparable to σ_n raised to the center dimension. For y in the rank-(n+1) base, consider the terminal rank-(n+1) fake u disk. One inverse step takes it into the disk at y_n of radius at most ν̂(p_n)τ_{n+1}<τ_n: use the contraction on the moving chart, τ_{n+1}=τ(p_n)τ_n, and connectedness of the plaque component containing y_n. The remaining n inverse steps are the same in both definitions. Thus the smaller-rank-radius construction really gives a subset of the preceding julienne. Here τ<σγ<1 by p. 474 and the center rate bounds. Only y in the smaller base is needed for this check; no use of a rank-(n+1) fiber outside its defining domain is necessary. The fibers are positive leaf-volume sets (diffeomorphic pullbacks of open disks). The local plaque products are measurable: one can use local foliation coordinates, with open center bases and connected local disks cut out by strict radius inequalities, and the continuous plaque maps and iterates. In particular these sets contain relative neighborhoods of their centers. Positive cu measure is also explicit on p. 475. Section 7 gives uniform fibers and a consecutive fiber-volume lower bound; on p. 481 bounded-Jacobian disintegration is for fake u inside fake cu. This proves the corrected criterion applies, given that disintegration input.
For BW's G_n, the bases are just those cu-juliennes. Above a base point persisting to rank n+1 the two fibers are genuine stable plaque disks with the same center and decreasing radii. Hence G_{n+1}⊆G_n. Stable fiber volumes are comparable to σ_n raised to the stable dimension in the local graphs. This gives c-uniformity, positive volume after disintegration, and the required fiber-volume ratio since inf σ>0. This time the bounded-Jacobian input is for genuine stable leaves transverse to the fake cu base. There is no use here of a regular sequence with zero transversal measure.
ABW p. 963 uses fake u over positive center disks in the fixed time-zero fake cu plaque; pp. 964, 966 use G_n made of genuine stable star-disks over that cu base. The fixed-chart disks are measurable, have positive induced volume and nest as their radii decrease (the bounds on σ_{n+1}/σ_n are stated on p. 964). Positive fibers and the disintegration invoked on pp. 963 and 966 give positivity in the two ambient measures. For nesting of the cu base of G_n, the geometric check in star notation is precisely f^{-1}(Ŵ^u_⋆(y_{n+1},τ_{n+1}))⊆Ŵ^u_⋆(y_n,τ_n) for y in the smaller center disk. With the local inverse contraction on the moving star charts and terminal radii chosen so that the contracted radius is ≤τ_n, the pullback and nesting proof above applies verbatim. Then G_n nests without any further dynamical estimate. This identifies a specific upstream condition, not a consequence of 5.2(1). The end of the ABW proof on p. 963 displays an inner disk in that inverse image to prove a lower volume estimate; that inclusion alone does not prove the nesting inclusion in the other direction. The earlier star-scale choices are not all in the supplied ABW pages. I find the intended positive disk/tube applications, not a flat-box counterexample realized by them; my ABW nesting check is conditional on that one-step input.
Premises used
- BW (2010), printed p. 460, (12): in the working box, for the chosen uniformly transverse disk, the integral of leaf volumes against transversal volume is within a fixed factor C of ambient volume. Use the corresponding property with the induced volume when the ambient space is a weak plaque.
- BW (2010), printed pp. 464–468, 473–476: locally invariant coherent graphs on moving orbit copies, strong plaque contraction, positive continuous scales, and uniform graph sizes on the working boxes. These give the disk inclusions independently of Proposition 2.4.
- For the volume-ratio applications, BW Proposition 5.4's unstable-fiber comparisons (printed pp. 480–481), and ABW's corresponding fiber estimates and fake subfoliation input invoked on printed p. 963. Neither nesting nor bounded distortion is deduced from the other.
- All density quotients are for positive denominators, in a box on which one disintegration constant works for all ranks. On a local interior plaque it suffices for density limits to work on a tail.
Missing inputs
- For a source-complete check in ABW, the definitions and estimates preceding printed p. 963, in particular Proposition 5.4's moving star-chart rates and the choice of τ_n. A sufficient explicit estimate is an inverse u-plaque factor a_n with a_n τ_{n+1}≤τ_n. Comparability of volumes, or exponential decay of a product from time zero, is not by itself this estimate.
- The original external proofs of the genuine bounded-Jacobian property for the transversals used and the branch-specific fake regularity theorem are not checked here. The abstract corrected measure criterion does not depend on their historical status.
Scope changes
- Regular need not mean that the intersection is a single point; no such extra condition is necessary for the corrected criterion.
- ABW p. 964's equal-base density comparison for C_n,D_n and p. 966's final saturated-set comparison are other uses of Proposition 5.2, not further uses of its false standalone regularity implication.
Located concern
BW (2010), printed pp. 459, 461–462, Proposition 2.4; ABW (2009), printed pp. 945–946, Proposition 5.2(1)
Two independent missing conditions in the standalone implication to regularity.
In a small Euclidean product box, use vertical leaves and a horizontal transversal (C=1). Alternate I_0×(-a,a) and I_1×(-a,a) for disjoint equal-length open subintervals I_0,I_1 of the transversal. Both hypotheses hold with c=δ=1 and positive volumes, but there is no nesting. Separately, take Y_n={0}×(-a2^{-n},a2^{-n}). These even nest at a point. Each positive-fiber base is {0}, c=1, the base inequality reads 0≥δ·0, and (ii) holds at 0 with δ=1/2. Every ambient volume is zero. For the ABW formulation one can take Z_n=Y_n.
Assume measurable nested Y_n in a box satisfying the bounded-Jacobian comparison, and m(Y_0)>0 (equivalently here, positive initial base measure); keep the stated c and δ hypotheses. For any positive fiber of length or leaf-volume l_n, Proposition 2.3 gives (Cc)^{-1} l_n m_τ(τ_{Y_n}) ≤ m(Y_n) ≤ Cc l_n m_τ(τ_{Y_n}). Use the two selected fibers and the two δ inequalities to get m(Y_{n+1}) ≥ δ²(Cc)^{-2} m(Y_n). This propagates positivity and proves regularity. Alternatively retain the old hypotheses and conclude only this last inequality.
False auxiliary formulations, with a valid volume calculation. The repair does not require a new bunching condition. The positive nested families described above are the relevant consumers; a dynamical conclusion does not fail just because the abstract assertion has more general counterexamples.
L2
Checked statement: The E,F,G internesting argument: the rank of the transported point, the lower radius of the strong fiber in the foliation product, and the scope of a time-zero chart proof in BW, ABW, and HT.
Returned assessment: repairable_gap
BW: write J_n(x) for the fake cu-julienne, D_s(x,r) for the genuine stable disk (component convention), and r=σ_n. The common stable base in the E,F comparison is genuine as well as fake on the plaque through p. The fibers above x′ in this base are J_n(x′) and h^s_{x,x′}J_n(x), respectively. The genuine stable holonomy estimate and its inverse have a fixed rank loss. Applying them and including the smaller stable base in the larger proves the E,F inclusions. In particular, if q∈J_n(x), the point q′=W^s_loc(q)∩Ŵ^{cu}(x′) belongs to J_{n-k}(x′), not by this argument to J_n(x′). For boundary points use the product map on a slightly larger open box, or take closures of the estimates. The diameter of J_n is O(r): add the center radius to the pulled-back unstable radius O(ν̂_n τ_n), with ν̂τ<σ. The shifted diameter is still O(r) since inf σ>0. Thus q,x,x′,q′ are all in an O(r) chart neighborhood of x. One can also get this localization for F without any rank assertion about q′: intersect the small transverse stable graph through q and cu graph through x′. With complementary coordinate planes and graph slopes <b<1, substitution of one graph in the other bounds both distances to the intersection by a constant times d(q,x′).
Here is the additional lower-bound proof. Use a single time-zero coordinate chart at x, S=T_xW^s and V=T_xŴ^{cu}, and let P project to S along V. Identify point differences with coordinate differences only in this chart. The angle bound controls the norm of P. On a boundary point x′ of D_s(x,r), d(x,x′)=r. The stable graph through x and chart metric comparison give |P(x′−x)|≥a r for some a>0. On the other hand q,x are on one fake cu graph and q′,x′ on another graph of the same time-zero foliation. The tangents in a ball of radius Cr approach V with modulus ω(Cr)→0. Join each pair in its graph by lifting a straight segment in the graph plane in a slightly enlarged chart; the paths have lengths O(r). Consequently |P(q−x)|+|P(q′−x′)|≤C′r ω(C′r)=o(r). It follows from P(q′−q)=P(x′−x)+P(q′−x′)−P(q−x) that d(q,q′)≥c r on the boundary for all sufficiently large n. The upper bound d(q,q′)≤C r holds over the whole closed disk. This uses continuity of the tangent field on the working box, not differentiability of cu holonomy across leaves. The uniform tangent modulus in BW Proposition 3.1(vi), compact chart and angle bounds, and true stable graph control make the estimates uniform on interior working plaques.
For fixed q the F fiber is the image of D_s(x,r) by fake-cu product holonomy from the stable plaque through x onto a neighborhood in the stable plaque through q. This homeomorphism sends x to q and, on a box containing the closed disk in its interior, sends its relative boundary to the relative boundary of the image. Thus the image contains D_s(q,c r): a path in that component from q to the outside would cross the boundary. The image is connected and contained in the component D_s(q,C′ r) for C′ larger than the outer bound. This is an inradius and an outradius argument in any stable dimension, not an inference about the volume or diameter of a possibly very thin fiber. If the strong dimension in such a comparison is zero, the fiber inclusions are trivial. For b_0=sup σ<1 take k so that b_0^k<min(c,1/C′). Then σ_{n+k}≤cσ_n and σ_{n-k}≥C′σ_n. Nesting the transverse J bases as well gives G_{n+k}⊆F_n and F_{n+k}⊆G_n. For a fixed interior x, tail comparisons suffice; finitely many well-defined earlier neighborhoods can be covered by increasing the shift since the families shrink to x and contain neighborhoods of x. For uniform working domains, take the stable centers in the interior of a slightly larger plaque on which the construction is made. Do not apply an estimate over W^s(p,1) to an uncontrolled plaque outside it.
ABW: use the star chart at time zero of the fixed reference orbit. Its locally flat norm is a single norm for this argument. Use the projected differences with S the true stable tangent and V the fake cu tangent there, with r=σ_n. The definitions of F,G on p. 964 are the same product and strong-disk definitions in that metric. The unstable pullback bound O(ν̂_n τ_n)=o(σ_n) used on p. 965 gives the required julienne size; Proposition 5.11 as invoked on p. 966 supplies the shifted transport. The scale ratios away from zero and one on p. 964 handle the shift in size and the inner and outer constants. On continuous-tangent graphs in this one chart the same o(r) calculation and boundary argument apply. This proves the local repair given ABW's plaque and rank estimates; constants can depend on the reference adapted chart. Nothing here compares all future star metrics by a single smooth ambient metric or proves uniformity over all nonuniformly bunched orbits.
HT: reverse the strong direction (now u), use r_n=δ_n, and do not put tangent coordinates on M_f. E,F are compared over the real unstable disk by the unstable transport of the rsc-julienne, with a shift from Proposition 7.1 (pp. 32–33). For F,G, fix q∈Ĵ^{rsc}_n(x), x′ in the closed unstable disk through x, and let q′ be the rsc-product image of x′ on the unstable disk through q. Equivalently, q′ is unstable transport of q to the rsc plaque at x′. The p. 33 use of 7.1 makes sense with x′ in place of the undefined p′: shifted size control gives d(x′,q′)=O(r_n) as well as d(x,q)=O(r_n). No same-index assertion is needed or made there.
In a product neighborhood U×K (K a history fiber) put X=πx, Q=πq, X′=πx′, Q′=πq′. Two facts about the local model matter. First, an r leg has zero displacement in M. For an r-extension of a fake sc plaque in U, its projection lies in the sc plaque: if the plaque is D in a reference sheet, r-saturating it gives points over D, not over a translated disk. Equivalently one may draw the sc graphs in the common forward chart and lift them to the history sheets. Histories over the same chart center have identical forward fiber maps; the forward-only sc graph transform with a prescribed point gives the same projected sc graph, by uniqueness, when constructed with those same maps. This explains why one can use a single sc foliation here. Part (viii) on p. 15 states the equality for centered plaques; it does not assert equality of center or unstable plaques for all histories. Second, x,x′ are in one local scu sheet and q,q′ are in one local scu sheet, possibly a different one. The metric convention used on p. 32 gives d=d_M after projection for such same-sheet pairs. A mere homeomorphism in p. 16 without this metric control would not suffice.
In the base chart project onto T_X πW^u(x) along the tangent of the projected fake sc plaque through X. For the above r-extension, Q−X and Q′−X′ are chords of projected sc graphs in an O(r_n) box. In particular their unstable coordinate projections are o(r_n), including when a history displacement accounts for most of the full distance d(x,q). In contrast |P(X′−X)|≥a r_n on the unstable boundary, since d_M(X,X′)=r_n there. It follows that c r_n≤d_M(Q,Q′)=d(q,q′)≤C r_n on the image boundary. All small-error estimates can be made in the one projected sc chart, uniformly in q, not by claiming transverse smoothness in a Cantor parameter. The projected unstable disks are used sheet by sheet with uniform local product domains; there need not be an unstable foliation on M. The product homeomorphism and component argument now give the two inclusions on p. 33 after fixed shifts. One needs the shrink and fixed-shift bounds on δ_n and the corresponding inclusions for the rsc bases (including their history cylinders), not just a bound on a diameter. Fixed shifted base inclusion suffices if one takes multiples of its shift. This validates a sheetwise repair for the stated product construction, conditional on the precise r-extensions and uniform scale/holonomy inputs; the limited HT excerpts do not certify all those inputs. No estimate of a lower length in a history fiber is called for.
Premises used
- BW Proposition 5.3 is used with a shift uniform in the two centers on the working plaque, not a shift depending without bound on x′ as n varies. Its proof on printed pp. 477–479 seeks exactly such a k. Lemma 6.1 uses uniform Lipschitz fake stable holonomy on center disks inside fake cs, not Lipschitz holonomy on whole cu transversals. This costs a fixed shift because sup σ<1.
- Check on the other leg in BW §6: track the center base point y and a point of its rank-n fake unstable fiber in the thin orbit boxes. The transported center base point on fake stable, and the transported fiber point on genuine stable, remain within O(ν_j) of the respective tracked points by contraction, with a box margin. At time n the two transported points are within O(τ_n), since the original terminal u leg is at most τ_n and ν<τ. Intersect fake u of the transported fiber point with fake cs of the transported base there. Local graph bounds give O(τ_n) displacements; coherence in their common fake cu gives a point on the destination center. Pull back on fake u to stay in the boxes. The center displacement at time zero is O(τ_n/γ_n)=O(σ_n), by local invariance on the moving copies. At time n−k the remaining u displacement divided by τ_{n-k} is at most C[sup(ν̂τ)]^k. This gives the uniform rank loss after including the shifted center offset, also with centers exchanged. This terminal-intersection justification uses the uniform plaque and thin-box inputs, not Lemma 8.2 or a same-rank hypothesis.
- BW Proposition 3.1(i),(iv)–(vi) and the product-box convention on printed p. 468 give the transverse graphs and continuous tangent fields. The metric projection in the new F,G proof is along the actual fake weak tangent, not necessarily along a summand of the genuine invariant splitting.
- In ABW, take the locally controlled continuous-tangent plaque system of Proposition 5.4 and the uniform-within-the-working-plaque shifted holonomy estimate of Proposition 5.11 as inputs, as used on printed pp. 963–966. A fixed time-zero Riemannian/star norm comparison does not change density-one tests there.
- HT pp. 10, 14–16 distinguish r fibers, scu sheets and diffeomorphic fiber maps over the invertible orbit shift. Pp. 28, 32–33 use the product, its holonomies and the shifted transport bound. For the metric repair one needs only projection of the rsc incidences to sc plaques, not identical projections of all juliennes over all histories or constancy of unstable directions over a fiber.
Missing inputs
- ABW: the full statements/construction of Propositions 5.4 and 5.11 and the preceding nonuniform orbit estimates are not supplied. The local calculation verifies what follows from them, not their proofs. Base nesting is subject to the check in L1.
- HT: the full product/metric definitions and statements behind pp. 14, 16, 28 and 32, the definitions of Ŵ^{rsc}, Ĵ^{rsc}n and the δ scales, and the full hypotheses of Proposition 7.1 are not supplied. To certify the original application, check the r-extension in the common chart (also for off-center x′), the shifted size bound uniform in the histories used, and the base inclusions. Sufficient scale control is 0<a≤δ{j+1}/δ_j≤b<1 for the radii along the orbit. Arbitrarily recentering unrelated fake sc plaques in different sheets would not be justified by (viii) alone.
- The results giving the plaque systems, true strong graphs and their uniform sizes are used as geometric inputs. The fixed-chart argument supplies the missing lower bound; it is not a proof of all those existence and regularity theorems.
Scope changes
- The chart proofs are additional arguments, not readings of the word 'Hence' as a proof. They use no new dynamical bunching once the appropriate local plaque and scale inputs are in place.
- Within-chart or within-orbit uniformity suffices in a pointwise density argument. A global constant across all adapted ABW metrics or a differentiable structure across HT histories has not been established and is not used.
Located concern
BW (2010), printed p. 485, Lemma 8.2; ABW (2009), printed pp. 965–966, Lemma 5.21
The destination membership is written with the same n; the cited transport result is an internesting result. Even an upper size bound at both ends does not prove the asserted positive lower radius.
The product definition locates q′ on the destination fake cu plaque, and shifted control bounds its offset from x′. Subtracting two O(σ_n) offsets from d(x,x′)=σ_n need not leave any positive multiple of σ_n. There is no demonstrated counterexample to the internesting conclusion.
Use shifted transport or local graph intersection for localization. Use the projection estimate, image-of-boundary argument, and nested-base inclusions in the argument field for the two-sided comparison.
A local proof completion for both smoothness branches of BW given their respective plaque inputs, and for ABW in its fixed time-zero chart given its own inputs. It does not strengthen quasi-preservation to exact preservation.
Located concern
HT 2025 preprint, printed p. 28, Lemma 9.4 and pp. 32–33, Appendix B
Undefined p′ in an upper bound, followed by an unsupported lower bound on an unstable fiber.
Proposition 7.1 is actually invoked here. Correcting p′ to x′ still gives only an O(δ_n) upper estimate. The two rsc legs can involve different histories, so four points cannot simply be subtracted in a smooth chart of the orbit space.
Use the projected sc chords and the two same-sheet unstable pairs as above. Verify the history-product and rank hypotheses for the specific families before using the repaired inclusions.
The HT proof shares the lower-bound problem, not BW's exact same-index claim. The geometric repair does not establish the separate orbit-space measure and saturation arguments.
L3
Checked statement: Which historical proof inputs support the four selected downstream assertions, what is conditional in the lower-regularity extensions, and which additional BW geometry/density qualifications affect these uses.
Returned assessment: unresolved
The proof edges are identifiable; unconditional support for all the extensions is not. BW journal p. 467 makes the branch split explicit. For k≥2 it uses a bundle/graph-transform argument with [PSW97, PSW00] for the difficulty of the merely C¹ weak leaf. For 1<k<2 it instead invokes the holonomy result [BW06b] for the fiber extension F_r. The statement obtained on p. 465(vii) is uniform strong C¹ subfoliation inside fake weak plaques. Its two important uses here are uniform Lipschitz control between center transversals (p. 478) and bounded-Jacobian disintegration of fake u within fake cu (p. 481), also with time reversed. It is genuine stable holonomy between the full fake cu transversals in Proposition 5.3 and the density transfer on p. 477. Thus neither the fake regularity claim alone nor a theorem just about an ergodic conclusion suffices for all the density comparisons.
HHU: Proposition B.7 on p. 374 attributes to [14, Theorem 0.2] a uniform C¹ conclusion for the stable foliation restricted to a local center-stable manifold everywhere tangent to E^s⊕E^c, for C^{1+α} with the indicated bunching. This is not a theorem about every disk in a cone. Lemma B.8.1 explicitly takes the Lipschitz bound between center curves in the same such manifold, uses scale comparability and absorbs the constant in a shift of B^c_n. This is one part of Proposition B.8, later used on p. 380 with a two-sided volume bound for stable holonomy. HHU p. 374 expressly offers [26, Theorem 2.1] for the C², one-dimensional-center case. Reference [26] is Pugh–Shub (1972), p. 381; [14] is the holonomy preprint, distinct from the submitted ergodicity paper [13], p. 380. Theorem B, p. 356, concerns PH²_m with one-dimensional center, essential accessibility implying Kolmogorov, hence ergodic. HHU describes its own use of integral center curves and mentions the lower-differentiability improvement of BW separately. The low-regularity assertion of B.7 has a direct note dependency. It is not established that HHU's C² Theorem B has no alternate route; nor does the mere citation prove that [26] covers the local weak manifolds and uniform bounds for all the center-curve choices in B.8.1.
ABW: the analytical Theorem B, p. 936, assumes a C² partially hyperbolic diffeomorphism and a bi-essentially saturated X. It asserts stable saturation of the ambient density representative intersected with CB⁺ and unstable saturation after intersection with CB⁻. These are directional nonuniform bunching loci, not necessarily the entire manifold. The comparison in pp. 963–966 is in the proof of Proposition 5.13 for a fully stable-saturated, essentially unstable-saturated representative. There is direct reuse of the Section 2.3 measure criterion and of the julienne proof step (L1–L2). There is no inference from the C¹ genericity problem of the paper to a use of BW's C^{1+δ} clause in this analytical theorem.
ASV: p. 16 first gives ordinary center bunching (2.5) in the C² setting, then explicitly refers to [10, Theorem 0.3] and says all the results extend to C^{1+α} with strong center bunching. This is a historical lower-regularity dependency through BW, not a separate proof of lower-regularity holonomy on this page. Replacing the input would require the density/saturation and holonomy control actually used for the extensions, with their uniform and measure hypotheses, not simply ergodicity of an essentially accessible volume-preserving base. The page does not identify a failure of a separately stated C² result, or by itself verify every application in the paper.
HT: Proposition 3.1(vii), p. 15, asserts both BW clauses. The construction replaces the base of the disjoint tangent-fiber model by the discrete set of full orbits, with fiber T_{πp}M; the fiber maps are diffeomorphisms over the invertible shift although f need not be invertible on M. The proof appeals to the rest of BW's argument and its uniform estimates. Thus the C^{1+α} strongly bunched clause inherits a need for the lower-regularity input in that orbit-family model. Theorem 1.1, p. 2, assumes C², volume preservation, partial hyperbolicity, ordinary center bunching, constant Jacobian and essential accessibility, and concludes ergodicity. It needs the C² branch, not the extra low-regularity assertion. It does use density comparisons, including the local geometry at issue in Lemma 9.4. Pp. 10–11 explain why even a repaired diffeomorphism proof is not by itself the endomorphism proof: constant Jacobian supplies a local product measure, a good sheet can be used for stable saturation, but a prescribed unstable segment need not be on a good sheet. The r-extended juliennes enter the latter argument. The K assertion discussed on p. 11 concerns the invertible natural extension and a separate Appendix A argument; it is not the statement of Theorem 1.1.
For BW itself the remaining deduction is specific: Proposition 5.5 changes ambient to cu density for a genuine stable saturate still essentially genuine unstable saturated. The C_n,D_n comparison uses essential genuine u-saturation, the D_n,E_n change is geometric, and the last G_n comparison needs full genuine s-saturation (pp. 482, 484–486). One may change the representative modulo zero in ambient volume, not arbitrarily on a cu transversal. Bounded volume factors for genuine h^s, internesting and regularity then transfer complement density zero to the destination juliennes (p. 477). The inverse dynamics supply unstable saturation. Theorem 5.1 and Corollary 5.2 on pp. 471–472 need no invariant probability. With an invariant probability equivalent to volume, forward and backward Birkhoff sublevel sets for continuous functions give the two essential saturates; the density representative and essential accessibility make their measures zero or one. This gives ergodicity. The Pinsker-set saturation input cited as [BP74, Proposition 5.1] on p. 472 is a separate premise for K.
Premises used
- Historical cutoff 2005-10-11 (journal received date). The dated June 2005 note is pre-cutoff evidence, not something first available merely because a journal reference key contains '06'. The title/key/version in a later bibliography does not establish line-for-line identity with that note. The October 2006 author text, printed pp. 6–7, explicitly describes plugging the lower-regularity result into the proof; the corresponding journal introduction on p. 455 is reported in the packet, not a supplied page image here.
- BW C² regularity interface: use the corrected strong-subfoliation argument for the uniformly C² fiber maps and the inverse maps, with strict center bunching. For u inside cu, at exponent one the domination divided by the weak-leaf base conorm is bounded with slack by ν̂/(γγ̂)<1 in a splitting-adapted metric; for s inside cs the inverse test uses ν/(γγ̂)<1. One needs the ambient bundle/weak-leaf version, not differentiation of a merely C¹ weak graph twice. BW printed pp. 465–467 describe the uniform extension; the underlying PSW proofs are not audited in this packet.
- BW lower branch: C^{1+δ}, the strong-rate bounds μ, μ̂, and a common 0<θ<δ with ν^θ,ν̂^θ<γγ̂ and ν/γ<μ^θ, ν̂/γ̂<μ̂^θ (author October 2006, printed pp. 6–7). The strict rate tests transfer with slack to a sufficiently close fiber extension. That fact verifies numerical applicability, not the asserted uniform holonomy theorem for the extension. No Brown or Saghin theorem is used as a premise here.
- For the later density deductions assume local two-sided bounded-Jacobian genuine s,u holonomy/disintegration on uniformly transverse C¹ disks, not just preservation of null sets by a single holonomy. BW printed pp. 460–461 state the needed properties and distinguish constant-rate references from the pointwise-rate setting. The Birkhoff theorem is used for integrable functions of an invertible probability-preserving system, the Lebesgue density theorem for Riemannian volume, and Pinsker saturation in the version for the particular invariant probability. I checked the deductions on BW pp. 471–472, 476–477, not the proofs of those external inputs.
Missing inputs
- The focused sources give the June note witness and transcription, not a primary page-by-page note proof or a collation with [BW06b] or HHU [14]. No verification that a particular later note text either retains or fixes every step. A valid uniform fake-holonomy or subfoliation theorem in the needed model, in both directions, is required to remove the low-regularity proof condition.
- Original statements/proofs of HHU [26, Theorem 2.1], the corrected PSW theorem, the uniform graph-transform imports and the full pointwise-rate genuine absolute-continuity and Pinsker inputs. In particular, matching C² and center dimension one does not by itself verify the weak-leaf adaptation of [26]. Source access limits are not negative mathematical results.
- ASV's individual later proof uses beyond printed p. 16, ABW's preceding nonuniform estimates, and HT's full natural-extension density proof are not present. Their unexamined hypotheses cannot be supplied just by citing BW.
- The reported later Brown stronger-bunching repair and Saghin fake-holonomy result are modern replacement candidates only. For an application one would still have to match the rates, the precise fake plaque construction on uniformly bounded orbit charts, forward and inverse bounds, and the absolute-continuity interfaces. I have not checked those papers' proofs or substituted a later theorem for a pre-cutoff proof.
Scope changes
- The unresolved verdict concerns unconditional proof support, not the existence of the identified citations and uses. C² does not need the disputed lower-regularity step, but it does not by itself discharge L2, a nonuniform chart estimate, an endomorphism measure argument or a Pinsker theorem.
- This is a non-blind adjudication of the selected passages. The Sol inventory's Lemmas 8.4–8.5 do not exist in BW §8: printed p. 486 finishes Lemma 8.3 and Proposition 5.5 and starts §9. That is not a defect in BW.
Located concern
June 11, 2005 supporting note, printed p. 6, Lemma 1.1(2) as given in the packet; BW (2010), printed pp. 465, 467, Proposition 3.1(vii)
The disputed iterated estimate is an auxiliary assertion about projective-bundle distance including base distance. The assertion of low-regularity holonomy is a proof input; it is not the same assertion as that estimate.
On the given flat toral product the stable rate λ=(3−√5)/2 lies strictly between 3/10 and 2/5, the unstable rate is its reciprocal, the center derivative is 1, and the center line is constant. With γ=γ̂=.9, the extra tests include √.4<.81 and (.4/.9)²<.3, with θ=1/2<3/4. Put the initial separation ηλ^n along stable with identical fiber direction, fixing η below any proposed radius and the injectivity scale. All the backward separations through n are at most η. The transcribed conclusion gives η≤√η(√λ/.81)^n, impossible since √λ=(√5−1)/2<.81. This is an all-radii argument, not the finite sample scalar check.
Do not iterate a one-step Hölder bound with an unchanged exponent in that manner. Supply valid holonomy estimates in the version and uniform application required by the consumer, or restrict to a branch with an independently applicable theorem.
A false estimate on the reported June note statement. The product actually has smooth center holonomies and preserves the circle coordinate on su-paths, so is not essentially accessible. It refutes neither the holonomy conclusion nor a principal ergodicity theorem. The failed derivation cannot certify the lower-regularity input.
Located concern
BW author October 2006, printed pp. 3, 9 (adapted metric convention); BW journal (2010), printed pp. 464–468, especially the weak-plaque chord bounds on p. 468
Sol's weak-sum metric objection matters to the orbit estimates, but not as an obstruction to choosing a working metric or as a requirement of orthogonality in the L2 projection proof.
Bounds on the three summands alone do not bound the weak sums in every such metric. In the same diagonalizable product take unit e_c,e_s with inner product ρ near 1, and e_u orthogonal to them. The squared norm ratio on e_c−e_s is (1+λ²−2λρ)/(2−2ρ), unbounded as ρ tends to 1, with the separate rates unchanged. Scalar rescaling does not fix this.
Choose first the continuous metric orthogonalizing the splitting and retaining the norms on the summands. In it the weak-sum operator bound is the maximum of the relevant summand bounds. Smoothly approximate the metric in C⁰ using the uniform strict margins, including the bunching and lower strong rates. Small cone graphs, secants in a small exponential chart, and uniform closeness of the chart map of f to its derivative in Lipschitz error then give the one-step chord bounds in this metric (also for f^{-1}). Only then choose or rescale the working boxes.
A normalization available without changing the smooth volume class, density points or bunching hypothesis. This is not a license to replace all of ABW's moving norms by one uniformly equivalent smooth norm.
Located concern
BW (2010), printed pp. 464–465, 468, 473, 478–481, 484–485; BW author October 2006, printed pp. 23–25 for the thin-neighborhood estimate
Orbit-indexed fake geometry has to be kept in the distortion and intersection arguments. A fake Jacobian need not be a single function on overlapping balls of M.
At time j the bundle in the Jacobian is the one in the p_j copy, locally invariant on the tracked points. Neither that bundle nor an arbitrary fake intersection is globally invariant under unlimited iteration in M.
In Lemma 7.1 put α_j(z)=Jac(Df(z) on Ê^u in copy j). These factors have a common positive lower bound and uniform within-copy Hölder bounds by invertibility on compact M, C^{1+δ} and (vi). For two points in T_n the log product difference is bounded by H Σ_{j<n} d(y_j,y′_j)^β ≤ H(2C)^β/(1−κ^β) using the thin tubes. This proves the needed distortion by the chain rule without identifying overlapping copies; for real fibers use E^u instead. For Lemma 8.1, intersect the alternate unstable graph with the tracked fake cs graph at time n, within O(τ_n) by the local graph estimate; pull back by unstable contraction in the boxes. Local invariance (or the invariant global fiber graphs restricted to those boxes) identifies the initial intersection. Both initial unstable offsets are O(ν̂_n τ_n)=o(σ_n); fixed shifts absorb the terminal and base constants. Use the analogous terminal check for §6 as in L2. Read iterated plaques with moving p_j subscripts. Slips in bound variables or destination names do not provide independent counterexamples to these arguments.
The distortions and localized identifications have short completions given the plaque, rate and thin-tube inputs. They do not supply (vii) for C^{1+δ}, and the disputed note estimate is not needed for this log-distortion calculation.
Outside the audit
- Full proof and version history of the supporting note, full pre-cutoff external holonomy, graph-transform, genuine absolute-continuity and Pinsker sources, and full modern Brown and Saghin proofs.
- Nonuniform star-metric construction and all earlier ABW scale choices; full HT definitions and proofs for history cylinders, transport, product measures, Lemma 9.5 and Appendix A.
- HHU's alternate 1972 theorem in its original text, other steps of HHU Theorem B, and ASV applications beyond the supplied Section 2.1 passage.
- Any exhaustive citation graph, any unassigned BW examples or open questions, and any inference about authors' awareness.
Additional supporting archive items are identified in the original report. View the manuscript hashes and repository snapshot.