{"id":"0d3941a0-69ca-4db9-803c-1020464abc16","arxiv_id":"2411.17192","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For Bollobás systems of d-tuples, the conjectured bound sum of inverse multinomials <= 1 is false; the paper proves an asymptotically tight upper bound for d=3 and tight uniform skew bounds.","lead":"This paper refutes a 2024 conjecture about Bollobás-type inequalities for collections of d-tuples of sets, and proves new upper bounds for the sum of inverse multinomial coefficients. It also gives tight bounds on the maximum size of uniform skew Bollobás systems of d-tuples, for both sets and vector spaces.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Empty-middle components in Theorem 1.6 are projected to a d=2 Bollobás system, but the projection need not preserve the Bollobás property (explicit two-triple counterexample), so the main upper bound lacks a valid proof as written.","rationale":"The reader correctly flags that empty components are mishandled, but the specific degeneration they name (A_i^(1)=A_i^(3)=∅) is not the core obstruction. The load-bearing gap is the asserted reduction: the paper deletes an empty middle component and claims the remaining pairs form a Bollobás system. This is false, as the two-triple example shows; the original triple system is Bollobás, yet the projected pair system is not. This directly affects Theorem 1.6, which is the paper's main quantitative result and the basis for saying the Hegedüs–Frankl conjecture is false and Example 2 is asymptotically extremal. The counterexample to the conjecture itself is sound, and Theorems 1.8, 1.9, and 1.13 may be correct, so the paper has merit and a conditional verdict is appropriate pending a corrected proof of Theorem 1.6. The reader's second concern about equation (1) in Theorem 1.13 is not load-bearing: the skew condition only needs intersections with p<q, where the general-position dimension preservation does hold, so the overstatement for p=q is harmless.","tokens_in":14,"tokens_out":14111,"duration_ms":458124,"concrete_test":"Verify the two-triple system F={({1},∅,{2}), ({3},{1,2},{4})} on [4]. Check that Definition 1.5 holds for both ordered pairs (1,2) and (2,1), and then check that the projected pair family {({1},{2}), ({3},{4})} fails Definition 1.1. If confirmed, the reduction step in Theorem 1.6 is invalid; a repaired proof must handle A_i^(2)=∅ without relying on Theorem 1.2.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Section 2, the proof of Theorem 1.6 states: 'For A_i^(2)=∅, the collection {(A_i^(1), A_i^(3)) | A_i^(2)=∅} forms a Bollobás system (of d=2).' This claim is false. Let i=({1},∅,{2}) and j=({3},{1,2},{4}) on [4]. This is a Bollobás triple system: for (i,j), A_i^(1)∩A_j^(2)={1}; for (j,i), A_j^(2)∩A_i^(3)={2}. However, the projected pairs ({1},{2}) and ({3},{4}) fail the d=2 Bollobás condition in both directions: A_i^(1)∩A_j^(3)=∅ and A_j^(1)∩A_i^(3)=∅. Thus Theorem 1.2 cannot be invoked for the empty-middle component, and the d=3 bound is unproved. The same defect occurs in the induction step for arbitrary d: deleting a component A_i^(k) can destroy witnesses that used that component, so the resulting (d-1)-tuple family is not necessarily Bollobás. The counterexample to Conjecture 1 remains valid, but the asymptotic optimality claim for Theorem 1.6 is not established by the written argument.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies Bollobás-type theorems for families of d-tuples of sets and of vector subspaces. It refutes a conjecture of Hegedűs and Frankl by constructing a Bollobás triple system whose inverse multinomial sum is floor(n/2)+1, and proves upper bounds for this sum for arbitrary d (Theorem 1.6), which is asymptotically tight for d=3. It also improves a skew Bollobás inequality (Theorem 1.8) and determines the maximum size of uniform skew Bollobás systems of d-tuples of sets and spaces (Theorems 1.9 and 1.13). The proofs use random permutations and exterior algebra.","tokens_in":12688,"tokens_out":30061,"duration_ms":238227,"significance":"If the results hold, the paper settles a natural conjecture in the negative and provides the first nontrivial upper bounds for the generalized Bollobás sum. The d=3 bound is asymptotically tight, and the exterior algebra proof of Theorem 1.13 is a clean linear independence argument. The probabilistic techniques are coherent, and the paper correctly acknowledges independent work by Tian and Wu.","major_comments":[{"comment":"The displayed estimate for a fixed k states that the sum over tuples with A_i^(k)=∅ is at most 1/(d-2) binom(n+d-3,d-3) + O(n^{d-4}). The subsequent 'Hence' line adds a single binom(n+d-3,d-3) term to the bound for tuples with all middle components nonempty. To cover all tuples with at least one empty middle component, the proof must sum the displayed estimates over all k in {2,...,d-1}; the factor (d-2) then cancels the 1/(d-2), yielding exactly the term shown. As written, the step is a non sequitur, though the repair is straightforward.","section":"Section 2, induction step of Theorem 1.6"}],"minor_comments":[{"comment":"The statement that the subfamily with A_i^(2)=∅ projects to a Bollobás pair system is true, but the proof should justify it: any witness for two such triples cannot involve the empty second component, so it survives projection.","section":"Section 2, d=3 case of Theorem 1.6"},{"comment":"The definition of E_i is ambiguous about whether the delimiters must appear in a fixed order. If a specific order is meant, the probability should include a 1/(d-1)! factor; if the union over all orders is meant, this should be stated. The current text is open to misinterpretation.","section":"Section 2, proof of Theorem 1.8"},{"comment":"The claim 'A ∩ (B′ ∪ C′) ≠ ∅' when A≠A′ is false in general (e.g., if A=∅ or A⊆A′). The example is still valid because in the exceptional case B∩C′≠∅ provides the witness; the proof should be corrected.","section":"Example 2"},{"comment":"There is a typo in the family notation: the closing parenthesis is missing in '{(A_i^(1), ..., A_i^(d) | i ∈ [m])}'.","section":"Theorem 1.6 statement"},{"comment":"There are several minor typos, e.g., 'boun d' in the abstract and '/greaterorequalslant' in the body; and the symbols '⋆' in Table 2 are only explained in the final remark.","section":"Abstract and text"}],"recommendation":"minor_revision","confidential_remarks":"The stress-test concern about the projection of empty-middle components appears to be based on a misreading: the counterexample uses a tuple with a nonempty middle component, which is not in the subfamily being projected. Within the subfamily where all tuples have the same component empty, the projection does preserve the Bollobás property. The only substantive issue I found is the terse induction step in Theorem 1.6, which needs an explicit summation over k."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. The counterexample to Hegedüs and Frankl's conjecture is real, and the uniform skew Bollobás bounds (Theorems 1.9 and 1.13) look correct. But the proof of the main upper bound, Theorem 1.6, has a load-bearing gap: the empty-middle-component case is not handled by the argument given.\n\nWhat's good. Example 2 is a clean construction: for d=3, the sum of inverse multinomials is about n/2, so the conjectured bound of 1 is wrong. Theorem 1.8's random-permutation argument is sound and genuinely sharpens the skew Bollobás inequality; the disjointness of the events is straightforward. The exterior algebra proof of Theorem 1.13 is a valid linear-independence argument and gives the tight uniform bound for spaces, which also implies Theorem 1.9 for sets.\n\nWhere it breaks. In the proof of Theorem 1.6, after handling tuples with nonempty middle component, the paper says that when A_i^(2)=∅, the projected pairs (A_i^(1), A_i^(3)) form a d=2 Bollobás system, so Theorem 1.2 applies. That is false. Take i=({1},∅,{2}) and j=({3},{1,2},{4}). The triple system is Bollobás: i∩j via {1}∩{1,2}, and j∩i via {1,2}∩{2}. But the projected pairs ({1},{2}) and ({3},{4}) fail in both directions. So the bound (n+3)/2 is not established. The same defective projection appears in the induction step for d>3 when deleting a component. This is not a minor technicality; it is the step that lets the proof reach the asserted bound.\n\nThe paper is honest about Tian and Wu's independent work, so the counterexample and Theorem 1.8 are not novel to this manuscript. What remains new is Theorem 1.6 (needs a repaired proof) and the uniform bounds.\n\nBottom line: the counterexample and the uniform bounds are worth a serious referee. The author should be asked to fix the empty-component argument, or clearly restrict Theorem 1.6 to tuples with all internal components nonempty. Until then, Theorem 1.6 should not be taken as proved.\n\nRecommendation: send to peer review, but with the expectation that the proof of Theorem 1.6 must be revised.","headline":"Nice counterexample and sound uniform bounds, but the proof of the main upper bound has a real gap in the empty-middle-component case.","tokens_in":13253,"tokens_out":4398,"would_cite":true,"duration_ms":36958,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05D05","05A10","05A20"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper refutes the Hegedüs–Frankl conjecture that d-tuple Bollobás sums stay at most 1, proves an asymptotically tight (n+3)/2 bound for triples, and gives multinomial bounds for skew systems of sets and spaces.","keywords":["Bollobás systems","skew Bollobás systems","d-tuples","multinomial coefficients","random permutations","exterior algebra","extremal set theory"],"falsifier":"An explicit Bollobás system of triples on a five-element ground set with inverse-multinomial sum exceeding 4 would refute the d=3 bound; exhaustive search for small n could settle it.","tokens_in":12197,"feed_emoji":"🧮","tokens_out":6530,"duration_ms":59533,"temperature":0.7,"pith_summary":"In 1965 Bollobás proved that for a family of disjoint set-pairs with the required cross-intersections, the sum of inverse binomial coefficients is at most 1. The paper extends this question to d-tuples of pairwise disjoint sets and shows that the natural multinomial generalization is false. It constructs a triples example whose inverse-multinomial sum is floor(n/2)+1, and proves that for triples the sum can never exceed (n+3)/2, so the construction is asymptotically tight. For general d, the paper gives a polynomial upper bound whose leading term is 1/(d-1) times the binomial coefficient (n+d-2 choose d-2). It also sharpens the known inequality for skew Bollobás systems and determines the exact maximum size of uniform skew systems on both sets and spaces.","feed_headline":"A 1965 bound breaks for triples: inverse sums reach n/2","feed_subtitle":"The classical Bollobás bound said such sums stay at 1; for triples the true maximum grows linearly and is nearly sharp.","key_machinery":"The load-bearing mechanism is a random-permutation encoding: take a uniformly random permutation of [n] together with d-1 extra delimiter elements, and require the components of a tuple to appear in the prescribed order with delimiters separating specified adjacent blocks. The probability of such an event is the reciprocal of a product of a multinomial coefficient and a binomial factor, and the Bollobás condition forces events belonging to different tuples to be disjoint, giving the sum bound. For skew systems the same encoding yields the weighted inequality of Theorem 1.8. For the space version, exterior algebra replaces permutations: each subspace is represented by a wedge product, and the skew cross-intersection condition makes certain linear functionals vanish in a triangular way, yielding the multinomial size bound.","core_discovery":"The central discovery is that the multinomial analogue of Bollobás's inequality fails for d-tuples, and the correct order of growth is polynomial in the ground-set size. The counterexample collects all disjoint triples of type (l, n-2l, l) for l=0,...,floor(n/2); each triple contributes an inverse multinomial coefficient, and the total is floor(n/2)+1. Theorem 1.6 proves for d=3 that no Bollobás system of triples on [n] can make the sum exceed (n+3)/2, so the construction is asymptotically extremal. For general d, the same random-permutation argument yields the bound 1/(d-1) times (n+d-2 choose d-2) plus O($n^{{d-3}}$). For skew systems, the paper proves a refined inequality with an extra binomial factor and shows that uniform skew systems of d-tuples, of sets or of subspaces, have size at most the corresponding multinomial coefficient.","pith_inferences":["The d=3 upper bound of (n+3)/2 suggests that extremal constructions concentrate on middle components of intermediate size; a testable conjecture is that the maximum is attained by types with all components as equal as possible.","The proof gap for tuples with empty first or last components might be closed by a limiting or convexity argument; if the gap cannot be closed, the bound could still be valid by a different route.","The exterior-algebra proof for uniform skew systems of spaces may extend to affine subspaces or matroids, paralleling the classical Lovász extension of Bollobás-type theorems.","For d≥4, sharpening the O(n^{d-3}) error term would require precisely counting tuples with empty middle components, which the current induction only estimates coarsely."],"forward_implications":["The Hegedüs–Frankl conjecture is false: for d=3 the inverse-multinomial sum can be as large as floor(n/2)+1, so the original bound of 1 does not extend to d-tuples.","For triples, the maximum sum is pinned between roughly n/2 and (n+3)/2, so the paper's construction is asymptotically extremal.","For d≥4, the leading term 1/(d-1) times (n+d-2 choose d-2) is the first general upper bound, and it remains open whether the true maximum is smaller.","For skew Bollobás systems, the weighted inverse sum is at most 1, which immediately implies the unweighted inverse sum is at most (n+d-1 choose d-1).","Uniform skew systems of d-tuples, whether of sets or of subspaces, have size at most the multinomial coefficient (a_1+...+a_d choose a_1,...,a_d), and this is tight by the explicit disjoint-tuples construction."],"supporting_citations":[{"why":"Supplies the original 1965 set-pair theorem, used directly in the proof for the case A_i^(2)=empty.","marker":"[3]"},{"why":"Introduces Bollobás systems and skew Bollobás systems of d-tuples, states the conjecture that Theorem 1.6 refutes, and provides the skew inequality that Theorem 1.8 sharpens.","marker":"[7]"},{"why":"Independently noticed the invalidity of the conjecture and of the earlier skew inequality; the paper compares its counterexample with theirs.","marker":"[20]"},{"why":"Supplies the probabilistic method with random permutations that this paper generalizes to d-tuples.","marker":"[23]"},{"why":"Provides the general-position lemma and the exterior-algebra conventions used in the proof of Theorem 1.13.","marker":"[2]"},{"why":"The Lovász matroid extension of Bollobás's theorem that Theorem 1.13 generalizes to d-tuples of spaces.","marker":"[12]"}],"fun_headline_variants":["Triples break Bollobás bound: max sum grows as n/2","Bollobás d-tuple bound refuted; linear for triples","For triples, Bollobás sum reaches (n+3)/2, almost tight","Bollobás d-tuple conjecture false; asymptotically tight bound","Uniform skew Bollobás systems bounded by multinomial coefficient"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The counting argument assumes that the permutation events for different tuples cannot overlap, and that this remains true even for tuples with empty first or last components.","fun_headline_variants_meta":{"raw":{"variants":["Triples break Bollobás bound: max sum grows as n/2","Bollobás d-tuple bound refuted; linear for triples","For triples, Bollobás sum reaches (n+3)/2, almost tight","Bollobás d-tuple conjecture false; asymptotically tight bound","Uniform skew Bollobás systems bounded by multinomial coefficient"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001323,"raw_usage":{"total_tokens":5417,"prompt_tokens":1010,"completion_tokens":4407,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":626,"completion_tokens_details":{"reasoning_tokens":4308}},"tokens_in":626,"tokens_out":4407,"duration_ms":30964,"temperature":1.0,"reasoning_tokens":4308,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T12:29:32.791127+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"An explicit Bollobás system of triples on a five-element ground set with inverse-multinomial sum exceeding 4 would refute the d=3 bound; exhaustive search for small n could settle it.","supporting_citations":[{"cited_title":"Bollob´ as, On generalized graphs, Acta Math","cited_arxiv_id":null,"evidence_quote":"Supplies the original 1965 set-pair theorem, used directly in the proof for the case A_i^(2)=empty."},{"cited_title":"Heged¨ us, P","cited_arxiv_id":null,"evidence_quote":"Introduces Bollobás systems and skew Bollobás systems of d-tuples, states the conjecture that Theorem 1.6 refutes, and provides the skew inequality that Theorem 1.8 sharpens."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Independently noticed the invalidity of the conjecture and of the earlier skew inequality; the paper compares its counterexample with theirs."},{"cited_title":"Some new Bollob\\'as-type inequalities","cited_arxiv_id":"2405.17639","evidence_quote":"Supplies the probabilistic method with random permutations that this paper generalizes to d-tuples."},{"cited_title":"Babai, P","cited_arxiv_id":null,"evidence_quote":"Provides the general-position lemma and the exterior-algebra conventions used in the proof of Theorem 1.13."},{"cited_title":"Lov´ asz, Flats in matroids and geometric graphs, Proc","cited_arxiv_id":null,"evidence_quote":"The Lovász matroid extension of Bollobás's theorem that Theorem 1.13 generalizes to d-tuples of spaces."}],"review_version":1}