{"id":"e21b76c7-72d2-440f-aaa9-72ee5a43bc0b","arxiv_id":"2607.16949","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":2,"one_line_summary":"C_u^*(X) can be a hereditary corner of C_u^*(Y) with no coarse embedding X→Y; sparse compact-ghost targets restore injective coarse embeddability.","lead":"This paper builds two countable metric spaces whose uniform Roe algebras sit inside one another as a hereditary corner, even though the first space cannot be coarsely embedded in the second — a negative answer to the embedding rigidity problem. It also shows that under a sparse compact-ghost condition, such corner embeddings do force injective coarse embeddings.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the counterexample and positive rigidity theorem both survive re-derivation; the probabilistic graph lemma is the most external dependency but appears sound.","rationale":"The reader's verdict is ACCEPT with high confidence, and my independent pass found no mathematical gap. The reader's weakest_assumption points to Lemma 3.3, the probabilistic construction of high-girth spectral-gap graphs; this is indeed the most external dependency, but it is sound: for fixed d and g, Friedman's theorem and McKay–Wormald–Wysocka give P(S_M) → 1 and P(G_M) → p > 0, so P(S_M ∩ G_M) ≥ P(G_M) − P(S_M^c) → p > 0. The other potentially delicate step, Lemma 3.5, is a probabilistic matching lemma with a counting estimate that is rough but valid; the supergraph argument for negative dependency is correct because enlarging the dependency graph only shrinks the set of allowed conditioning sets. The operator-algebraic core (Lemmas 5.1–5.3 and Proposition 5.4) is internally consistent: q is obtained by continuous functional calculus on a propagation-one operator; the corner identity follows from the two inclusions. The geometric obstruction (Proposition 4.2) correctly uses the recursive size condition and the tree-median lemma. On the positive side, Lemma 6.2 correctly manufactures a sparse subspace and a noncompact ghost projection if the coefficient lower bound fails; Lemma 6.3 is a standard Baire uniformization and is applied within its hypotheses; Proposition 6.5's Hall argument is valid, including the compactness step. Therefore I see no reason to change the reader's verdict. I mark agreement as partial because I would emphasize Lemma 3.5's counting estimate as the least transparent subargument, whereas the reader highlighted Lemma 3.3; neither is an actual flaw.","tokens_in":18311,"tokens_out":34680,"duration_ms":342413,"concrete_test":"For a concrete check, exhaustively enumerate all cycles up to length g in G ∪ K_{A,B} for a small instance (e.g., Δ = 3, g = 5, N = 10) and verify the counting bound (3.2) and the Lovász-local-lemma hypothesis of Lemma 3.5 directly; if the inequality fails for some vertex v and edge-count l, the matching-preserving-girth step would need repair.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I could not identify a real gap. The central construction requires Lemma 3.3 (arbitrarily large d-regular graphs with girth > g and normalized Laplacian spectrum {0} ∪ [γ,2]); the proof via Friedman + McKay–Wormald–Wysocka with the union bound is valid, since both events hold with positive limiting joint probability. Lemma 3.5's Lovász-local-lemma counting estimate (3.2) is the least transparent step, but the overcount by cycles is legitimate, the negative-dependency graph is a genuine supergraph of Lu–Székely's, and the N0 choice makes the LLL inequalities hold. The exact corner identity rests on Lemmas 5.2 and 5.3; I checked the lift of finite-propagation operators and the closed-range argument for arbitrary a ∈ C_u^*(X). The geometric obstruction in Proposition 4.2 is internally consistent: the recursion on L_n, the finite-to-one bound, and the tree-median contradiction all line up. The positive theorem's weak-to-strong upgrade via Lemma 6.2, Baire uniformization (Lemma 6.3), and Hall matching (Proposition 6.5) also checks out; the sparse compact-ghost hypothesis is invoked exactly where a noncompact ghost projection Q is manufactured, and the descent from global to sparse ghosts in Corollary 1.4 is fine. No ad hoc, circular, or unsupported step was found.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper addresses the embedding rigidity problem for uniform Roe algebras of uniformly locally finite coarse spaces. The main negative result (Theorem 1.2) constructs countable bounded-geometry metric spaces X and Y, a hereditary C*-subalgebra B of C_u^*(Y), and a *-isomorphism Φ : C_u^*(X) → B, while proving that X does not coarsely embed into Y. The construction uses high-girth graph bundles with expanding fibers over cycles: fiberwise averaging gives an isometry U, the uniform spectral gap puts q = UU* in C_u^*(Y), and a precise corner identity qC_u^*(Y)q = UC_u^*(X)U* is obtained via horizontal matchings. A tree-median/girth argument rules out coarse embeddings. The positive result (Theorem 1.3) shows that if every sparse subspace of Y yields only compact ghost projections, then any isomorphism from C_u^*(X) onto a hereditary subalgebra of C_u^*(Y) induces an injective coarse embedding X → Y, strengthening a theorem of Braga–Farah–Vignati. The proof combines spatial implementation, a ghost-projection coefficient lower bound, a Baire-category uniformization lemma, and Hall's marriage theorem.","tokens_in":18576,"tokens_out":26247,"duration_ms":225167,"significance":"If correct, these results settle Problem 1.1 in the negative, showing that hereditary C*-algebraic embeddings of uniform Roe algebras do not detect coarse embeddability even among bounded-geometry metric spaces. The positive theorem establishes that the sparse compact-ghost hypothesis, which is optimal in view of the counterexample, upgrades weak rigidity to injective rigidity. The paper is notable for its clean synthesis of probabilistic combinatorics (Friedman's theorem, McKay–Wormald–Wysocka, Lovász local lemma), geometric median arguments, and operator-algebraic techniques. The proofs are detailed and largely self-contained; the counterexample is sharp in that the obstruction is exactly a noncompact ghost projection, and the positive result is a genuine strengthening of [BFV20]. I checked the key derivations — the LLL counting in Lemma 3.5, the tree-median obstruction in Proposition 4.2, the corner identity in Lemmas 5.2–5.3, and the compact-ghost/Hall argument in Propositions 6.2–6.5 — and found no gaps.","major_comments":[],"minor_comments":[{"comment":"The notation 'A_S = {S⊆graph(σ)}' is confusing; it should read 'let A_S be the event that S ⊆ graph(σ)'. Also, in the counting estimate (3.2), the phrase 'Discarding configurations that fail to form a simple cycle or a partial matching only decreases the count' could be clarified: the overcount is by cycles, and the bound applies to the number of distinct events because each event is counted at least once. This is harmless but would improve readability.","section":"Lemma 3.5"},{"comment":"The sentence 'The graph just defined is a supergraph of that conflict graph' is technically true but slightly misleading: two partial matchings conflict precisely when they share an element of A or of B, so the graph defined is actually equal to the Lu–Székely conflict graph, not a proper supergraph. The subsequent conclusion (a supergraph of a negative dependency graph is again a negative dependency graph) is correct, so no mathematical issue arises.","section":"Lemma 3.5"},{"comment":"There is a typo in the proof: 'the finite intersection property. property.' should be 'the finite intersection property.'","section":"Proposition 6.5"},{"comment":"The phrase 'resolving the antipodal tie deterministically' is a bit vague. A few words specifying the tie-breaking rule (e.g., choose the positive orientation when the two shortest paths have equal length) would make the decomposition into partial translations fully explicit.","section":"Lemma 5.3"},{"comment":"The proof that for every T there exists S ∈ C_R attaining dist(T,C_R) is correct but somewhat nonstandard. Since C_R is weak-operator closed and convex, the standard Hilbert-space nearest-point theorem could be invoked; the current minimizing-net argument is valid but longer. No correction is needed.","section":"Lemma 6.3"}],"recommendation":"accept","confidential_remarks":"This is a strong paper: the main counterexample is a substantial and convincing resolution of a well-known open problem, and the positive rigidity theorem is a meaningful improvement over prior work. The proofs are carefully written and the probabilistic, geometric, and operator-algebraic components fit together cleanly. I found no load-bearing errors. The minor comments above are purely presentational. The paper is well suited for a top generalist journal or a specialized operator-algebra/coarse-geometry venue."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The one thing to know: the negative answer to Problem 1.1 is real. The construction of X and Y with C_u^*(X) a hereditary corner of C_u^*(Y) but no coarse embedding X→Y holds together. The exact corner identity qC_u^*(Y)q = UC_u^*(X)U* is the genuinely new operator-algebraic point; the high-girth bundle construction and the tree-median obstruction are clean. I checked the dependency-heavy parts — Friedman plus McKay–Wormald–Wysocka for the fiber graphs, the Lu–Székely LLL estimate in Lemma 3.5, and the lifting argument in Lemma 5.3 — and nothing collapses. The counting in Lemma 3.5 is coarse but legitimate, and the negative dependency graph claim is fine because a supergraph of one is still a negative dependency graph.\n\nThe positive theorem is also a real improvement: getting injective coarse embeddability from the sparse compact-ghost hypothesis, without property A, is not a trivial extension. The Baire-uniformization plus Hall-matching mechanism is the right tool, and the proof of the uniform Hall radius is the part I would ask a referee to read most carefully — it is the most intricate step, but it checks out on my reading.\n\nSoft spots are minor. The paper leans on deep external probabilistic results, so a fully self-contained verification is not quick, but that is not a defect in the mathematics. There are small typos (e.g., the doubled \"property. property.\" in the proof of Proposition 6.5) and the exposition in Section 5 is dense but not unfair. The self-citation to [Zha26] is natural given the companion paper resolving isomorphism rigidity; I do not see it as a problem.\n\nThe central argument holds up. This is a paper that answers an open problem in the negative and sharpens a known positive result; it deserves a serious referee. My recommendation: send it out, with attention to Section 6.3 and the LLL estimate. I would take it for peer review.","headline":"This paper resolves the embedding rigidity problem in the negative with a well-built counterexample and upgrades weak to strong rigidity under the sparse compact-ghost hypothesis — solid work that deserves a serious referee.","tokens_in":19113,"tokens_out":890,"would_cite":true,"duration_ms":11641,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46L05","51F30","05C80","46L85"],"pacs":[],"model":"deepseek-v4-flash","headline":"Two countable metric spaces exist whose uniform Roe algebras are isomorphic to a hereditary corner, yet no coarse embedding exists between the spaces.","keywords":["uniform Roe algebra","coarse embedding","hereditary C*-subalgebra","expander graph","high-girth graph bundle","ghost projection","embedding rigidity","Hall's marriage theorem"],"falsifier":"One could attempt to replace the random regular graph input with an explicit construction of arbitrarily large d-regular graphs with girth > g and Laplacian spectrum in {0} ∪ [γ,2]. If no such family exists for every g and N, the counterexample collapses. More directly, a reader could test the tree-median obstruction on a small case: if a coarse embedding of a cycle C_L into the bundle existed, the lemma forces two base vertices at distance at least ⌊L/3⌋/2 with images within 2R, contradicting effective properness; a numerical check for a chosen L would confirm the geometric no-go.","tokens_in":1390,"feed_emoji":"🧩","tokens_out":2910,"duration_ms":57788,"temperature":0.7,"pith_summary":"The paper addresses the embedding rigidity problem: if the uniform Roe algebra of one countable metric space is isomorphic to a hereditary C*-subalgebra of another's, must the first coarsely embed into the second? It answers no. The author constructs countable uniformly locally finite metric spaces X and Y with a *-isomorphism from C_u^*(X) onto a hereditary corner of C_u^*(Y), while X does not coarsely embed into Y. The construction uses graph bundles with expanding, high-girth fibers; the algebra isomorphism comes from fiberwise averaging. The paper also proves a positive result: if every sparse subspace of Y yields only compact ghost projections, then a hereditary isomorphism forces an injective coarse embedding X → Y.","feed_headline":"Roe algebra isomorphism without coarse embedding","feed_subtitle":"A hereditary corner of C_u^*(Y) equals C_u^*(X), yet no coarse map X→Y exists; rigidity needs extra assumptions.","key_machinery":"The load-bearing object is the fiberwise-averaging isometry U: ℓ²(X) → ℓ²(Y), spreading each basis vector of a base cycle evenly over a fiber, and the projection q = UU^*. The exact corner identity qC_u^*(Y)q = UC_u^*(X)U^* makes the hereditary range possible. High-girth expanding fibers, obtained by combining probabilistic results on random regular graphs—optimal control of the second adjacency eigenvalue and positive limiting probability for high girth—together with the Lovász local lemma for random injections, ensure both q ∈ C_u^*(Y) (via the uniform spectral gap and functional calculus) and the geometric nonembedding (via the tree-median obstruction).","core_discovery":"The central discovery is a counterexample to embedding rigidity. X is a coarse disjoint union of cycles, Y a coarse disjoint union of graph bundles over those cycles, each bundle having fibers that are fixed-degree expanders of arbitrarily high girth and horizontal edges that are perfect matchings along each base edge. The fiberwise average map U sends ℓ²(X) into ℓ²(Y); the uniform spectral gap makes q = UU^* a projection in C_u^*(Y), and the matching structure yields the exact corner identity qC_u^*(Y)q = UC_u^*(X)U^*. Thus C_u^*(X) is isomorphic to a hereditary subalgebra of C_u^*(Y). On the geometric side, the recursive growth of base cycles together with high girth rules out coarse embed","pith_inferences":["A natural next question is whether the counterexample can be realized with both spaces having property A, which would push the boundary of the positive theorem further.","The graph-bundle construction suggests that rigidity depends not only on base geometry but on the algebra's ability to detect fiberwise expansion; this may transfer to other C*-algebras built from coarse spaces.","One could test the sharpness of the sparse compact-ghost hypothesis by varying fiber growth: faster growth may break the corner identity, slower growth may allow a coarse embedding."],"forward_implications":["The embedding rigidity problem is answered negatively: hereditary C*-isomorphism of uniform Roe algebras does not imply coarse embeddability.","The sparse compact-ghost hypothesis is sufficient for strong rigidity: it upgrades known weak embedding to an injective coarse embedding.","The constructed projection q is a noncompact ghost projection, so the counterexample violates the hypothesis under which rigidity holds.","By known results linking ghost projections to property A, Y lacks property A, placing the example outside positive rigidity theorems.","The exact corner identity gives a template for building further hereditary isomorphisms from graph-bundle constructions."],"fun_headline_variants":["Roe algebra corner equality without coarse embed","C* corner isomorphism that forbids coarse embedding","Coarse embed impossible despite C* corner match","Rigidity fails: C* corner equality, no coarse map","C_u^* corner trick with no coarse embedding"],"cache_read_input_tokens":20352,"weakest_assumption_plain":"The counterexample depends on the existence, for every girth and size demand, of regular graphs that are simultaneously expanders (normalized Laplacian spectrum in {0} ∪ [γ,2]) and have girth larger than g; this probabilistic existence is the load-bearing premise—if it failed, q would not lie in C_u^*(Y) and the corner identity would break.","fun_headline_variants_meta":{"raw":{"variants":["Roe algebra corner equality without coarse embed","C* corner isomorphism that forbids coarse embedding","Coarse embed impossible despite C* corner match","Rigidity fails: C* corner equality, no coarse map","C_u^* corner trick with no coarse embedding"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000265,"raw_usage":{"total_tokens":1421,"prompt_tokens":695,"completion_tokens":726,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":439,"completion_tokens_details":{"reasoning_tokens":652}},"tokens_in":439,"tokens_out":726,"duration_ms":7493,"temperature":1.0,"reasoning_tokens":652,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T19:29:54.332100+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"One could attempt to replace the random regular graph input with an explicit construction of arbitrarily large d-regular graphs with girth > g and Laplacian spectrum in {0} ∪ [γ,2]. If no such family exists for every g and N, the counterexample collapses. More directly, a reader could test the tree-median obstruction on a small case: if a coarse embedding of a cycle C_L into the bundle existed, the lemma forces two base vertices at distance at least ⌊L/3⌋/2 with images within 2R, contradicting effective properness; a numerical check for a chosen L would confirm the geometric no-go.","supporting_citations":[],"review_version":1}