{"id":"8b54656d-3db8-46a0-8248-ef853eb14432","arxiv_id":"1909.02628","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"In dimensions n at least 4, S^2 x S^(n-2) is not cancellable, so connected sum decomposition is not unique; for simply connected manifolds the same failure is claimed for n at least 17.","lead":"This paper studies when closed manifolds in dimensions four and higher have unique prime decompositions under connected sum, as three-manifolds do by the Kneser-Milnor theorem. It proves that uniqueness fails broadly, including for many simply connected manifolds, by constructing explicit non-cancellable summands.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 6.1's proof invokes Corollary 4.7 outside its hypotheses: C_alpha and C_beta are 8-dimensional but the corollary is applied with n=4, yielding S^8 x S^(k-8), not the stated S^5 x S^(k-5).","rationale":"The reader's weakest assumption identifies exactly the same proof-statement mismatch: Corollary 4.7 is applied to 8-dimensional complexes with n = 4, outside the corollary's hypothesis that the complexes have dimension at most n. The forced correction n = 8 produces S^8 × S^(k-8) rather than S^5 × S^(k-5), so Theorem 1.4 as stated is not proved. I checked whether an alternative construction could justify the S^5 summand: the mapping cones are built from maps S^7 → S^4, so the top cell is in dimension 8, and the wedge sphere produced by Theorem 6.2 is S^8. There is no step that lowers this to a 5-dimensional sphere. The main theorem Theorem 1.3 is unaffected because it uses 2-dimensional presentation complexes and a legal application of Corollary 4.7 with n = 2, yielding S^2 × S^(k-2). The qualitative conclusion that MCat,sc_n is not a unique factorisation monoid for large n remains credible, since the corrected S^8-based construction still gives a non-cancellable element in that monoid. Thus the appropriate disposition is the same as the reader's: conditional acceptance pending a corrected statement and proof for the simply connected case. My agreement is with the reader's identification of the weak assumption; no additional load-bearing concern emerged from the rest of the paper. The proofs relying on Metzler's complexes, Wall's classification, and the thickening technology are internally coherent, and the main non-simply-connected theorem has independent support from cited theorems.","tokens_in":22458,"tokens_out":6348,"duration_ms":67288,"concrete_test":"Locate the application of Corollary 4.7 in the proof of Theorem 6.1 and rewrite it with the legally forced parameter n = 8 (matching dim C_alpha = dim C_beta = 8 and the wedge sphere S^8). Verify that the conclusion is M^k(C_alpha)#(S^8 × S^(k-8)) ≅ M^k(C_beta)#(S^8 × S^(k-8)) for k ≥ 17. Then search the proof for any additional step that replaces S^8 by S^5; if no such step exists, Theorem 1.4 must be restated with the summand S^8 × S^(k-8) rather than S^5 × S^(k-5).","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof of Theorem 6.1 constructs alpha, beta in pi_7(S^4), so the mapping cones C_alpha and C_beta have cells in dimensions 0, 4, and 8 and are therefore 8-dimensional. Theorem 6.2 gives C_alpha ∨ S^8 ≃_s C_beta ∨ S^8. The proof then says Theorem 1.4 follows from Corollary 4.7 applied to these 8-dimensional complexes with n = 4 and k ≥ 17. This is not a legal application: Corollary 4.7 requires dim(X), dim(Y) ≤ n, and the wedge sphere in its hypothesis is S^n. The data supplied have dim(C_alpha) = dim(C_beta) = 8 and wedge sphere S^8, forcing n = 8, not n = 4. With n = 8, Corollary 4.7 concludes M^k(C_alpha)#(S^8 × S^(k-8)) ≅ M^k(C_beta)#(S^8 × S^(k-8)), which proves non-cancellability of S^8 × S^(k-8), not S^5 × S^(k-5). No later argument in Section 6 or elsewhere converts the S^8 summand into an S^5 summand; the two are not diffeomorphic in the relevant dimensions. Therefore Theorem 6.1 and Theorem 1.4 as stated are unsupported. This does not affect Theorem 1.3, whose use of Corollary 4.7 with genuine 2-complexes is legitimate, and it does not destroy the qualitative conclusion that simply connected high-dimensional monoids are not unique factorisation monoids, since a corrected statement with S^8 would suffice. But the specific published claim about S^5 × S^(n-5) requires either a new construction or a revised statement.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the monoids of oriented closed manifolds under connected sum in the categories Top, PL, and Diff, for dimensions n ≥ 1. It formalizes unique factorization monoids and, after establishing an existence result for prime decompositions (Proposition 1.2), proves that uniqueness fails in high dimensions: Theorem 1.3 states that for every n ≥ 4 the manifold S^2 × S^{n−2} is not cancellable in the reduced monoid, so the monoid is not a unique factorization monoid; Theorem 1.4 claims the analogous statement for simply connected manifolds for n ≥ 17 using S^5 × S^{n−5}. The proofs use Wall's thickening construction, Metzler's and Browning's non-cancellation results for 2-complexes, and Hilton's mapping cones of sphere maps. The paper also surveys Wall's classification results for highly connected manifolds, proves that the Wu manifold is prime in the simply connected 5-dimensional monoid, and discusses the D2 problem and aspherical manifolds.","tokens_in":22766,"tokens_out":6825,"duration_ms":60328,"significance":"The paper's main message—that unique connected-sum factorization, valid in dimension 3, fails in high dimensions and even in the simply connected case—is interesting and important. The method of converting non-cancellation phenomena for CW-complexes into non-cancellation of manifolds via thickenings is elegant and carefully explained. The proof of Theorem 1.3 appears sound. However, the proof of Theorem 1.4 contains a dimension error: Corollary 4.7 is applied outside its hypotheses. The data only support a corrected statement with S^8 × S^{k−8} instead of S^5 × S^{k−5}. Since the qualitative conclusion (non-UFD for simply connected high-dimensional monoids) survives this correction, the paper's central thesis is defensible, but the stated Theorem 1.4 requires revision.","major_comments":[{"comment":"Corollary 4.7 is applied to the 8-dimensional complexes Cα and Cβ with n = 4. This violates the hypothesis of Corollary 4.7 that the complexes have dimension at most n. Moreover, Theorem 6.2(2) gives Cα ∨ S^8 ≃s Cβ ∨ S^8, so the sphere in the wedge is S^8, not S^4; hence the only legal application is with n = 8. With n = 8 the conclusion is M^k(Cα)#(S^8 × S^{k−8}) ≅ M^k(Cβ)#(S^8 × S^{k−8}), which would prove non-cancellability of S^8 × S^{k−8}. No argument is supplied that converts the S^8 summand into an S^5 summand, and the two are not diffeomorphic in the relevant range.","section":"§6, proof of Theorem 6.1"},{"comment":"Because the proof of Theorem 6.1 only supports the S^8 version, Theorem 1.4 as stated (\"S^5 × S^{n−5} is not cancellable for n ≥ 17\") is not established. The error does not affect Theorem 1.3, whose application of Corollary 4.7 to genuine 2-complexes is legitimate. I recommend either supplying a new construction that yields S^5 summands or restating the theorem with S^8 × S^{n−8}; the latter still gives the main qualitative conclusion.","section":"§1, Theorem 1.4; §6, Theorem 6.1"}],"minor_comments":[{"comment":"\"Theorem 4.7\" should be \"Corollary 4.7\".","section":"§6, first paragraph"},{"comment":"The cross-reference \"page ??\" is unresolved; it should point to the relevant remark in Section 6.","section":"§8, Question 8.3"},{"comment":"There are several typographical errors such as \"orient ed\" and \"definition\" that should be corrected in a final revision.","section":"Abstract and Section 1"},{"comment":"The displayed expression with braces under \"(r+1)·(S2×S2)\" is garbled; it should clearly state that the connected sum is taken r+1 times.","section":"§5.2, proof of Theorem 5.3"}],"recommendation":"major_revision","confidential_remarks":"The flaw in Section 6 is localized and fixable: the authors should either restate Theorem 1.4 with S^8 in place of S^5 or provide a genuinely new construction. I would not reject on this basis, but the published version must correct the statement."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take on arXiv:1909.02628. The main result is Theorem 1.3: S^2 × S^(n−2) is not cancellable in any of the monoids for n ≥ 4, so none of these monoids is a unique factorisation monoid. The proof is a clean application of Wall thickenings to Metzler's non-homotopy-equivalent 2-complexes with a common wedge with S^2. That argument checks out—the use of Corollary 4.7 is legitimate there because the complexes are genuinely 2-dimensional. This is a satisfying negative answer to a natural generalization of Kneser–Milnor, and it is new, as far as I can tell.\n\nThe simply connected part is where I part ways with the current text. The proof of Theorem 6.1 builds 8-dimensional complexes C_α and C_β and then applies Corollary 4.7 with n = 4. That violates the hypothesis: the corollary requires the complexes to have dimension at most n, and the wedge sphere needs to be S^n. A legal application uses n = 8 and yields diffeomorphisms after connected sum with S^8 × S^(k−8), not S^5 × S^(k−5). So Theorem 1.4 and Theorem 6.1 as stated are not supported. The qualitative claim that simply connected monoids fail unique factorisation for large n is probably still true, since an S^8 version would give the same conclusion, but the authors need to either correct the statement or supply a different construction.\n\nThe rest of the paper is largely fine. The complexity argument for decompositions, the Wall review, the prime Wu manifold, and the Borel conjecture remark are all useful. The citation pattern is honest; the central arguments rest on external theorems, not on the authors' own work. Minor issues: there is a broken page reference '??' in Section 8, and an 'orientation presentation' typo in the proof of Theorem 5.3, but those are cosmetic.\n\nNet: conditional. Send it to a referee. Theorem 1.3 alone justifies that, and the simply connected gap is isolated and likely fixable. I'd bring it to the reading group, but flag the proof issue in advance.","headline":"Main theorem 1.3 is solid and new; Theorem 1.4 is unsupported as written due to an illegal application of Corollary 4.7 — fixable, but needs a revision.","tokens_in":23410,"tokens_out":2829,"would_cite":true,"duration_ms":28883,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["57R65"],"pacs":[],"model":"deepseek-v4-flash","headline":"Connected-sum decompositions of closed manifolds lose uniqueness in every dimension above three, in all three manifold categories.","keywords":["connected sum","unique factorization monoid","cancellation","high-dimensional manifolds","simply connected manifolds","CW complexes","thickenings","Kneser-Milnor theorem"],"falsifier":"Recompute the proof of Theorem 6.1 with k = 17: Corollary 4.7, applied honestly with n = 8 to the 8-dimensional mapping cones $C_\\mu$ and $C_{5\\mu}$, gives a diffeomorphism $M^{17}(C_\\mu)\\#(S^8 \\times S^9) \\cong M^{17}(C_{5\\mu})\\#(S^8 \\times S^9)$ together with non-homotopy of the two 17-manifolds; the stated theorem requires the same with $S^5 \\times S^{12}$. A direct surgery-theoretic computation of whether $M^{17}(C_\\mu)\\#(S^5 \\times S^{12})$ and $M^{17}(C_{5\\mu})\\#(S^5 \\times S^{12})$ are diffeomorphic would settle Theorem 1.4 as stated, and a negative answer would disprove it.","tokens_in":22180,"feed_emoji":"🧩","tokens_out":13053,"duration_ms":132583,"temperature":0.7,"pith_summary":"In three dimensions, the Kneser-Milnor theorem gives every closed oriented connected manifold a unique decomposition into prime summands under connected sum. This paper asks how much of that structure survives in higher dimensions and answers: very little. It proves that for every n ≥ 4 and in each of the smooth, piecewise-linear, and topological categories, the manifold $S^2 \\times S^{n-2}$ is not cancellable in the connected-sum monoid, and consequently none of these monoids is a unique factorisation monoid; the same failure occurs among simply connected manifolds for n ≥ 17 with $S^5 \\times S^{n-5}$ in place of $S^2 \\times S^{n-2}$. The upshot is that the clean uniqueness of 3-dimensional prime decompositions has no straightforward extension to high-dimensional manifolds.","feed_headline":"High-dimensional manifolds break unique prime decomposition","feed_subtitle":"In dimensions 4 and up, adding the same summand can make distinct manifolds diffeomorphic.","key_machinery":"The engine is Wall's thickening construction: for a finite n-dimensional complex X and k ≥ 2n+1, k ≥ 6, there is an essentially unique smooth k-manifold $N^k(X)$, a 'thickening' with trivial tangent bundle and a simple homotopy equivalence $X \\to N^k(X)$; its boundary $M^k(X) = \\partial N^{k+1}(X)$ is a closed k-manifold that respects wedges, $M^k(X \\vee Y) \\cong M^k(X)\\#M^k(Y)$, and, when k ≥ 2n+1, remembers the homotopy type of X. This turns purely homotopy-theoretic non-cancellation of complexes—$X \\not\\simeq Y$ but $X \\vee S^r \\simeq_s Y \\vee S^r$—into diffeomorphism-level non-cancellation of manifolds. The complex pairs are produced by two machines: presentations with equal deficiency but distinct homotopy types, stabilised by a theorem on finite 2-complexes with the same finite fundamental group and Euler characteristic, and a mapping-cone criterion applied to the elements $\\mu$ and $5\\mu$ of order 12 in $\\pi_7(S^4)$, using Toda's calculation that $\\mu$ suspends to generate the relevant torsion.","core_discovery":"The paper's central discovery is a systematic mechanism for manufacturing non-cancellation in the monoids of closed oriented n-manifolds. Starting from pairs of finite cell complexes X, Y that are not homotopy equivalent but become simple-homotopy equivalent after wedging with a sphere, Wall's thickening construction produces closed n-manifolds $M^n(X)$, $M^n(Y)$ that are not homotopy equivalent yet satisfy $M^n(X)\\#(S^2 \\times S^{n-2}) \\cong M^n(Y)\\#(S^2 \\times S^{n-2})$ for n ≥ 5, with a stabilized analogue for n = 4. The needed complex pairs come from known presentations of $(\\mathbb{Z}/p)^s$ with equal deficiency but different homotopy type, combined with a theorem that such complexes become simple-homotopy equivalent after wedging with $S^2$. For simply connected manifolds, the paper uses mapping cones of the elements $\\mu$ and $5\\mu$ in $\\pi_7(S^4) \\cong \\mathbb{Z}/12$, whose wedge-with-$S^8$ stabilisations are simple-homotopy equivalent, to obtain the analogous result with $S^5 \\times S^{k-5}$ for k ≥ 17. The proof shows that failure of unique factorisation is detected already by the additive structure of connected sums, before any discussion of primeness.","pith_inferences":["The construction is a template: any pair of finite complexes X, Y with $X \\not\\simeq Y$ but $X \\vee S^r \\simeq_s Y \\vee S^r$ yields, via the same thickening corollary, a non-cancellable summand $S^r \\times S^{k-r}$ in some high dimension; the paper's two theorems are two instances of one mechanism.","Because the simply connected bound n ≥ 17 comes from the particular mapping-cone pair, analogous pairs in lower homotopy stems, or algebraic 2-complexes over groups with periodic cohomology, should push the failure down to much lower dimensions; the search is naturally phrased in homotopy theory rather than manifold topology.","The gap flagged in the proof of Theorem 6.1 matters: if it cannot be repaired, the honest output of that argument is non-cancellation of $S^8 \\times S^{k-8}$ rather than $S^5 \\times S^{k-5}$, leaving Theorem 1.4 true in spirit but unproven in the stated form.","In dimension 4, gauge-theoretic and topological classification results prevent the simply connected analogue: distinct smooth simply connected 4-manifolds that become diffeomorphic after summing with $S^2 \\times S^2$ are already homeomorphic, so any 4-dimensional non-cancellation has to exploit fundamental groups, exactly as the paper's Theorem 1.3 does."],"forward_implications":["For n ≥ 4, no monoid of closed oriented connected n-manifolds in the smooth, PL, or topological category has unique prime factorisation: the summand $S^2 \\times S^{n-2}$ can be added to different manifolds and make them diffeomorphic.","Cancellation fails in the strongest concrete form: an equality $M\\#(S^2 \\times S^{n-2}) \\cong N\\#(S^2 \\times S^{n-2})$ does not imply $M \\cong N$, even when M and N have different homotopy types.","The failure persists inside the simply connected submonoid for n ≥ 17, with $S^5 \\times S^{n-5}$ as the non-cancellable summand.","Uniqueness is not totally absent: in highly connected even dimensions with $k \\equiv 3, 5, 7 \\bmod 8$ and $k \\neq 15, 31, 63$, the monoid is isomorphic to $\\mathbb{N}$ via half the rank of the middle homology, and the Wu manifold $SU(3)/SO(3)$ is prime in the simply connected 5-dimensional monoid.","Assuming the Borel conjecture, the monoid generated by aspherical topological manifolds of dimension at least 4 is a unique factorisation monoid, so any higher-dimensional failure of uniqueness must come from non-aspherical summands."],"supporting_citations":[{"why":"Supplies the first presentations with equal deficiency but non-homotopy-equivalent Cayley complexes, detected by the bias invariant.","marker":"[Me76]"},{"why":"Provides Theorem 5.4: finite 2-complexes with the same finite fundamental group and Euler characteristic become simple-homotopy equivalent after wedging with S^2.","marker":"[Bro79]"},{"why":"Develops the thickening theory used to turn CW-complex non-cancellation into manifold non-cancellation via Corollary 4.7.","marker":"[Wa66a]"},{"why":"Supplies Proposition 4.6(4), used to conclude thickened boundaries are not homotopy equivalent, and the stable classification theorem used in the 4-dimensional stabilized case.","marker":"[KS84]"},{"why":"Gives Theorem 6.2: mapping cones of finite-order suspension elements are non-equivalent yet become equivalent after wedging with a sphere.","marker":"[Hi67]"},{"why":"Provides the calculation $\\pi_6(S^3) \\cong \\mathbb{Z}/12$ and the suspension facts identifying the elements $\\mu$ and $5\\mu$ used in the simply connected construction.","marker":"[To62]"},{"why":"Wall's classification of highly connected 2k-manifolds underlies the positive unique-factorisation cases and the non-cancellation of $S^{2k} \\times S^{2k}$ in highly connected monoids.","marker":"[Wa62]"},{"why":"The Kneser-Milnor unique factorisation theorem for 3-manifolds is the classical baseline whose higher-dimensional analogue the paper disproves.","marker":"[Mi62]"}],"fun_headline_variants":["Adding same summand can make distinct manifolds diffeomorphic","Prime decomposition not unique for high-dimensional manifolds","Unique connected sum decomposition fails beyond dimension 3","Distinct high-dimensional manifolds can share a summand"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof of the simply connected case applies a cell-complex-to-manifold construction to 8-dimensional complexes while claiming only 4 dimensions are needed; the construction as written would produce sums with $S^8 \\times S^{k-8}$, not $S^5 \\times S^{k-5}$, so the stated Theorem 1.4 relies on a repair that is not written down.","fun_headline_variants_meta":{"raw":{"variants":["Adding same summand can make distinct manifolds diffeomorphic","Prime decomposition not unique for high-dimensional manifolds","Unique connected sum decomposition fails beyond dimension 3","Distinct high-dimensional manifolds can share a summand"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001109,"raw_usage":{"total_tokens":4578,"prompt_tokens":859,"completion_tokens":3719,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":475,"completion_tokens_details":{"reasoning_tokens":3655}},"tokens_in":475,"tokens_out":3719,"duration_ms":32285,"temperature":1.0,"reasoning_tokens":3655,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T04:45:35.407824+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the proof of Theorem 6.1 with k = 17: Corollary 4.7, applied honestly with n = 8 to the 8-dimensional mapping cones $C_\\mu$ and $C_{5\\mu}$, gives a diffeomorphism $M^{17}(C_\\mu)\\#(S^8 \\times S^9) \\cong M^{17}(C_{5\\mu})\\#(S^8 \\times S^9)$ together with non-homotopy of the two 17-manifolds; the stated theorem requires the same with $S^5 \\times S^{12}$. A direct surgery-theoretic computation of whether $M^{17}(C_\\mu)\\#(S^5 \\times S^{12})$ and $M^{17}(C_{5\\mu})\\#(S^5 \\times S^{12})$ are diffeomorphic would settle Theorem 1.4 as stated, and a negative answer would disprove it.","supporting_citations":[],"review_version":1}