{"id":"d7d4e8e4-a509-40ab-9887-06eec783ffb5","arxiv_id":"1908.01677","paper_version":4,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Bounding low-degree Betti numbers of all intersections of a set family bounds its Radon number, giving an optimal surface fractional Helly theorem for b=0.","lead":"This paper proves that families of sets in Euclidean spaces, manifolds, or surfaces have bounded Radon numbers whenever the low-degree Betti numbers of their intersections are bounded. The result yields fractional Helly theorems and, for open sets on a surface, settles a conjecture of Holmsen, Kim, and Lee.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Surface conjecture resolution hinges on the unproved companion inequality Theorem 17; the main Euclidean Radon bound is internally supported.","rationale":"I agree with the reader's weakest-assumption analysis. The main Euclidean theorem and the surface Radon theorem are supported by proofs in this manuscript plus standard cited non-embeddability results. The only serious unproved ingredient is Theorem 17, imported from the author's companion paper with Kalai, and it is essential specifically for the optimal fractional Helly number and the conjectured (p,q)-theorem. This is a legitimate citation dependency, not an internal inconsistency, so it does not by itself overturn the ACCEPT verdict; the reader already flagged it as a caveat. The abstract typo (b=1 versus b=0) is cosmetic and does not affect the mathematics. If a future audit of [KP19] found the inequality false or inapplicable, the conjecture resolution would fail, which is why the concrete test is to verify that theorem directly.","tokens_in":12216,"tokens_out":31543,"duration_ms":329669,"concrete_test":"Independently verify [KP19, Theorem 4]: confirm that it applies to finite families of open sets in a compact surface with only TC_1 ≤ b (bounded number of connected components of all intersections), and recover the exact linear inequality and k-range used in Theorem 17; then recompute the constant t in Proposition 16 using those c1, c2. If the inequality is false, or if it requires bounding beta_1 rather than only beta_0, Theorem 4 and the HKL-conjecture resolution do not follow.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central Radon bound (Theorem 1) is backed by Proposition 13, whose proof is in this manuscript, together with the cited non-embeddability of the k-skeleton of the (2k+2)-simplex in R^{2k}; I do not see an internal gap in that argument. The genuinely load-bearing soft spot is the optimal surface result. Theorem 4 and the resolution of the Holmsen-Kim-Lee conjecture are obtained by bootstrapping Proposition 16 from Theorem 17, a reformulation of [KP19, Theorem 4] that is not proved here. If that inequality — f_{k+1}=0 implies f_k ≤ c1 f_{k-1} + c2 for open sets in a surface with TC_1 ≤ b — is not available with the stated k-ranges (k(0) ≥ 2, k(b) ≥ 2b+3), the bootstrapping from k0 = 3 or k0 = 2b+4 fails and the optimal fractional Helly bound (and hence the (p,q)-theorem for b=0) does not follow. The manuscript even labels the reformulation weaker, so the exact hypotheses and constants cannot be audited from this text. This is a dependency rather than a demonstrated internal error, but it is load-bearing for the paper's headline applications.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper proves general topological bounds on Radon numbers. The main result (Theorem 1) states that if a finite family F of subsets of R^d has all reduced Betti numbers beta_i(∩G; Z2) for 0 ≤ i < ceil(d/2) bounded by b, then the Radon number r(F) is bounded by a function of b and d. Analogous statements are given for simplicial complexes, smooth manifolds, and surfaces (Theorem 2), where on a surface it suffices to bound the number of connected components of all intersections. Via the Holmsen–Lee theorem, these Radon bounds imply fractional Helly theorems (Theorem 3) and hence weak epsilon-nets and (p,q)-theorems. In Section 4, for open subsets of a surface with TC_1 ≤ b, the fractional Helly number is improved to 3 for b=0 and to 2b+4 for b≥1, with the b=0 case resolving a conjecture of Holmsen–Kim–Lee. The proof is built on a constrained-chain-map construction (Proposition 13), a Ramsey-type combinatorial lemma (Proposition 15), and known non-embeddability results.","tokens_in":12447,"tokens_out":16660,"duration_ms":135616,"significance":"The main Euclidean bound is a substantial and credible advance: it strengthens the earlier Helly-number bounds of [GPP+17] to Radon numbers, and the proof cleanly separates the topological non-embeddability statement from the combinatorial Ramsey argument. Proposition 13 is proved in detail within the manuscript, and the application to Theorem 1 uses only the cited non-embeddability of the k-skeleton of the (2k+2)-simplex. The optimal fractional Helly number 3 for b=0 on surfaces would be a sharp conjecture resolution. However, the surface-optimality part (Theorem 4 and the HKL conjecture) relies on Theorem 17, which is only cited as a weaker reformulation of [KP19, Theorem 4] and is not proved or fully specified in this manuscript. This dependency is the main weakness.","major_comments":[{"comment":"The proof of Proposition 16, and therefore Theorems 4 and 6, depends entirely on Theorem 17, which is stated as a reformulation of [KP19, Theorem 4] and explicitly acknowledged to be weaker than that result. The manuscript does not prove Theorem 17, nor does it specify the constants c1 and c2, the exact ranges of k and b, or how the reformulation follows for open subsets of a surface with TC_1(A) ≤ b. Because the bootstrapping in Proposition 16 starts from k0 = 3 (b=0) or k0 = 2b+4 (b≥1), the claimed optimal fractional Helly number and the resolution of the Holmsen–Kim–Lee conjecture cannot be audited from this text. Please add a self-contained proof of Theorem 17 (or of the weaker reformulation actually used), or state Theorem 17 with complete hypotheses and a derivation from [KP19], so that the k-range and constants can be verified.","section":"Section 4, Theorem 17"}],"minor_comments":[{"comment":"The sentence 'for b=1 we get that the fractional Helly number is at most three' is inconsistent with Theorem 4, which gives k=3 for b=0 and k=2b+4 for b≥1; this should be corrected to b=0.","section":"Abstract"},{"comment":"The phrase 'Let the cardinality of F be large enough' appears to be a typo for 'Let the cardinality of P be large enough'; the role of P in the induction hypothesis should be stated explicitly, and the definition of rK(b) in terms of the chosen s and the induction constants should be spelled out.","section":"Section 3.4, proof of Proposition 13"},{"comment":"The statement that TC_1(F) ≤ b bounds 'the number of connected components' should say that it bounds the reduced Betti number beta_0, i.e., the number of connected components is at most b+1; the current wording has an off-by-one discrepancy.","section":"Abstract and Section 2"},{"comment":"In the inequality verification, the expression 'tk+1' should be written as 't^{k+1}' to denote exponentiation; as typeset, it could be misread as a product.","section":"Section 4, proof of Proposition 16"}],"recommendation":"major_revision","confidential_remarks":"The main concern is the reliance on the companion paper [KP19] for the headline surface result. If the companion paper is under review or not yet accepted, the editor should consider whether the dependency is acceptable. The main Euclidean theorem is solid and would support publication even if the optimal surface result is weakened to a conditional statement."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper proves a genuine new theorem: bounded mid-level Betti numbers imply bounded Radon numbers for finite families in R^d (Theorem 1). This is the topological Radon analogue of the Helly-number bound from [GPP+17], and the proof is mostly self-contained. Proposition 13, the technical heart, is proved by a clean induction with a Ramsey-type lemma (Proposition 15); the reduction to Radon via constrained chain maps (Proposition 14) is short and correct. The applications to fractional Helly, weak eps-nets, and (p,q) theorems follow from Holmsen-Lee as stated. The surface version (Theorem 2) also works, using a non-almost-embeddable graph from [GMP+17].\n\nThe genuine soft spot is Theorem 4, the optimal fractional Helly bound for open sets on surfaces: fractional Helly number at most three when b=0, settling the Holmsen-Kim-Lee conjecture. This relies on Theorem 17, restated from the companion paper Kalai-Patáková [KP19]. That inequality (fk+1=0 implies fk ≤ c1 f_{k-1} + c2) is not proved here, and the manuscript notes the reformulation is slightly weaker, so the exact hypotheses cannot be audited from this text. This is a load-bearing dependency for the paper's headline applications. It is not an internal error, but a referee should verify [KP19] carefully. The main Euclidean Radon bound is independent of this.\n\nMinor issues: the abstract says b=1 where the body correctly says b=0, and in the proof of Proposition 13 the phrase 'Let the cardinality of F be large enough' should presumably be 'P'. Both are typos, not substantive.\n\nI agree with the reader's take: the central theorem is well-supported, the writing is clear, and the use of the companion paper is honest. The paper deserves a serious referee. If I were handling it, I would send it out, asking the referee to check the [KP19] dependence for Theorem 4 and the k-ranges in Proposition 16.","headline":"Solid new Radon bound via Betti numbers; the optimal surface result leans on a companion theorem that a referee must check.","tokens_in":12999,"tokens_out":2820,"would_cite":true,"duration_ms":25615,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["52A35","05D10","55N10","57N05"],"pacs":[],"model":"deepseek-v4-flash","headline":"Bounded Betti numbers force bounded Radon numbers in Euclidean space and on surfaces.","keywords":["Radon numbers","Betti numbers","topological complexity","fractional Helly theorem","(p,q)-theorem","weak epsilon-nets","surfaces","homological almost-embeddings"],"falsifier":"Take a finite family of open sets on the $2$-sphere whose every intersection is either empty or path-connected ($TC_1=0$), with a constant fraction of intersecting triples but where no point belongs to more than $o(n)$ sets; Theorem 4's $b=0$ case forbids this, so such a construction would refute it. Likewise, an explicit family in $\\mathbb{R}^d$ with $TC_{\\lceil d/2\\rceil}\\le b$ but with Radon number exceeding any prescribed bound would refute Theorem 1.","tokens_in":11999,"feed_emoji":"📐","tokens_out":8857,"duration_ms":76584,"temperature":0.7,"pith_summary":"This paper proves that a family of sets in Euclidean space with bounded homological complexity—meaning the first $\\lceil d/2\\rceil$ reduced Betti numbers of all intersections stay below a fixed constant—must have a bounded Radon number. The same kind of statement holds on surfaces when only the number of connected components of intersections is bounded. These Radon bounds combine with known reductions to yield fractional Helly theorems, weak $\\varepsilon$-nets, and $(p,q)$-theorems. For open sets on a surface, the fractional Helly number is at most $2b+4$ when intersection components are bounded by $b$, and at most $3$ for $b=0$, which is optimal and settles a conjecture about a $(p,q)$-theorem for open subsets of a surface.","feed_headline":"Bounded Betti numbers force bounded Radon numbers","feed_subtitle":"A new proof gives Radon and Helly bounds in Euclidean space and on surfaces, settling a surface conjecture.","key_machinery":"The central mechanism is the constrained chain map: a nontrivial chain map from the simplicial chains of a complex $K$ to singular chains of the ambient space whose support over each simplex lies inside the relative convex hull of a prescribed set of points, together with a map $\\Phi$ that preserves intersections and emptiness. Proposition 13 shows that a family with bounded topological complexity admits such a constrained map from any complex $K$ of dimension at most that level, via an induction that uses a Ramsey-type proposition to homogenize boundary homology classes. The Radon bounds then follow by combining this map with the non-embeddability of the $k$-skeleton of the $(2k+2)$-simplex in $\\mathbb{R}^{2k}$ (for Euclidean space) or with the nonexistence of almost-embeddings of suitable graphs in a surface. For the sharp surface bound, an additional ingredient (Theorem 17) supplies a linear inequality relating the face numbers of nerves of open sets, which drives the bootstrapping that lowers the fractional Helly number.","core_discovery":"The central claim is Theorem 1: for each $b$ and $d$ there is a number $r(b,d)$ such that any finite family $\\mathcal{F}$ in $\\mathbb{R}^d$ with $TC_{\\lceil d/2\\rceil}(\\mathcal{F}) \\le b$ satisfies $r(\\mathcal{F}) \\le r(b,d)$. The paper also proves the surface analogue (Theorem 2): if all intersections of subfamilies have at most $b$ connected components, the Radon number is bounded by a function of $b$ and the surface alone. For open sets on a surface, Theorem 4 improves the fractional Helly number from the general bound to $2b+4$, and to $3$ when $b=0$; this is optimal and yields the conjectured $(p,q)$-theorem for such families.","pith_inferences":["The companion inequality used for surfaces is quoted rather than proved here; if it generalizes, the same bootstrapping might give fractional Helly numbers for open sets in higher-dimensional manifolds, where the paper leaves bounded Radon numbers open.","The author conjectures that the surface fractional Helly number is $3$ for every $b$; the linear bound $2b+4$ in Theorem 4 may be an artifact of the proof rather than the true threshold.","The constrained chain-map method is formulated for any ambient space with a non-embeddability result, so it may yield Radon bounds beyond the listed spaces (e.g., for $\\mathbb{Z}_2$-acyclic spaces or spaces with a fixed topological type), though no such extension is claimed in the paper."],"forward_implications":["Every family in $\\mathbb{R}^d$ whose first $\\lceil d/2\\rceil$ intersection Betti numbers are bounded by $b$ has Radon number bounded in terms of $b$ and $d$; consequently its Helly, Tverberg, colorful Helly, and fractional Helly numbers are also bounded.","On any compact surface, bounding the number of connected components of all intersections bounds the Radon number, with no control needed on higher Betti numbers.","For open sets on a surface, the fractional Helly number is at most $2b+4$ when intersection components are at most $b$, and at most $3$ for $b=0$; the latter is best possible.","The fractional Helly results deliver weak $\\varepsilon$-nets and a $(p,q)$-theorem for these families, including the previously conjectured $(p,q)$-theorem for open subsets of a surface."],"supporting_citations":[{"why":"Supplies the non-embeddability of the $k$-skeleton of $\\Delta_{2k+2}$ into $\\mathbb{R}^{2k}$ (Theorem 10) and the original constrained-chain-map technique.","marker":"[GPP+17]"},{"why":"Provides the linear nerve inequality for open sets on a surface (Theorem 17) that drives the bootstrapping for the sharp fractional Helly bound.","marker":"[KP19]"},{"why":"Gives the reduction from bounded Radon number to fractional Helly theorem and colorful Helly theorem, used to derive Theorem 3.","marker":"[HL19]"},{"why":"Yields weak $\\varepsilon$-nets and $(p,q)$-theorems from fractional Helly theorems, giving Theorems 5 and 6.","marker":"[AKMM02]"},{"why":"Supplies a finite graph that does not almost-embed into a given surface, used in the proof of Theorem 2.","marker":"[GMP+17]"},{"why":"Gives the supersaturation theorem for hypergraphs that converts many intersecting $k$-tuples into many intersecting $(k+1)$-tuples in Proposition 16.","marker":"[ES83]"},{"why":"Contains the conjecture about a $(p,q)$-theorem for open subsets of a surface, settled here for the case $b=0$.","marker":"[HKL19]"},{"why":"Is the classical Radon theorem whose far-reaching generalization motivates the paper's definition of Radon number.","marker":"[Rad21]"}],"fun_headline_variants":["Betti numbers bound Radon numbers in all dimensions","Topological proof bounds Radon numbers, settles conjecture","Radon bounded by Betti: new proof, surface conjecture solved","Homology controls Radon: sharp bounds on surfaces","Betti to Radon: universal bound, plus Helly on surfaces"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The optimal surface bound and the resolved conjecture depend on a linear inequality from a companion paper (Theorem 17) that is stated but not proved in this manuscript; if that inequality fails for open subsets of a surface with bounded first Betti number, the sharp fractional Helly bound and the conjecture resolution do not follow.","fun_headline_variants_meta":{"raw":{"variants":["Betti numbers bound Radon numbers in all dimensions","Topological proof bounds Radon numbers, settles conjecture","Radon bounded by Betti: new proof, surface conjecture solved","Homology controls Radon: sharp bounds on surfaces","Betti to Radon: universal bound, plus Helly on surfaces"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000624,"raw_usage":{"total_tokens":2938,"prompt_tokens":1044,"completion_tokens":1894,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":660,"completion_tokens_details":{"reasoning_tokens":1811}},"tokens_in":660,"tokens_out":1894,"duration_ms":13823,"temperature":1.0,"reasoning_tokens":1811,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:07:00.539952+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a finite family of open sets on the $2$-sphere whose every intersection is either empty or path-connected ($TC_1=0$), with a constant fraction of intersecting triples but where no point belongs to more than $o(n)$ sets; Theorem 4's $b=0$ case forbids this, so such a construction would refute it. Likewise, an explicit family in $\\mathbb{R}^d$ with $TC_{\\lceil d/2\\rceil}\\le b$ but with Radon number exceeding any prescribed bound would refute Theorem 1.","supporting_citations":[],"review_version":1}