{"id":"db86bdb1-c594-4f39-9e58-cd374682d032","arxiv_id":"2506.00405","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Conservative partially hyperbolic diffeomorphisms homotopic to the identity on closed 3-manifolds with non-virtually-solvable fundamental group are always accessible and hence ergodic.","lead":"The paper proves that any volume-preserving partially hyperbolic diffeomorphism of a closed 3-manifold that is homotopic to the identity must be accessible, and therefore ergodic, unless the fundamental group is virtually solvable. This resolves the Hertz-Hertz-Ures ergodicity conjecture within the homotopy class of the identity.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The proof's central reduction to ideal-boundary geometry rests on an unproved extension of Candel's uniformization to C1 branching foliations and laminations; if that extension fails, Sections 3-7 do not apply to W^cs or Λsu.","rationale":"The reader's weakest_assumption identifies exactly the same load-bearing concern: the application of Candel's uniformization to the C1 branching foliation W^cs and to the su-lamination Λsu. I agree that this is the most serious gap. Without uniform Gromov hyperbolicity of the leaves, the paper's abstract results on ideal boundaries, quasi-geodesic subfoliations, degenerate limit sets, and non-separated leaves cannot be brought to bear on the partially hyperbolic system. The paper itself flags the issue in Remark 8.7, but gives no proof that [Cal01] covers branching foliations in the sense of Definition 2.13 or that the uniformity constants survive the approximations used in Section 8.4. The reliance on the unpublished preprint [BFP25] in Theorem 7.1 is a secondary concern: it affects one step (transferring non-separated rays from one leaf to another), whereas the hyperbolicity assumption underpins the whole geometric framework. Because the reader already conditioned the verdict on supplying this missing justification, my assessment does not change the recommended verdict: the paper remains credible, with a specific, checkable condition attached. If the Candel-extension assumption were disproved, the verdict would drop to REJECT; if it were supplied, the paper would move toward ACCEPT. Thus UNCHANGED is appropriate relative to the reader's CONDITIONAL verdict.","tokens_in":42397,"tokens_out":4701,"duration_ms":47140,"concrete_test":"Independently derive Proposition 8.9 for the branching foliation W^cs arising in the incoherent example of [BGHP20]: construct the approximating foliations W^cs_epsilon, leafwise smooth them, and check whether the bi-Lipschitz distortion to H^2 remains bounded independently of epsilon and of the branching points. If the constants blow up as epsilon tends to 0 or depend on the branch locus, the uniformity claim in Theorem 8.6 fails and the proof of Theorem 1.3 collapses. If the constants stay bounded, the Candel-extension assumption is validated at least in that representative class; a written proof covering branching foliations in the sense of Definition 2.13 would settle the concern generally.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The decisive step is in Sections 8.3-8.4: Theorem 8.6 and Proposition 8.9 assert that the su-lamination Λsu and the epsilon-approximating foliations W^cs_epsilon of the C1 branching foliation have uniformly Gromov hyperbolic leaves. This is the entry point for the entire ideal-boundary machinery (Section 3.1), the bounded-distance dichotomy (Proposition 4.3), the quasi-geodesic and topological Anosov flow results (Sections 5-6), and the non-separated-leaves argument (Section 7). The proof invokes Candel's uniformization theorem [Can93] and asserts, via [Cal01], that it applies to surface laminations or foliations 'with no regularity requirement for leaves' (Remark 8.7) and to the branching foliation W^cs. But W^cs is not a genuine lamination: its leaves are C1-immersed and may merge (Definition 2.13), and no argument is given that Candel's construction, which produces uniform hyperbolic metrics on leaves of a smooth lamination, survives the branching or the epsilon-to-0 limit. Proposition 8.9 cites [FP22, BFP23], but those references do not address branching foliations either. If the uniform hyperbolicity constants depend on epsilon or on the merging structure, then the leafwise geodesics, ideal circles, and transverse continuity used in Propositions 4.2, 5.1, 6.3, and 7.4 are not available, and Theorem 1.3 does not follow. This is not a disagreement with consensus; it is a missing proof at a load-bearing junction.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims to prove that a C^1 partially hyperbolic diffeomorphism homotopic to the identity on a closed 3-manifold with non-virtually-solvable fundamental group and NW(f)=M is accessible (Theorem 1.3). Combining this with known results [HHU08a, BW10] yields ergodicity for Cr conservative such diffeomorphisms unless there is an embedded 2-torus tangent to Es⊕Eu (Theorem 1.2), giving a positive answer to the Hertz-Hertz-Ures Ergodicity Conjecture in the homotopy class of the identity. The proof reduces the problem to a study of two transverse R-covered foliations with Gromov hyperbolic leaves, their intersection foliation, and the structure of ideal boundaries. It introduces a number of intermediate results (Propositions 4.2, 4.3, 5.1, 5.7, 6.3, 7.1, Theorem 6.1) which are then applied in Section 8 to the su-lamination and the center-stable branching foliation.","tokens_in":42692,"tokens_out":9890,"duration_ms":87993,"significance":"If the main theorem is correct, this is a major advance: it settles the HHU Ergodicity Conjecture for the whole homotopy class of the identity, a class that includes the difficult cases of non-dynamically-coherent systems and non-periodic-point-free settings. The geometric machinery developed in Sections 3-7 is of independent interest, especially Theorem 1.7 on constructing a topological Anosov flow from a pair of transverse foliations. The paper is also notable for its ambition in removing geometric restrictions on the ambient manifold. However, the proof's reliance on a black-box extension of Candel's uniformization theorem to branching foliations and on several unpublished results [BFP25, FU24] means that the contribution is conditional on those external ingredients being fully supplied.","major_comments":[{"comment":"","section":"Section 8.3, Remark 8.7 and Proposition 8.9"},{"comment":"","section":"Section 8.4"},{"comment":"","section":"Theorems 7.1 and Lemma 6.6"},{"comment":"","section":"Section 4.2, Proposition 4.3 and Corollary 4.4"},{"comment":"","section":"Section 8.4, dense-limit-set case"}],"minor_comments":[{"comment":"","section":"Section 5.3, Proposition 5.7"},{"comment":"","section":"Section 6.3, proof of Theorem 6.1"},{"comment":"","section":"Throughout"},{"comment":"","section":"Theorem 7.1"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses a central open problem and the main strategy is coherent, but the proof as written has several load-bearing dependencies on unpublished work and on an unproved extension of Candel's theorem. The authors are clearly in a strong position to fill these gaps, and the results are likely to be correct, but the manuscript in its current form is not sufficiently self-contained for publication without substantial additions. I recommend major revision rather than rejection, provided the authors supply the missing statements/proofs or give precise references with verified hypotheses."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Pepe, here's the thing you need to know about arXiv:2506.00405. Feng and Ures claim the HHU Ergodicity Conjecture for every conservative partially hyperbolic diffeomorphism homotopic to the identity on a closed 3-manifold with non-solvable fundamental group. That is a big deal: it covers the whole identity homotopy class, not just Seifert or hyperbolic bases. The proof is a serious piece of machinery, and I think it is likely right, but it has two soft spots that a referee must push on.\n\nThe genuinely new part is Theorem 1.3, accessibility under NW(f)=M and non-solvable pi1. The strategy is to reduce to an invariant su-lamination, collapse complementary I-bundles, and then apply a general theory of transverse foliations with Gromov hyperbolic leaves. Sections 3-7 are a self-contained reservoir of such results: quasi-geodesic subfoliations, degenerate limit sets, topological Anosov flows, non-separated leaves. Some of this is adapted from the authors' earlier work [FU24] and from Fenley-Potrie, but the synthesis here is new and useful beyond the immediate application.\n\nThe proof is honest about its debts. It leans on a lot of prior work, some still in preprint: [BFP25], [FP23a], [FP23b], [FU24]. That alone is not a flaw, but it does make the chain of dependencies long.\n\nThe real question is Section 8.3. The whole ideal-boundary argument needs leaves of the su-lamination and of the approximating foliations W^cs_epsilon to be uniformly Gromov hyperbolic. The authors cite Candel's theorem and say, via Calegari [Cal01], that it applies to laminations and foliations with no regularity requirement on leaves. That may be true, but it is not proved, and the branching foliation W^cs is not a genuine foliation—leaves can merge. Even though the arguments are applied to W^cs_epsilon, which is a true foliation, the uniform constants have to be controlled as epsilon goes to zero, and that control is not shown. The stress-test note is right: this is a missing proof at a load-bearing junction, not a mere technicality.\n\nIf that junction holds, the rest of the proof is coherent. I did not find an internal contradiction or a circular step. The paper is dense but well organized, and the main theorems are new.\n\nWho is this for? Anyone working on accessibility, ergodicity, or 3D partially hyperbolic dynamics. It deserves a serious referee, and I would send it to a top dynamics journal with a referee asked to check the Candel extension and the status of [BFP25]. My own verdict is conditional: I would not desk-reject, but I would not accept until the hyperbolicity step is either proved or pinned down to a precise theorem in the literature.","headline":"A serious and likely correct proof of the HHU ergodicity conjecture in the identity homotopy class for non-solvable 3-manifolds, but the Gromov-hyperbolicity step for branching foliations is asserted rather than proved.","tokens_in":43315,"tokens_out":4670,"would_cite":true,"duration_ms":39872,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37A25","37C86","37D30","57R30"],"pacs":[],"model":"deepseek-v4-flash","headline":"A partially hyperbolic diffeomorphism homotopic to the identity on a closed 3-manifold is accessible whenever the fundamental group is not virtually solvable and every point is non-wandering.","keywords":["partial hyperbolicity","accessibility","ergodicity","foliations","3-manifolds","homotopy class of identity","non-wandering set","Gromov hyperbolic leaves"],"falsifier":"A counterexample would be a $C^1$ partially hyperbolic diffeomorphism homotopic to the identity on a closed 3-manifold with non-virtually solvable fundamental group and $NW(f)=M$ that has two points not joined by any stable-unstable path; the structure theory would then produce an invariant su-lamination, and one could check directly whether its lifted leaves are uniformly Gromov hyperbolic as Theorem 8.6 asserts.","tokens_in":42121,"feed_emoji":"🌀","tokens_out":11586,"duration_ms":100930,"temperature":0.7,"pith_summary":"This paper proves that partially hyperbolic diffeomorphisms homotopic to the identity on closed 3-manifolds are accessible — any two points can be joined by a path made of stable and unstable arcs — provided the fundamental group is not virtually solvable (no finite-index solvable subgroup) and every point is non-wandering. The authors then combine this with earlier ergodicity criteria to show that a volume-preserving $C^r$ such diffeomorphism is a $K$-system, a strong form of ergodicity, unless there is an embedded 2-torus tangent to the stable-unstable distribution $E^s \\oplus E^u$. This gives an affirmative answer to the Ergodicity Conjecture for partially hyperbolic diffeomorphisms in dimension three, within the homotopy class of the identity. The proof runs through a geometric analysis of the invariant su-lamination and of the branching foliation along the center-stable direction, showing that any failure of accessibility would force a closed stable leaf or contradict the translation behavior of a good lift.","feed_headline":"3D maps homotopic to identity are ergodic unless tori block it","feed_subtitle":"For volume-preserving maps, accessibility rules out non-ergodicity except for torus obstructions.","key_machinery":"The engine is the pair $(F^{su}, W^{cs}_\\epsilon)$: $F^{su}$ is the minimal foliation obtained by collapsing the complementary I-bundle regions of the invariant su-lamination $\\Lambda^{su}$, and $W^{cs}_\\epsilon$ is a well-approximated foliation of the invariant branching foliation $W^{cs}$ tangent to $E^s \\oplus E^c$. Both are uniform R-covered minimal foliations by non-compact Gromov hyperbolic leaves, meaning their lifted leaf spaces are lines, pairs of lifted leaves lie at bounded Hausdorff distance, and leaves behave coarsely like the hyperbolic plane; their intersection is a one-dimensional foliation $G = F^{su} \\cap W^{cs}_\\epsilon$. The argument tracks the ideal limit set of $G$ inside the Gromov boundaries, the ideal circles at infinity, of the leaves: a dense limit set combined with Hausdorff or non-Hausdorff leaf spaces forces a closed $G$-leaf, hence a closed stable leaf, which partial hyperbolicity forbids; a degenerate limit set makes the unstable foliation regulating and produces an invariant leaf for the good lift, contradicting the fact that the lift translates the leaf space. A supporting structural theorem says that suitable transverse minimal R-covered foliation pairs with Gromov hyperbolic leaves form the weak-stable and weak-unstable foliations of a transitive topological Anosov flow.","core_discovery":"The central claim is Theorem 1.3: a $C^1$ partially hyperbolic diffeomorphism homotopic to the identity on a closed 3-manifold is accessible whenever the fundamental group is not virtually solvable and $NW(f)=M$. Accessibility means that the stable and unstable foliations together connect any two points. From this, the paper derives Theorem 1.2: a $C^r$ conservative (volume-preserving) such diffeomorphism is a $K$-system, and hence ergodic, unless it admits an embedded 2-torus tangent to $E^s \\oplus E^u$. The proof treats the non-accessible case as a rigid geometric configuration: the structure theory of non-accessible diffeomorphisms produces a minimal invariant lamination tangent to $E^s \\oplus E^u$; collapsing its complementary I-bundle regions turns it into a uniform R-covered minimal foliation with Gromov hyperbolic leaves; intersecting that foliation with a well-approximated center-stable branching foliation yields a one-dimensional intersection foliation whose ideal-boundary behavior gives the contradictions that prove accessibility.","pith_inferences":["The authors leave implicit that the technical Theorem 1.7 is a purely foliation-theoretic statement: two transverse minimal R-covered foliations with Gromov hyperbolic leaves and suitable intersection behavior always carry a transitive topological Anosov flow, so the same machinery may apply to classification problems outside partial hyperbolicity.","The proof uses unpublished results on non-separated leaves of transverse foliations; if those results require additional hypotheses, the affected step would need a replacement, while the overall dichotomy strategy might survive.","A natural testable extension is to drop the non-wandering assumption: related work on systems without periodic points suggests accessibility may hold in broader identity-homotopy classes, and the current proof indicates where such a generalization would have to intervene.","The $C^1$ accessibility result combined with the $C^r$ ergodicity criterion suggests that regularity of the measure, not accessibility, is what separates accessibility from ergodicity in this setting."],"forward_implications":["Theorem 1.2: every $C^r$ conservative partially hyperbolic diffeomorphism homotopic to the identity on a closed 3-manifold is a $K$-system unless an embedded 2-torus tangent to $E^s \\oplus E^u$ exists.","Corollary 1.4: under the same $C^1$ hypotheses, the diffeomorphism is transitive.","Theorem 1.5: for $C^r$ conservative such diffeomorphisms, transitivity and ergodicity are equivalent.","Corollary 1.6: if $NW(f)=M$ or $f$ is dynamically coherent, and no iterate is a discretized suspension Anosov flow, then $f$ is accessible; if it is $C^r$ and conservative, it is a $K$-system.","The non-accessible case is confined to manifolds with virtually solvable fundamental group, so the obstruction to ergodicity in this homotopy class is purely algebraic."],"supporting_citations":[{"why":"Supplies the structure theorem that non-accessible $f$ with $NW(f)=M$ preserves a minimal lamination or foliation tangent to $E^s \\oplus E^u$.","marker":"[HHU08b]"},{"why":"Provides the existence of invariant branching foliations $W^{cs}$ and $W^{cu}$ well-approximated by foliations.","marker":"[BI08]"},{"why":"Gives leaf-wise uniform Gromov hyperbolicity of the su-lamination and the good-lift dichotomy used in Proposition 8.4.","marker":"[FP22]"},{"why":"Candel's uniformization theorem underlies the claim that leaves carry uniform hyperbolic metrics.","marker":"[Can93]"},{"why":"Extends the uniformization theorem to laminations and foliations without leaf regularity, as invoked in Remark 8.7.","marker":"[Cal01]"},{"why":"Provides results on non-separated leaves of transverse R-covered minimal foliations used in Theorem 7.1.","marker":"[BFP25]"},{"why":"Establishes that accessibility implies ergodicity for conservative partially hyperbolic systems with one-dimensional center.","marker":"[HHU08a]"},{"why":"Completes the ergodicity criterion from accessibility for $C^r$ conservative partially hyperbolic diffeomorphisms.","marker":"[BW10]"},{"why":"Earlier work used as a black box for convergence of rays to ideal points and for existence of periodic points under recurrence assumptions.","marker":"[FU24]"},{"why":"Supplies the characterization of topological Anosov flows and branching-foliation leaf-space theory used in the quasi-geodesic fan arguments.","marker":"[BFP23]"}],"fun_headline_variants":["Ergodic for 3D maps homotopic to identity, except torus block","Accessibility holds unless ambient group is virtually solvable","Volume-preserving 3D maps: ergodic if not torus-obstructed","Hertz-Hertz-Ures conjecture solved for identity-homotopic maps","Torus obstruction is only barrier to ergodicity in 3D"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the uniformization theorem for surface laminations applies to the $C^1$ branching foliation and to the collapsed su-lamination, so all their leaves are uniformly Gromov hyperbolic; if that uniformity fails at branching or collapse points, the ideal-boundary and Anosov-flow constructions in the proof do not get off the ground.","fun_headline_variants_meta":{"raw":{"variants":["Ergodic for 3D maps homotopic to identity, except torus block","Accessibility holds unless ambient group is virtually solvable","Volume-preserving 3D maps: ergodic if not torus-obstructed","Hertz-Hertz-Ures conjecture solved for identity-homotopic maps","Torus obstruction is only barrier to ergodicity in 3D"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000775,"raw_usage":{"total_tokens":3367,"prompt_tokens":824,"completion_tokens":2543,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":440,"completion_tokens_details":{"reasoning_tokens":2444}},"tokens_in":440,"tokens_out":2543,"duration_ms":16649,"temperature":1.0,"reasoning_tokens":2444,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T12:06:01.588047+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A counterexample would be a $C^1$ partially hyperbolic diffeomorphism homotopic to the identity on a closed 3-manifold with non-virtually solvable fundamental group and $NW(f)=M$ that has two points not joined by any stable-unstable path; the structure theory would then produce an invariant su-lamination, and one could check directly whether its lifted leaves are uniformly Gromov hyperbolic as Theorem 8.6 asserts.","supporting_citations":[],"review_version":1}