{"id":"38ae1d24-9ab8-4f24-8142-c37e90c71df4","arxiv_id":"2509.02883","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Higher-dimensional spherical links have Milnor invariants that grow at most polynomially with 1/thickness when a 1-dimensional component is present, and exponentially otherwise; both rates are asymptotically sharp, and a Freedman-Krushkal question is resolved.","lead":"The paper proves the first thickness bounds for Milnor linking invariants of higher-dimensional spherical links, showing a polynomial upper bound when a component is a circle and an exponential upper bound otherwise, with examples matching both rates. It applies these results to confirm a question of Freedman and Krushkal about 2-complexes whose embeddings are exponentially thin in 4-space.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Sharpness of exponential regime in Theorem A(iii) rests on the unproved embedding of the building block B; no thickness/reach control is supplied.","rationale":"The reader's verdict is CONDITIONAL, and I agree that CONDITIONAL is appropriate. However, the reader's own weakest_assumption (unproved Propositions 5.6 and 5.7) is not the most load-bearing point for the strongest claimed results. Those propositions affect repeated-index estimates and, in the reader's account, Theorem C, but for the m = 4 application the repeated components have p = m−2, so the needed doubling scale is Proposition 5.8, which is proved. The genuinely exposed load-bearing step for the paper's headline 'optimal asymptotic bounds' is the construction of the exponentially thick links in Theorem 2.3, specifically the informal embedding of the building block B. If that construction does not achieve thickness proportional to n^{-1}, the lower bound degrades and the exponential sharpness claim in Theorem A(iii) is unsupported. The reader did flag this as issue (3), but did not make it the primary concern; hence partial agreement. A rigorous construction or explicit coordinate model of B would settle the matter. Until then, the sharpness of the exponential regime should be regarded as conditional.","tokens_in":42039,"tokens_out":10200,"duration_ms":110285,"concrete_test":"Give an explicit embedding of the block B in the relevant dimension with the properties used in Theorem 2.3, together with a quantitative lower bound on its reach. For the special case S^2 ⊔ S^2 ⊔ S^3 → S^5, write coordinates for the embedded B in a ball of radius C n^{-1} whose outer and inner tori are isometric to the standard ∂(S^1 × D^2), and verify that the reach is ≥ c n^{-1}. For the general case, do the same for T1 = S^{p_d−q_1} × D^{q_1} and T′. If such a construction cannot be supplied with the stated thickness scale, the asymptotic sharpness of Theorem A(iii) should be weakened to a one-sided exponential upper bound.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim includes that the exponential upper bound exp(C(m,d)τ^{-m}) in Theorem A(iii) is asymptotically sharp. The lower-bound construction in Theorem 2.3 builds links with thickness εn^{-1} and Milnor invariant 2^{n^m} (under the natural exponential reading of '2nm'). This construction depends on embedding the block B = T1 \\ T′ in a small ball with two boundary components isometric to ∂T1, and then stacking n^m copies along a curve of thickness n^{-1}. For the special case S^2 ⊔ S^2 ⊔ S^3 → S^5, the only justification is the informal statement that one glues the standard embedding of B in R^3 with an isotopy that 'unwraps' the inner torus; for the general case, Theorem 2.3 says only 'as in the special case, we can embed B as a cobordism between two isometric copies of ∂T1.' No explicit embedding, no coordinate model, and no lower bound on the reach of the resulting embedding is provided. If the smallest ball containing such a B-embedding has radius R(n) ≫ n^{-1}, then the thickness τ of the constructed link is ≪ n^{-1}, and the invariant 2^{n^m} becomes exp(O(τ^{-m/α})) with α > 1, which does not attain the claimed exp(Θ(τ^{-m})) rate. Thus the optimality half of Theorem A(iii) is currently unproved. This is distinct from the repeated-index doubling issue: Theorem C's factor 10(q+1) in §7.2 comes from Proposition 5.8 (p = m−2 doubling, which is proved), not from the unproved Propositions 5.6–5.7. The exponential-construction gap directly affects the headline dichotomy for distinct indices.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops quantitative bounds relating the thickness (normal injectivity radius) of higher-dimensional spherical links to their Milnor invariants, defined via Massey products. The main result, Theorem A, gives upper bounds for invariants with distinct indices: polynomial in τ^{-(m+1)(d-1)} when at least one component is 1-dimensional, a separate linking-number bound for d=2, and an exponential bound exp(C τ^{-m}) otherwise, with all rates claimed to be asymptotically sharp. Theorem B extends these bounds to repeated indices, with a polynomial bound τ^{-2(m+1)(d-1)} in the codimension-2/1-dimensional case and an exponential bound otherwise. Theorem C applies the machinery to prove that certain Freedman–Krushkal 2-complexes have exponentially small thickness in any embedding into R^4. The paper also develops a detailed homotopy-period and Massey-product framework, including integral formulas and a comparison with Koschorke's invariants.","tokens_in":42331,"tokens_out":8672,"duration_ms":94134,"significance":"If the results hold, this is a substantial contribution: it supplies the first thickness–Milnor-invariant bounds for higher-dimensional links, establishes a polynomial/exponential dichotomy, and resolves a question from Freedman–Krushkal. The proof architecture is coherent and contains useful tools, including explicit coisoperimetric inequalities (Lemmas 6.6, 6.8, 6.9, 6.11), integral formulas for Massey products (Theorem 4.3), and a systematic treatment of homotopy periods. However, several load-bearing assertions are currently not proved, most notably the thickness-scale doubling claims in Propositions 5.6–5.7 and the quantitative embedding of the building block B in the exponential construction of §2.2. These gaps directly affect the main theorems as stated.","major_comments":[{"comment":"Both statements end with the end-of-proof box and contain no proof. Proposition 5.6 asserts the existence of a τ-thick double for p < m/2, and Proposition 5.7 asserts a τ^2-thick double with zero linking number for p=1, m=3. These are used in §6.2 in the proof of Theorem 6.1(d) to replace τ by τ^2, yielding the exponent in Theorem B(i), and Theorem C inherits this through §7.2. The paper must supply proofs or a different argument; without them, Theorem B(i) and the application as written are unsupported.","section":"§5.5, Propositions 5.6 and 5.7"},{"comment":"The sharpness of the exponential regime in Theorem A(iii) depends on embedding the block B = T1 \\ T' with quantitative control on thickness and enclosing radius. The text only says, for the general case, 'as in the special case, we can embed B as a cobordism between two isometric copies of ∂T1', and for the special case it refers to an isotopy that 'unwraps' the inner torus. No coordinate model, no bound on the enclosing ball, and no reach estimate are provided. If the smallest ball containing such a B-embedding has radius ≫ n^{-1}, then the constructed link's thickness is not ε n^{-1} and the claimed exp(Θ(τ^{-m})) rate in Theorem A(iii) does not follow.","section":"§2.2, Theorem 2.3"},{"comment":"The proof constructs maps f' and f'' and then 'create[s] our map g by connecting the punctured f' and f'' via a movie of this isotopy in S^{p+2}.' It is not shown that the resulting g is an embedding, nor is any thickness control supplied for it. The lemma is used in Lemma 5.2 and Theorem 2.3, so the gap propagates to those arguments. At minimum, the construction needs a precise statement of how injectivity and normal-injectivity bounds are preserved.","section":"§2.1, Lemma 2.2"},{"comment":"The proof of the double's rational triviality says that one can 'modify g on a small ball' by taking a connected sum with an embedding representing the inverse value, 'since such an embedding exists by Lemma 2.2.' Given that Lemma 2.2 is itself only sketched, this step is not established. Since Theorem 5.1 relies on Lemma 5.2 to reduce repeated-index Massey products to the distinct-index case, the foundational treatment of §5 is incomplete as written.","section":"§5.1, Lemma 5.2"},{"comment":"The proof invokes Theorem B(i), whose proof depends on the unproved Propositions 5.6–5.7. The link in question has one S^1 and q+1 S^2 components with distinct indices, so Theorem 6.1(c)/A(i) would suffice and would give the stronger exponent 5(q+1) instead of 10(q+1). The text should either prove the doubling propositions or switch to the distinct-index bound. As written, the application is contingent on the missing doubling estimates.","section":"§7.2, proof of Theorem C"}],"minor_comments":[{"comment":"The displayed equality reads '¯µ(1,...,d)=2nm'; this should be 2^{n^m} to match the surrounding text and avoid confusion with the product 2nm.","section":"Theorem 2.3 statement"},{"comment":"Typo: 'In either cae' should be 'In either case'.","section":"§5.1, Lemma 5.2"},{"comment":"The notation 'thickness 4 ∼ n−1' uses '∼' with an unexplained subscript; please clarify the intended meaning and ensure the constants are tracked consistently.","section":"§2.2"},{"comment":"In the derivation of the exponent (2m−5)(d−2), the text says 'two of the p_i will always be missing from the sum in the exponent' without spelling out which two, and the displayed intermediate exponent τ^{m+2d−3} appears before the final τ exponent is obtained. A short explanatory sentence would improve readability.","section":"§6.2, Proof of Theorem 6.1(e)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is ambitious and likely correct in broad outline, but the current version contains several load-bearing gaps that are appropriately described as missing proofs rather than mere presentation issues. I do not see a scope mismatch, and the gaps appear fixable, so major revision seems the right decision. The authors should be asked to provide proofs for Propositions 5.6 and 5.7, a quantitative embedding of the block B with reach control, and a complete verification that Lemma 2.2 produces an embedding of the stated thickness. Alternatively, they may weaken the claims accordingly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should read this paper, but don't take the 'asymptotically sharp' in Theorem A(iii) on faith. The distinct-index polynomial bounds and the upper-bound machinery are genuine advances; the exponential sharpness construction is not fully proved.\n\nWhat is actually new: this is the first thickness (ropelength) bound for Milnor invariants of higher-dimensional spherical links. The authors build a serious framework: Massey products and homotopy periods (Sections 3–5), coisoperimetric inequalities (Section 6), and explicit thick links with large invariants (Section 2). Their Theorem A(i)–(ii) recovers and generalizes [42] and the linking-number bound, and Theorem C resolves the Freedman–Krushkal question via the proved Proposition 5.8. The upper bounds for distinct indices look solid: they follow from norm estimates on the integrands and the coisoperimetric lemmas, which are mostly proved carefully.\n\nNow the soft spots, in order of size. The lower-bound construction for the exponential regime (Theorem 2.3) rests on an informal embedding of the building block B in a ball, with no coordinate model and no control on the reach. If the ball radius needed for B is much larger than n^{-1}, the thickness of the stacked link is correspondingly smaller, and the invariant is not exp(Theta(tau^{-m})). So the claimed optimality of exp(C tau^{-m}) is currently unproved. This is not a technicality; it is the sharpness half of the headline dichotomy.\n\nSecond, Propositions 5.6 and 5.7 are stated with end-of-proof boxes but no proof. They both concern doubling components and explicitly carry the m=3 case of Theorem B(i). Theorem B(i) for m>=4 is covered by the cited, but not fully derived, integer coisoperimetric estimate in Proposition 5.8. The reader's report is slightly unfair there: Theorem C does not inherit the unproved 5.6/5.7 gap; it uses 5.8.\n\nThird, the exponent in Theorem 2.3 appears as '2nm', which must be 2^{n^m}; as typeset it is ambiguous and should be fixed.\n\nNone of this breaks the upper-bound story, and the central result is the polynomial/exponential dichotomy as an upper bound, not its sharpness. But a referee should ask for a complete proof of the exponential construction and for proofs of 5.6/5.7 before accepting the full claims. The paper otherwise shows honest engagement with the literature and a clear proof architecture.\n\nVerdict: send to peer review, with major comments. I would bring it to reading group.","headline":"First thickness bounds for higher-dimensional Milnor invariants with a real upper-bound dichotomy, but the exponential sharpness construction is not yet proved.","tokens_in":42999,"tokens_out":5739,"would_cite":true,"duration_ms":61978,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["57K45","55P62","57N35"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves sharp bounds: higher-dimensional thick links have Milnor invariants at most polynomial in τ^{-1} when a component is a circle, and at most exponential otherwise, with both rates attained.","keywords":["Milnor invariants","link thickness","ropelength","Massey products","higher-dimensional links","coisoperimetric inequalities","Brunnian links","homotopy periods"],"falsifier":"Construct a τ-thick circle in R^3 and measure the largest thickness of a parallel circle that is unlinked with the original. If that thickness is much smaller than a constant times τ², Proposition 5.7 is false and the repeated-index polynomial bound of Theorem B(i) does not follow.","tokens_in":41739,"feed_emoji":"🔗","tokens_out":7019,"duration_ms":78354,"temperature":0.7,"pith_summary":"The paper establishes the first quantitative bounds connecting the thickness of a higher-dimensional spherical link to the size of its Milnor invariants, the algebraic numbers that detect higher-order linking. For a link of d spheres in S^m satisfying the relevant dimension condition, the main theorem says the top Milnor invariant grows at most like a constant times τ^{-(m+1)(d-1)} when at least one component is one-dimensional, and at most like exp(C τ^{-m}) otherwise; the two-component case is exactly the linking number, bounded by C τ^{-(m+1)}. The estimates are asymptotically sharp: explicit thick links are constructed that realize each growth rate. As an application, the paper answers a previously open question about a family of 2-complexes in R^4 by showing that any embedding of them has exponentially small thickness.","feed_headline":"Thick links: Milnor invariants grow polynomially or exponentially","feed_subtitle":"First optimal thickness bounds for higher-dimensional spherical links, with explicit examples reaching both rates.","key_machinery":"The proof has three connected engines. Milnor invariants are computed as coefficients of Massey products of Alexander-dual cohomology classes of the link components, turning an invariant into an integral of products of differential forms. These forms are built inductively as primitives, and coisoperimetric inequalities bound the sup-norm of a primitive by the norm of the form times a constant that depends on thickness, volume, and dimension. On a τ-thick submanifold these constants are polynomial or exponential in τ^{-1}, and substituting them into the integral formula yields the upper bounds. Sharpness comes from explicit embeddings whose last component represents a large multiple of an ite","core_discovery":"The central claim is that thickness controls Milnor invariants of higher-dimensional spherical links in a precise, dimension-dependent way. For a d-component link with trivial Milnor invariants on proper subsequences, the top Milnor invariant satisfies three regimes: if some component is S^1, |μ̄| ≤ C τ^{-(m+1)(d-1)}; if d=2, |μ̄| is the linking number and is ≤ C τ^{-(m+1)}; otherwise |μ̄| ≤ exp(C τ^{-m}). The paper constructs links with thickness ε n^{-1} and Milnor invariant n^{(m+1)(d-1)} in the first regime, and links with Milnor invariant 2 n^m in the exponential regime, proving optimality for every combination of dimensions. It also proves a companion theorem for Milnor invariants with","pith_inferences":["The repeated-index bounds depend on two doubling claims that are stated without proof; a direct computation of the optimal thickness of an unlinked parallel copy of a thick circle in R^3 would either confirm or break the τ² scale that drives Theorem B(i).","The exponential sharpness examples suggest that thickness alone is a weak measure of embedding complexity when components have high codimension; combining thickness with a bounded local bilipschitz constant may be the more natural quantitative setting for further results.","The coisoperimetric-integral strategy may extend to other invariants defined by iterated integrals or Massey products, such as higher-order linking invariants or finite-type invariants of higher-dimensional links."],"forward_implications":["When at least one component is one-dimensional, the d-component Milnor invariant is O(τ^{-(m+1)(d-1)}), recovering and extending the classical ropelength-based bound.","When all components have codimension at least 3, the upper bound jumps to O(exp(C τ^{-m})); the paper's examples with invariant 2 n^m at thickness ε n^{-1} show this exponential rate is the true one.","For two components, the invariant is exactly the linking number and obeys |Lk| ≤ C τ^{-(m+1)}, an estimate that serves as a building block for the higher-order cases.","Adding a local bilipschitz bound on the embedding restores polynomial growth, so the exponential regime is specific to thickness alone rather than to all geometric complexity measures.","The complex-embedding question is settled: the constructed 2-complexes in R^4 are exponentially thin, with thickness at most c^{-l} for some constant c > 1 depending on the linking order."],"supporting_citations":[{"why":"Supplies the classical m=3 bound for Milnor invariants in terms of ropelength that Theorem A(i) generalizes to higher dimensions.","marker":"[42]"},{"why":"Provides the two-component linking-number bound and poses the question about exponentially thin 2-complexes that Theorem C answers.","marker":"[16]"},{"why":"Gives the Massey-product definition of classical Milnor invariants, which the paper extends verbatim to higher-dimensional spherical links.","marker":"[60]"},{"why":"Introduces Milnor invariants and the Milnor-group computation used to verify the sharpness examples in the polynomial regime.","marker":"[53]"},{"why":"Supplies the theory of homotopy periods and minimal models used to evaluate Milnor invariants as homotopy periods of the last component.","marker":"[67]"},{"why":"Provides the Andrews–Arkowitz formula pairing indecomposables of the minimal model with Whitehead products, used to compute the pairing in Proposition 3.5.","marker":"[2]"},{"why":"Supplies the second quantitative Poincaré lemma, a key coisoperimetric tool used to bound primitives of exact forms on thick submanifolds.","marker":"[48]"},{"why":"Establishes the lower-central-series comparison used to identify homotopy periods of link complements with invariants of the fundamental group.","marker":"[66]"}],"fun_headline_variants":["Thickness yields optimal Milnor bounds for spherical links","Polynomial vs exponential: thickness controls link invariants","Optimal thickness bounds on Milnor invariants in all dimensions","Spherical links: thickness dictates Milnor growth rate","Milnor invariants: thickness forces polynomial or exponential growth"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The sharp polynomial exponents for repeated-index Milnor invariants are carried by two claims about doubling a thick embedded sphere—Propositions 5.6 and 5.7—which are stated with end-of-proof boxes but no proof; if the true thickness of a parallel unlinked copy is smaller than claimed, the exponents in Theorem B(i) and the constant in Theorem C degrade.","fun_headline_variants_meta":{"raw":{"variants":["Thickness yields optimal Milnor bounds for spherical links","Polynomial vs exponential: thickness controls link invariants","Optimal thickness bounds on Milnor invariants in all dimensions","Spherical links: thickness dictates Milnor growth rate","Milnor invariants: thickness forces polynomial or exponential growth"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000206,"raw_usage":{"total_tokens":1196,"prompt_tokens":672,"completion_tokens":524,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":416,"completion_tokens_details":{"reasoning_tokens":455}},"tokens_in":416,"tokens_out":524,"duration_ms":6037,"temperature":1.0,"reasoning_tokens":455,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T11:21:43.784820+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Construct a τ-thick circle in R^3 and measure the largest thickness of a parallel circle that is unlinked with the original. If that thickness is much smaller than a constant times τ², Proposition 5.7 is false and the repeated-index polynomial bound of Theorem B(i) does not follow.","supporting_citations":[{"cited_title":"Komendarczyk and A","cited_arxiv_id":null,"evidence_quote":"Supplies the classical m=3 bound for Milnor invariants in terms of ropelength that Theorem A(i) generalizes to higher dimensions."},{"cited_title":"Freedman and V","cited_arxiv_id":null,"evidence_quote":"Provides the two-component linking-number bound and poses the question about exponentially thin 2-complexes that Theorem C answers."},{"cited_title":"Porter, Milnor’s ¯µ-invariants and Massey products , Trans","cited_arxiv_id":null,"evidence_quote":"Gives the Massey-product definition of classical Milnor invariants, which the paper extends verbatim to higher-dimensional spherical links."},{"cited_title":"Milnor, Link groups, Ann","cited_arxiv_id":null,"evidence_quote":"Introduces Milnor invariants and the Milnor-group computation used to verify the sharpness examples in the polynomial regime."},{"cited_title":"Sullivan, Infinitesimal computations in topology , Inst","cited_arxiv_id":null,"evidence_quote":"Supplies the theory of homotopy periods and minimal models used to evaluate Milnor invariants as homotopy periods of the last component."},{"cited_title":"Andrews and M","cited_arxiv_id":null,"evidence_quote":"Provides the Andrews–Arkowitz formula pairing indecomposables of the minimal model with Whitehead products, used to compute the pairing in Proposition 3.5."},{"cited_title":"Manin, Plato’s cave and differential forms , Geom","cited_arxiv_id":null,"evidence_quote":"Supplies the second quantitative Poincaré lemma, a key coisoperimetric tool used to bound primitives of exact forms on thick submanifolds."},{"cited_title":"Stallings, Homology and central series of groups , J","cited_arxiv_id":null,"evidence_quote":"Establishes the lower-central-series comparison used to identify homotopy periods of link complements with invariants of the fundamental group."}],"review_version":1}