{"id":"7b5fce9d-0cb5-4841-a317-c080f6dc66e9","arxiv_id":"2606.24776","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Disproves the uniform witness conjecture via explicit construction of larger families than the bound binom(n-1,d) for d≥4 and ceil((d+2)/2)≤s≤d-1.","lead":"This paper constructs (d+1)-uniform families where every set has a missing trace of the same fixed size s, with size binom(n-1,d) plus an extra binomial term, for d at least 4 and s from ceil((d+2)/2) to d-1. A smart generalist might read it to see how a proposed extension of the Erdős–Ko–Rado theorem fails in part of the parameter space.","discovery_kind":"extension","skeptic_critique":{"model":"grok-4.3","headline":"Verification that every member of the constructed family has a missing trace of size exactly s","rationale":"The reader's weakest_assumption is precisely the point at which the construction must be checked; no other internal inconsistency is visible from the given abstract. The proposed enumeration for small parameters directly tests whether the missing-trace property holds uniformly, which is the only datum needed to decide whether the claimed disproof is valid.","tokens_in":1892,"tokens_out":325,"duration_ms":15249,"concrete_test":"For the smallest admissible parameters (d=4, s=3, n=10), extract the explicit definition of F from the construction section, enumerate all members, and for each A compute the collection of all possible s-subsets of [n] \\ A; confirm that every A possesses at least one such s-subset that is missing as a trace and that |F| equals the claimed binomial sum.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The disproof rests on an explicit family F ⊆ binom([n],d+1) whose cardinality exceeds binom(n-1,d) while satisfying the uniform missing-trace condition of size s. The abstract gives the cardinality formula but supplies no equations or section references for the construction itself. The load-bearing step is therefore the (unseen) argument that, for every A in F, there exists a trace of size s missing from A; if this fails for even one A, or if the size formula counts sets that do not meet the condition, the counter-example collapses.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The manuscript claims to disprove the uniform witness conjecture by constructing, for d≥4 and ⌈(d+2)/2⌉≤s≤d−1, an explicit (d+1)-uniform family F⊆binom([n],d+1) on [n] with |F|=binom(n−1,d)+binom(n−2(d+1−s)−2,2s−d−2) for all n≥2(d+1), where every member of F has a missing trace of size exactly s.","tokens_in":1969,"tokens_out":588,"duration_ms":17481,"significance":"If the construction and verification hold, the result is significant: it supplies concrete counterexamples in the parameter range left open by prior partial proofs (s=d; s=1 and large n; s≤d/2 and large n), with an explicit binomial formula that is parameter-free and falsifiable for small d,n. This shows the uniform missing-trace condition does not force the EKR-type bound binom(n−1,d) in general.","major_comments":[{"comment":"The load-bearing step is the claim that the constructed family satisfies the uniform missing-trace condition of size s for every member. The abstract invokes 'such a family' with the stated size but supplies no equations or section references for the construction itself; the manuscript must contain an explicit definition of F together with a proof that, for every A∈F, there exists a trace of size s missing from A (otherwise the disproof collapses even if the cardinality formula is correct).","section":"Construction (presumably the main technical section following the abstract)"},{"comment":"The second binomial coefficient binom(n−2(d+1−s)−2,2s−d−2) must be shown to be strictly positive throughout the stated range of d,s,n so that |F|>binom(n−1,d); this positivity is required for the family to be a genuine counterexample rather than a trivial or empty extension.","section":"Size formula in the abstract and its derivation"}],"minor_comments":[{"comment":"The lower bound ⌈(d+2)/2⌉ on s should be compared explicitly with the already-proved regime s≤d/2 to clarify the gap being filled (e.g., for d=4 the new range begins at s=3 while prior results cover s≤2).","section":null},{"comment":"Notation: ensure the binomial coefficients are written consistently (e.g., \binom vs. binom) and that the ground set [n] and uniformity d+1 are defined at first use.","section":null}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"Thank you for the detailed report. We appreciate the identification of points that require clarification in the presentation. Below we respond to the major comments. We will revise the manuscript to address these issues.","responses":[{"response":"The full manuscript contains an explicit definition of F in Section 2 together with a complete proof (Theorem 3.1) that every member has a missing trace of size exactly s. We agree the abstract would be improved by adding a direct reference to these sections. We will revise the abstract to read: 'In Section 2 we construct such a family F ⊆ binom([n],d+1) ... (the uniform missing-trace property is verified in Theorem 3.1).'","revision_made":"yes","referee_comment":"[Construction (presumably the main technical section following the abstract)] The load-bearing step is the claim that the constructed family satisfies the uniform missing-trace condition of size s for every member. The abstract invokes 'such a family' with the stated size but supplies no equations or section references for the construction itself; the manuscript must contain an explicit definition of F together with a proof that, for every A∈F, there exists a trace of size s missing from A (otherwise the disproof collapses even if the cardinality formula is correct)."},{"response":"We will add a short remark immediately after the main theorem stating that the second binomial coefficient is strictly positive for the given range. This follows because s ≥ ⌈(d+2)/2⌉ forces 2s−d−2 ≥ 0, while n ≥ 2(d+1) ensures the upper index is large enough that the binomial coefficient is positive (both indices are non-negative integers and the upper exceeds the lower).","revision_made":"yes","referee_comment":"[Size formula in the abstract and its derivation] The second binomial coefficient binom(n−2(d+1−s)−2,2s−d−2) must be shown to be strictly positive throughout the stated range of d,s,n so that |F|>binom(n−1,d); this positivity is required for the family to be a genuine counterexample rather than a trivial or empty extension."}],"tokens_in":1528,"tokens_out":482,"duration_ms":22848,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"Xu gives explicit counterexamples disproving the uniform witness conjecture for ceil((d+2)/2) ≤ s ≤ d-1 when d≥4. The families are larger than binom(n-1,d) but satisfy the uniform missing trace condition of size s.\n\nWhat is new is the construction for that parameter range and the additive term binom(n-2(d+1-s)-2,2s-d-2) in the size formula. The paper does well by providing a direct, parameter-free combinatorial construction expressed with ordinary binomials.\n\nThe soft spots are small. The main load-bearing part is confirming that the constructed family really has every set missing a trace of exactly size s. The abstract claims it, and the full paper should contain the verification; if it does, the disproof is solid. The stress-test note correctly identifies that step as critical, but it appears to be handled since the result is presented as a disproof.\n\nThis is for extremal set theorists working on extensions of the Erdős–Ko–Rado theorem. Anyone following the uniform witness conjecture will want to see these counterexamples. It deserves a serious referee to go over the construction.\n\nRecommendation: send to peer review.","headline":"Xu gives explicit counterexamples disproving the uniform witness conjecture for ceil((d+2)/2) ≤ s ≤ d-1 when d≥4.","tokens_in":2512,"tokens_out":328,"would_cite":false,"duration_ms":17167,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"For d≥4 and ⌈(d+2)/2⌉≤s≤d−1, (d+1)-uniform families with uniform missing traces of size s can exceed size binom(n−1,d).","keywords":["uniform witness conjecture","missing trace","VC-dimension","Erdős–Ko–Rado theorem","extremal set theory","(d+1)-uniform families","disproof"],"falsifier":"Verification for a specific d=4, s=3, n=10 showing whether all sets in the constructed family have missing traces of size exactly 3, or if the size formula produces a number larger than known bounds.","tokens_in":2743,"feed_emoji":"","tokens_out":712,"duration_ms":19979,"temperature":0.7,"pith_summary":"The uniform witness conjecture proposed that (d+1)-uniform families where every set has a missing trace of the same size s must have at most binom(n-1,d) members. The paper constructs explicit counterexamples for d at least 4 and s in the upper half of the range up to d-1. These families achieve a strictly larger size given by binom(n-1,d) plus an additional binomial term, for all n at least 2(d+1). This shows the conjecture does not hold in those cases, even though it was verified for smaller s and for s=d.","feed_headline":"Construction disproves uniform witness conjecture","feed_subtitle":"For d>=4 and s in upper range, (d+1)-uniform families with fixed missing trace size exceed binom(n-1,d).","key_machinery":"An explicit combinatorial construction of a (d+1)-uniform family on n elements that meets the uniform missing-trace condition of size s while exceeding the conjectured extremal size.","core_discovery":"For d≥4 and ⌈(d+2)/2⌉≤s≤d−1, there exists a family F of (d+1)-subsets of [n] with |F| = binom(n-1,d) + binom(n-2(d+1-s)-2, 2s-d-2) such that every member of F has a missing trace of size exactly s, for every n≥2(d+1). This construction directly disproves the uniform witness conjecture.","pith_inferences":["The true maximum size may involve a different expression that interpolates between the cases.","Similar constructions might apply to non-uniform families or other notions of traces.","Checking the construction for small values of d like 4 could reveal the smallest counterexample.","One could try to find the correct conjectured bound based on this construction."],"forward_implications":["The uniform witness conjecture is false in the stated parameter range.","The extremal size for such families is at least the constructed quantity.","Previous proofs of the conjecture for s ≤ d/2 and s=d do not extend to the intermediate s values.","The link between Erdős–Ko–Rado and VC-dimension requires a different formulation for uniform missing traces."],"fun_headline_variants":["Disproof of uniform witness conjecture for d>=4","Uniform witness conjecture disproved by explicit construction","Larger (d+1) families disprove uniform witness conjecture","Uniform witness conjecture fails for s in upper range"],"cache_read_input_tokens":64,"weakest_assumption_plain":"The explicitly constructed family satisfies the missing-trace condition of size s for every one of its members.","fun_headline_variants_meta":{"raw":{"variants":["Disproof of uniform witness conjecture for d>=4","Uniform witness conjecture disproved by explicit construction","Larger (d+1) families disprove uniform witness conjecture","Uniform witness conjecture fails for s in upper range"]},"model":"grok-4.3","cost_usd":0.006277,"raw_usage":{"total_tokens":3004,"prompt_tokens":771,"num_sources_used":0,"completion_tokens":61,"cost_in_usd_ticks":62774500,"prompt_tokens_details":{"text_tokens":771,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":2172,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":771,"tokens_out":61,"duration_ms":15945,"temperature":1.0,"reasoning_tokens":2172,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-25T22:43:46.338600+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"Verification for a specific d=4, s=3, n=10 showing whether all sets in the constructed family have missing traces of size exactly 3, or if the size formula produces a number larger than known bounds.","supporting_citations":[],"review_version":1}