{"id":"2d406066-83d3-4ca0-a02e-f04cc2a14739","arxiv_id":"2507.17557","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"If a compact ANR is a Gromov-Hausdorff limit of closed n-manifolds with bounded contractibility functions, then it is an open cell-like image of each sufficiently close approximating manifold, resolving Moore's conjecture in this setting.","lead":"This paper proves that a compact space which is a Gromov-Hausdorff limit of n-manifolds with a uniform contractibility function is itself a cell-like image of those manifolds, provided the limit is an ANR. This resolves a 1991 conjecture of Moore in the finite-dimensional ANR setting and yields new topological obstructions for approximating spaces by PL or Riemannian manifolds.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The paper invokes Grove–Petersen–Wu stability without verifying that its hypotheses are met, since Theorem 2.40 requires a finite-dimensional limit while Theorem A only assumes X is an ANR.","rationale":"The reader's verdict of CONDITIONAL and the identification of the GPW stability theorem as load-bearing are, in my view, correct. My read agrees that the central claim is plausible and that the first proof is a logical chain of cited results. However, I would sharpen the concern: the specific missing ingredient is not merely whether stability holds for non-smooth or four-dimensional manifolds, but whether the hypotheses of Theorem A match the theorem being cited. The paper's own Theorem 2.40 explicitly includes finite-dimensionality of X, while Theorem A only assumes X is an ANR. Since compact metric ANRs can be infinite-dimensional, the proof silently relies on an unstated implication. The second proof has additional gaps, such as the removal of the circle factor, but those gaps are secondary because the first proof is intended to be sufficient. If the stability theorem truly requires finite-dimensionality and the missing implication is false, the theorem would need revision; if the implication is true, the paper must supply the proof. In either case, the existing CONDITIONAL verdict stands, and the authors should be asked to state the precise stability theorem used and to address the finite-dimensionality of X. I do not see grounds for rejection, because the overall approach is coherent and the gap appears repairable.","tokens_in":22616,"tokens_out":15990,"duration_ms":177087,"concrete_test":"Extract the exact statement of the Grove–Petersen–Wu stability theorem from [51, 52] and compare it with the sequence {Z_i} in the first proof of Theorem A. Specifically, verify whether the theorem requires the GH limit to be finite-dimensional and whether it covers arbitrary metric topological n-manifolds (not just Riemannian) in dimension n=4. Then test the missing implication: prove that an ANR which is a GH limit of n-manifolds with a uniform contractibility function has covering dimension at most n, or construct an infinite-dimensional ANR counterexample analogous to Example 2.37. If neither can be done, the first proof's invocation of GPW stability is unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the first proof of Theorem A (Section 3, after the definition of {Z_i}), the authors write: 'The Grove–Petersen–Wu stability theorem [51, 52] implies that, for large i, j, the space Z_i is homeomorphic to Z_j.' This is the only step that converts Gromov–Hausdorff closeness into actual homeomorphisms, so the theorem's validity is genuinely load-bearing. But the version of the GPW theorem quoted in the paper itself, Theorem 2.40, includes the hypothesis that the limit X is finite-dimensional; Theorem A assumes only that X is an ANR. A compact metric ANR need not be finite-dimensional (for example, the Hilbert cube is an infinite-dimensional compact ANR), so the cited theorem does not obviously apply to the interleaved sequence {Z_i} constructed in the proof. The first line of the proof, 'By [52], X is a resolvable homology manifold,' has the same issue, since the paper's stated GPW resolvability result also assumes finite-dimensionality. The authors do not supply an argument that 'ANR plus GH limit of n-manifolds with a uniform contractibility function' implies finite-dimensionality. A related but secondary gap is that the stability theorem was originally proved for dimensions n≥5 (or with the α-approximation refinement needed for n=4), and the proof does not indicate which version is being cited. These are missing hypotheses in the argument, not internal contradictions, so the theorem may still be true, but the proof as written is incomplete.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves Theorem A: if {X_i^n} is a sequence of closed metric topological n-manifolds (n ≥ 4) with a uniform contractibility function converging in the Gromov–Hausdorff sense to a space X that is an ANR, then for all sufficiently large i, X is a cell-like image of X_i, and the cell-like map can be chosen open. Two proofs are offered: a short proof using an interleaving argument and the Grove–Petersen–Wu stability theorem, and a longer proof using the α-approximation theorem and controlled surgery. The paper derives several corollaries, including a proof of Moore's conjecture for finite-dimensional ANR limits, characterizations of Gromov–Hausdorff limits of PL and Riemannian manifolds, a Quinn-obstruction criterion, and applications to Alexandrov spaces, Wasserstein spaces, and diffeomorphism stability.","tokens_in":22865,"tokens_out":8771,"duration_ms":90310,"significance":"If the main theorem and corollaries are correct, the paper would unify and extend known results on resolutions of Gromov–Hausdorff limits, giving a short proof of Moore's conjecture in the finite-dimensional ANR setting and connecting Quinn's resolution obstruction to metric approximation. The paper is clearly written, contains useful background, and the mapping-cylinder lemma (Lemma 2.33) is a nice tool. However, the central proof relies on hypotheses that are not verified and, in one place, appears to apply the α-approximation theorem to a codomain that is not known to be a manifold. Because these gaps are load-bearing for Theorem A, the claims as stated are not established.","major_comments":[{"comment":"The proof begins 'By [52], X is a resolvable homology manifold.' This is the paper's Theorem 2.40, which requires the Gromov–Hausdorff limit X to be finite-dimensional. Theorem A only assumes X is an ANR, and a compact metric ANR need not be finite-dimensional (for example, the Hilbert cube). Example 2.37 even shows that infinite-dimensional limits of n-manifolds with a uniform contractibility function exist. The proof does not establish finite-dimensionality of X, so the existence of the resolution M and the subsequent application of the Grove–Petersen–Wu stability theorem are not justified. If the intended stability theorem has weaker hypotheses, it needs to be stated precisely and its hypotheses verified.","section":"Section 3, first proof"},{"comment":"The step 'By the α-approximation theorem, ... each (f_i, id, id) : X_i × S^1 × S^1 → X × S^1 × S^1 is ε-close to a homeomorphism F_i' is not supported by the stated Chapman–Ferry theorem (Theorem 2.22), which requires both domain and codomain to be closed metric n-manifolds. At this point X × S^1 × S^1 is not known to be a manifold; indeed, showing that X is resolvable (equivalently, that X × R^2 is a manifold, by Quinn's Theorem 2.13) is part of what the proof aims to achieve. This appears circular. The proof should either first establish that X × S^1 × S^1 is a manifold by a separate argument, or replace the target by a resolution M of X with a clear explanation of why that replacement is legitimate for the later diagram.","section":"Section 3, second proof"},{"comment":"The passage 'Repeating the argument ... we can remove the last circle factor ... and obtain a homeomorphism ĥ_1 : X_i → X_j' delegates a nontrivial step to references [93, 51, 52]. Since the desired conclusion is a homeomorphism between the original manifolds X_i and X_j, not between their products with S^1, this step must be stated and proved or the relevant lemma from the references should be quoted in sufficient detail. This is especially important in dimension n = 4, where product-structure phenomena differ from higher dimensions.","section":"Section 3, second proof"},{"comment":"The assertion that the cell-like map can be taken to be open is attributed to Walsh [92] without stating the applicable theorem or checking its hypotheses. Because openness is an explicit part of Theorem A, the relevant result from Walsh's paper should be quoted and its conditions (such as dimension and ANR hypotheses) verified.","section":"Section 3, first proof, final sentence"}],"minor_comments":[{"comment":"The abstract emphasizes 'finite-dimensional Gromov–Hausdorff limits,' but Theorem A states only that X is an ANR. If the finite-dimensional hypothesis is in fact needed for the proof, the theorem statement and all dependent statements should be aligned with the abstract.","section":"Abstract and Theorem A"},{"comment":"In the definition of ANR, 'M is a retract of some open subset U of Z' should specify that U contains M (i.e., U is an open neighborhood of M), which is the standard definition.","section":"Definition 2.1"},{"comment":"The notation h_i : X_{η_i} → X_{η_{i+1}} is slightly confusing because h_i is indexed by the subsequence index rather than by the original sequence index; a different indexing notation would improve readability.","section":"Section 3, second proof"},{"comment":"The name 'Philipp Reiser' is spelled 'Phillip Reiser' in the acknowledgements; the spelling should be made consistent.","section":"Acknowledgements"},{"comment":"The statement 'This example is true in all dimensions' is vague; it would be clearer to specify which assertions in the example are meant.","section":"Example 5.2"}],"recommendation":"major_revision","confidential_remarks":"The central issue is that Theorem A as stated may require an additional finite-dimensionality hypothesis, and the second proof contains a potentially circular application of the α-approximation theorem. These are fixable if the theorem is restated and the proofs are expanded, so I do not recommend rejection, but the current version should not be accepted without substantial revision. The paper's extensive applications rely heavily on Theorem A, so the proof gaps affect the entire contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a real result. The paper proves Moore's 1991 conjecture for finite-dimensional GH limits in the ANR setting — that a limit of closed n-manifolds with a uniform contractibility function is a cell-like image of the approximating manifolds, not just of some manifold. That sharpening is new, and the alternating-trick proof is genuinely elegant. Corollary D, the H^4 obstruction for PL-approximability, is also new and is a nice payoff. The paper gives two proofs, and the first one is short enough to present in a seminar.\n\nThe main soft spot is the step where the first proof cites Grove–Petersen–Wu for the homeomorphism between the interleaved manifolds. The version of GPW quoted in the paper (Thm 2.40) assumes the limit is finite-dimensional; Theorem A only assumes the limit is an ANR, and the authors never say why that is enough. The gap is probably repairable — a limit of LGC(n, ρ) spaces is LGC(n, ρ'), and an ANR that is LGC(n,ρ) has covering dimension at most n, so finite-dimensionality follows — but the argument is not in the paper. The same missing line affects the opening sentence 'By [52], X is a resolvable homology manifold.' This is a missing hypothesis check, not a flaw in the central idea.\n\nThe second proof is sketchier: the removal of the circle factor is deferred to references, and the map's openness is just a citation to Walsh. These are minor for a paper whose main claim is credible, but they should be tightened before publication.\n\nOne more comment: the abstract says 'finite-dimensional Gromov–Hausdorff limits, in the ANR setting,' while Theorem A omits 'finite-dimensional.' Aligning the statement with the proof would help — either add finite-dimensional as a hypothesis or spell out the implication above.\n\nOverall: the central theorem is likely true, the paper is honest about what it does, and the applications are interesting. It deserves a serious referee and, after a modest revision, publication. I would take it.","headline":"A credible, substantial proof of Moore's conjecture in the ANR setting, with a repairable gap in the first proof's use of Grove–Petersen–Wu.","tokens_in":23434,"tokens_out":6212,"would_cite":true,"duration_ms":66140,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53C23","54C55","57P05"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that Gromov–Hausdorff limits of closed n-manifolds with a uniform contractibility function are cell-like images of the approximating manifolds when n≥4 and the limit is an ANR.","keywords":["Gromov–Hausdorff convergence","uniform contractibility function","cell-like maps","homology manifolds","resolutions","ANR","Alexandrov spaces","PL-smoothing obstruction"],"falsifier":"The theorem predicts that every such sequence is eventually homeomorphic: there exists $N$ such that all $X_i$ with $i\\geq N$ are homeomorphic. A single explicit counterexample — a Gromov–Hausdorff-convergent sequence of closed topological $n$-manifolds with a uniform contractibility function and an ANR limit whose terms are not eventually all homeomorphic — would settle the claim, so the test is to construct or rule out such a sequence, beginning with $n=4$.","tokens_in":22376,"feed_emoji":"🧩","tokens_out":19182,"duration_ms":179503,"temperature":0.7,"pith_summary":"The paper proves a structural rigidity statement about Gromov–Hausdorff convergence in the presence of a uniform contractibility function. Fix $n\\geq 4$; if closed $n$-dimensional topological manifolds $X_i$, all sharing one contractibility function, converge to a compact space $X$, and $X$ happens to be an absolute neighborhood retract (ANR), then for every sufficiently large $i$ the limit $X$ is a cell-like image of $X_i$ — that is, $X$ is obtained from $X_i$ by crushing contractible-looking sets to points, and the crushing map can be chosen open. This settles, in the finite-dimensional ANR setting, a 1991 conjecture about the structure of such limits. The theorem yields exact characterizations of which metric spaces can be approximated by PL or Riemannian manifolds with a uniform contractibility function, including a cohomological obstruction in $H^4(X;\\mathbb{Z}_2)$ and applications to Alexandrov spaces, Wasserstein spaces, and exotic spheres.","feed_headline":"Manifold limits are cell-like images of the tail","feed_subtitle":"For n≥4, a Gromov–Hausdorff limit that is an ANR is a crushed version of almost every approximating manifold.","key_machinery":"The load-bearing object of the first proof is the interleaving, or alternating trick: take a resolution of the limit $X$ by a manifold $M$, use the mapping-cylinder metrization of a cell-like map to give $M$ metrics so that it also converges to $X$ with a contractibility function, and weave the two sequences into one sequence $\\{Z_i\\}$ sharing a single contractibility function. A cell-like map is one whose point preimages are null-homotopic in every neighborhood — roughly, it crushes compact contractible-looking sets to points. The stability theorem for sequences with a uniform contractibility function then forces $Z_i$ to be homeomorphic to $Z_j$ for all large $i,j$, so each $X_i$ is eventually homeomorphic to $M$ and $X$ is a cell-like image of $X_i$. The second proof replaces this with a direct controlled-topology construction: epsilon-homotopy equivalences from the approximating manifolds to $X$, the $\\alpha$-approximation theorem to lift them to homeomorphisms after crossing with $S^1\\times S^1$, the thin h-cobordism theorem to delete the torus factors, and a diagonal subsequence argument in the style of controlled surgery to produce a limiting cell-like resolution.","core_discovery":"On the paper's own terms, the central discovery is Theorem A: fix $n\\geq 4$ and let $\\{X_i^n\\}$ be a sequence of closed metric topological $n$-manifolds with a uniform contractibility function converging in the Gromov–Hausdorff sense to a compact space $X$. If $X$ is an ANR, then for all sufficiently large $i$ the space $X$ is a cell-like image of $X_i$, and the cell-like map can be chosen open. Every such limit was already known to be a resolvable homology manifold; the new content is that the resolving manifold can be taken to be the approximating manifolds themselves, not merely some abstract manifold, and that this holds with no curvature assumption beyond the contractibility function.","pith_inferences":["The alternating trick uses only the existence of a resolution, metric closeness, and a shared contractibility function, so the same interleaving should apply to any class of spaces with a reliable resolution theory; a testable next step is to run it for sequences of Alexandrov spaces rather than manifolds.","Computing the PL-smoothing obstruction $\\Delta(X)$ of the resolution gives a concrete way to decide, from purely topological data, whether a geodesic metric space is a Riemannian Gromov–Hausdorff limit with uniform contractibility; the paper leaves such computations for explicit examples open.","The failure of diffeomorphism stability under a uniform contractibility function suggests that smooth structures are not invariant under Gromov–Hausdorff closeness in this setting, so any stability theorem that preserves smooth structure must use curvature bounds rather than contractibility bounds alone."],"forward_implications":["For $n\\geq 4$, a compact $n$-dimensional metric space is a Gromov–Hausdorff limit of closed PL $n$-manifolds with a uniform contractibility function if and only if it is a cell-like image of such a manifold.","For $n\\geq 5$ resolvable ANR homology manifolds, there is a well-defined class $\\Delta(X)$ in $H^4(X;\\mathbb{Z}_2)$ — the PL-smoothing obstruction of the unique resolution — that vanishes exactly when $X$ admits approximation by PL $n$-manifolds with a uniform contractibility function.","For compact geodesic spaces of dimension $n\\geq 5$, being a Gromov–Hausdorff limit of closed Riemannian $n$-manifolds with a uniform contractibility function is equivalent to being resolvable by a smooth manifold and also equivalent to having $X\\times \\mathbb{R}^k$ smoothable for some $k\\geq 2$.","Manifolds lying close to the boundary of the class of $n$-manifolds with a fixed uniform contractibility function are eventually cell-like-related over the limit space itself, a refinement of earlier results that holds for $n\\geq 4$.","The diffeomorphism stability conjecture fails if the lower curvature bound is replaced by a uniform contractibility function: exotic spheres can converge, with such a function, to the standard sphere, so the tail is homeomorphic but not diffeomorphic."],"supporting_citations":[{"why":"supplies the stability theorem that interleaved sequences with a uniform contractibility function converging to the same limit are eventually homeomorphic, and shows limits are resolvable homology manifolds.","marker":"[52]"},{"why":"supplies the 1991 conjecture and the mapping-cylinder construction by which a cell-like image can be metrized into a path of manifolds converging to it with a uniform contractibility function.","marker":"[72]"},{"why":"provides the epsilon-homotopy equivalences between Gromov–Hausdorff close spaces with a uniform contractibility function, used to start the second proof.","marker":"[77]"},{"why":"gives the alpha-approximation theorem that turns controlled homotopy equivalences between manifolds into homeomorphisms, the main tool for lifting maps in the second proof.","marker":"[21]"},{"why":"extends the alpha-approximation theorem to dimension 4, making the n=4 case of Theorem A accessible.","marker":"[42]"},{"why":"provides the thin h-cobordism theorem and the uniqueness of resolutions used to remove torus factors and to make the invariant $\\Delta(X)$ well-defined.","marker":"[81]"},{"why":"gives the four- and five-dimensional thin h-cobordism and uniqueness-of-resolutions inputs needed for the n=4 case and for the definition of $\\Delta(X)$.","marker":"[82]"},{"why":"supplies the controlled-surgery construction of resolutions from converging maps that the second proof adapts to build the cell-like resolution directly.","marker":"[14]"},{"why":"gives the approximation of cell-like maps over geodesic spaces by Riemannian metrics with a uniform contractibility function, used in Corollary E and the smooth examples.","marker":"[41]"},{"why":"shows cell-like maps between ANRs can be modified to be open, which adds the openness conclusion in Theorem A.","marker":"[92]"}],"fun_headline_variants":["Manifold limits as cell-like images: Moore's conjecture solved","Gromov-Hausdorff limits of manifolds are cell-like images","ANR manifold limits are cell-like images of approximating manifolds","For n≥4, manifold limits are cell-like images"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the stability theorem's guarantee that two sequences of manifolds that share a contractibility function and both get arbitrarily close to the same limit in the Gromov–Hausdorff sense must be homeomorphic for all large indices; the first proof collapses if that guarantee fails for merely topological manifolds or in dimension 4.","fun_headline_variants_meta":{"raw":{"variants":["Manifold limits as cell-like images: Moore's conjecture solved","Gromov-Hausdorff limits of manifolds are cell-like images","ANR manifold limits are cell-like images of approximating manifolds","For n≥4, manifold limits are cell-like images"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000271,"raw_usage":{"total_tokens":1605,"prompt_tokens":898,"completion_tokens":707,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":514,"completion_tokens_details":{"reasoning_tokens":632}},"tokens_in":514,"tokens_out":707,"duration_ms":7301,"temperature":1.0,"reasoning_tokens":632,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T14:47:52.839268+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"The theorem predicts that every such sequence is eventually homeomorphic: there exists $N$ such that all $X_i$ with $i\\geq N$ are homeomorphic. A single explicit counterexample — a Gromov–Hausdorff-convergent sequence of closed topological $n$-manifolds with a uniform contractibility function and an ANR limit whose terms are not eventually all homeomorphic — would settle the claim, so the test is to construct or rule out such a sequence, beginning with $n=4$.","supporting_citations":[{"cited_title":"Controlled boundary and h-cobordism theorems","cited_arxiv_id":null,"evidence_quote":"gives the alpha-approximation theorem that turns controlled homotopy equivalences between manifolds into homeomorphisms, the main tool for lifting maps in the second proof."},{"cited_title":"Approximating topological metrics by Riemannian metrics","cited_arxiv_id":null,"evidence_quote":"extends the alpha-approximation theorem to dimension 4, making the n=4 case of Theorem A accessible."},{"cited_title":"Stability, Finiteness and Dimension Four","cited_arxiv_id":"2006.02450","evidence_quote":"provides the thin h-cobordism theorem and the uniqueness of resolutions used to remove torus factors and to make the invariant $\\Delta(X)$ well-defined."},{"cited_title":"Ends of maps. III. Dimensions4 and 5","cited_arxiv_id":null,"evidence_quote":"gives the four- and five-dimensional thin h-cobordism and uniqueness-of-resolutions inputs needed for the n=4 case and for the definition of $\\Delta(X)$."},{"cited_title":"Erratum: “Topology of homology manifolds","cited_arxiv_id":null,"evidence_quote":"supplies the controlled-surgery construction of resolutions from converging maps that the second proof adapts to build the cell-like resolution directly."},{"cited_title":"Topological finiteness theorems for manifolds in Gromov-Hausdorff space","cited_arxiv_id":null,"evidence_quote":"gives the approximation of cell-like maps over geodesic spaces by Riemannian metrics with a uniform contractibility function, used in Corollary E and the smooth examples."},{"cited_title":"Isotoping mappings to open mappings","cited_arxiv_id":null,"evidence_quote":"shows cell-like maps between ANRs can be modified to be open, which adds the openness conclusion in Theorem A."}],"review_version":1}