{"id":"446804c8-b358-42dc-a958-760820aef163","arxiv_id":"2507.10828","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A d-maximal subset of the n-letter Hamming cube that contains no unit ball has size at most roughly n^d, and related bounds show all finite d-maximal sets are bounded and finitely many up to isomorphism.","lead":"The paper proves that any collection of words over an n-letter alphabet that is maximally spread out, meaning its diameter cannot grow without breaking the set, is either small or contains a ball of radius one. These results settle a lifting-stability question from recent work and give the first general size bounds for such maximal sets.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 12's distance identity is false as stated; the pigeonhole step yields a stronger disjointness condition that repairs it, so the central theorem remains plausible but the proof needs this fix.","rationale":"I read the paper in good faith. Lemma 11's translation from set-sunflowers to word-sunflowers is sound: if two words share a coordinate value, that pair lies in the common C2-intersection, and p > n-1 forces a shared value; the reader's identified weak point is not where the proof breaks. The actual flaw I found is in Lemma 12's distance computation, a step the reader did not flag. It is a real false identity, but it is immediately repairable via the stronger disjointness condition that the same argument supplies. I checked the other main components - Proposition 8's averaging argument, the D_m count, the floating-coloring rounding, and the binary entropy argument - and found only the typos already noted by the reader. Therefore the theorem is believable, but Lemma 12 should be corrected before the paper is accepted.","tokens_in":11364,"tokens_out":31032,"duration_ms":384574,"concrete_test":"Formalize the proof of Lemma 12 and test the displayed distance identity under the paper's actual selection condition. A minimal counterexample is a single coordinate with w_i=0, a^j_i=1, z_i=2, with all other coordinates arranged so that the stated condition {i>m': a^j_i = z_i != w_i} subset [m'] holds. If the identity fails, rerun the argument with the stronger selection condition E_j ∩ {i>m': z_i != w_i} = empty; verify that the same pigeonhole bound yields such j and that the identity then holds. This settles whether Lemma 12 needs the stated correction.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 2, proof of Lemma 12: after selecting the sunflower a^1,...,a^p, the text chooses j with {i>m': a^j_i = z_i != w_i} contained in [m'] and then asserts dist(a^j,z) = dist(b_{<=m'}, z_{<=m'}) + dist(w_{>m'}, z_{>m'}) + dist(w_{>m'}, a^j_{>m'}) = dist(b,z) + k - m'. This identity is not implied by the stated condition. For a coordinate i>m' with w_i=0, a^j_i=1, z_i=2, the Hamming cost [a^j_i != z_i] is 1, whereas [z_i != w_i] + [a^j_i != w_i] = 2. Such a triple is compatible with the paper's condition because a^j_i != z_i, so the displayed equality fails exactly where the proof needs it: it is used to convert Ball(a^j, l) subset Ball(z,d) into the stronger Ball(b, l+k-m') subset Ball(z,d). This is the load-bearing step in Lemma 12, and hence in the sunflower-based bound on |S^(l)_k[t(m)]|. The fix is local: the same disjointness/pigeonhole argument that produced j actually proves the stronger statement E_j ∩ {i>m': z_i != w_i} = empty for some j, since the E_j are pairwise disjoint and z differs from w in at most d < p positions. Under this stronger condition the problematic coordinate case cannot occur, and the identity becomes correct. Thus the central theorem is not overturned, but the proof as written has a genuine gap in a central lemma.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies d-maximal subsets of the infinite Hamming cube [n]^∞. The main result, Theorem 1, states that a d-maximal set containing no 1-ball has size at most d^2(n+8n^{2/3})^d, which yields the dichotomy that every d-maximal set either has this bounded size or contains a nontrivial Hamming ball. The paper also proves a more general bound for the centers S^(ℓ), deduces finiteness of isomorphism types of d-maximal sets, gives a polynomial lower bound on the size of any d-maximal set, and provides improved upper and lower bounds for the binary alphabet. The proofs use regular templates, a generalized sunflower lemma, Kleitman's theorem, and Azuma's inequality.","tokens_in":11710,"tokens_out":38152,"duration_ms":423505,"significance":"If the main theorem is established, it resolves the lifting-stability question of Briggs, Feng and Wells and gives the first general upper bound for ball-free d-maximal sets. The finiteness of isomorphism types and the polynomial lower bound are also substantial additions. The argument is transparent, uses standard and independent tools, and contains no fitted parameters. The paper's main claim is plausible, but the proof of the central Lemma 12 contains a genuine gap that must be repaired before the main result can be considered proven.","major_comments":[{"comment":"The displayed distance identity after the pigeonhole step is not implied by the condition stated in the proof. For a coordinate i>m' with w_i=0, a^j_i=1, z_i=2, the condition that {i>m': a^j_i=z_i ≠ w_i} is contained in [m'] is satisfied, but [a^j_i ≠ z_i]=1 while [z_i ≠ w_i]+[a^j_i ≠ w_i]=2, so the equality fails. The same pigeonhole argument actually yields the stronger condition E_j ∩ {i>m': z_i ≠ w_i}=∅, because the sets E_j are pairwise disjoint and z differs from w in at most d<p positions. Under this stronger condition the identity becomes correct. This step is load-bearing: it is used to convert Ball(a^j,ℓ)⊂Ball(z,d) into Ball(b,ℓ+k−m′)⊂Ball(z,d), and hence it is essential for the sunflower-based bound on |S^(ℓ)_k[t(m)]|.","section":"§1, proof of Lemma 12"},{"comment":"The sentence 'Since C2(a1)=···=C2(ap)' is false as stated: the sunflower condition gives identical pairwise intersections and pairwise disjoint petals, not equality of the C2 sets. The intended inference is nevertheless valid, because the C2(a^j) form a sunflower: if (i,ℓ) belongs to two of them, it belongs to the common core and hence to all of them. This correction should be stated explicitly; as printed, the proof of the generalized sunflower lemma contains an incorrect assertion in a central lemma.","section":"§1, proof of Lemma 11"},{"comment":"The final contradiction with the definition of S^(ℓ) is not immediate as written. The definition forbids Ball(a^j,ℓ)⊂Ball(c,ℓ+1)⊂S, whereas the proof has established Ball(a^j,ℓ)⊂Ball(b,ℓ+k−m′)⊂S with possibly k−m′>1. For k−m′≥1, one can choose c on a shortest path from b to a^j with dist(b,c)=k−m′−1; then Ball(c,ℓ+1)⊂Ball(b,ℓ+k−m′)⊂S and Ball(a^j,ℓ)⊂Ball(c,ℓ+1), yielding the required contradiction. Without this additional argument, the final step of Lemma 12 does not follow from the definition of S^(ℓ).","section":"§1, proof of Lemma 12"}],"minor_comments":[{"comment":"The stopping condition 'as long as the number of floating variables is less than f + m' should read 'greater than f + m'; otherwise the process would not run when the number of floating variables is initially large.","section":"§3, floating-coloring argument"},{"comment":"The sentence 'for otherwise n + 8n^{2/3} ≤ 2n' is false for n=100; the correct inequality for n<512 is n+8n^{2/3} ≥ 2n. The reduction to Corollary 14 should be corrected, and in fact the derivation in §2 yields the stated bound for all n≥2, so the assumption n≥100 is unnecessary.","section":"§2, first paragraph"},{"comment":"The template t_i=1 used in Lemma 17 is regular with respect to w only after a coordinate-wise permutation making w_i=2 for every i. This WLOG reduction should be stated explicitly.","section":"§4, Lemma 17"},{"comment":"The equality S=⋃_{ℓ≤d}⋃_{a∈S^(ℓ)}Ball(a,ℓ) and the subsequent counting of isomorphism types are only sketched; a few more sentences justifying the finite support argument would improve readability.","section":"Corollary 5"}],"recommendation":"major_revision","confidential_remarks":"I agree with the conditional verdict. The main obstacle is the gap in Lemma 12, but the repair is local and I have verified that the strengthened pigeonhole condition makes the distance identity correct. The manuscript should be suitable after the authors fix the Lemma 12 proof, correct the Lemma 11 inference, and address the small typographical and clarity issues listed above."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know this paper is a serious step on the Briggs-Feng-Wells question. The main dichotomy—every d-maximal set either contains a non-trivial Hamming ball or has size at most (n+o(n))^d—is exactly the kind of structural result that was missing. The machinery is original: regular templates, the S^(ℓ) hierarchy, and the generalized sunflower lemma. The proofs are detailed and mostly rigorous, and the finiteness corollary plus the binary upper bound (4-10^{-10})^d are solid secondary results.\n\nThe soft spots are real but narrow. Lemma 12's proof has a gap: the pigeonhole step only guarantees a j with {i>m': a^j_i = z_i ≠ w_i} contained in [m'], but the distance identity used immediately after requires that the sets {i>m': a^j_i ≠ w_i} and {i>m': z_i ≠ w_i} be disjoint. As printed the identity can fail. The fix is local—the same counting with dist(w,z) ≤ d < p actually yields a j whose E_j is disjoint from {i>m': z_i≠w_i}. So the lemma statement and the main theorems survive. There are also two typos: the floating-coloring stopping condition says 'less than f+m' where it must be 'greater than f+m', and the summation thresholds in Section 4 don't align (199/200d vs 19999/20000d). Both are cosmetic. The abstract's (n+o(n))^d is slightly stronger than the displayed d^2 factor, but that's a constant and not concerning.\n\nThe lower-bound proof via fractional rounding and Azuma is neat and checks out, and the constructions in Theorem 2 are explicit. Citation practice looks fair; no circularity, no fitted parameters. The generalized sunflower translation is a genuinely novel step, even if it needs the p>n condition to work.\n\nFor whom? Anyone working on Hamming-space diameter problems, maximal sets, or sunflower-type arguments. This deserves a serious referee. I would send it out, with a request to fix Lemma 12 and the typos. The core results are likely correct and important, and the proof strategy will be useful beyond this paper. Once the corrected version appears, I would cite it.","headline":"A genuine advance on the lifting-stability problem, with a real but local gap in Lemma 12 that is fixable without changing the main results.","tokens_in":12221,"tokens_out":7890,"would_cite":true,"duration_ms":76820,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05D05","05C35","05D40"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every d-maximal Hamming-cube set is small unless it contains a radius-1 ball.","keywords":["Hamming cube","d-maximal set","Hamming ball","sunflower lemma","diameter","lifting stability","extremal set theory","binary alphabet"],"falsifier":"A single construction settles the dichotomy: a d-maximal set in $[n]^\\infty$ with no 1-ball and size above $d^2(n+8n^{2/3})^d$ would refute Theorem 1; since the critical Lemma 11 is the bridge from set-sunflowers to word-sunflowers, a smaller counterexample would be a family of more than $\\mathrm{Sun}(p,2k)$ words at distance $k$ from a base word that contains no p-sunflower of words for some $p>n$.","tokens_in":11165,"feed_emoji":"🎲","tokens_out":16212,"duration_ms":172691,"temperature":0.7,"pith_summary":"This paper studies d-maximal subsets of the infinite Hamming cube over an n-letter alphabet: sets whose diameter is exactly d and that cannot be enlarged without increasing the diameter. The main result is a dichotomy: any such set that contains no Hamming ball of radius 1 has at most $d^2(n+8n^{2/3})^d$ elements, so any d-maximal set larger than this must contain a non-trivial ball. The bound is asymptotically tight because the d-dimensional cube $[n]^d\\times\\{0\\}$ has $n^d$ elements and is d-maximal. The same template-and-sunflower machinery yields a polynomial lower bound on the size of every d-maximal set and shows that only finitely many isomorphism types exist for fixed d and alphabet size n. This answers the motivating lifting-stability question: large finite d-maximal sets that remain d-maximal under coordinate lifting cannot avoid balls.","feed_headline":"In Hamming cubes, big diameter-maximal sets hold a 1-ball","feed_subtitle":"The proof splits every such set into a small one or a ball, with the cube giving the tight size barrier.","key_machinery":"The argument is carried by a template-refinement scheme. Fix a word $w\\in S$; a regular template is a word over $[n]\\cup\\{\\star\\}$ that differs from $w$ wherever it is specified, and its weight is the number of specified positions. Starting from the empty template, the proof repeatedly replaces one wildcard by a letter, keeping a nontrivial fraction of the original set that fits the template; Proposition 8 controls this fraction, and Lemma 9 is the only place where the absence of $(L+1)$-balls enters, guaranteeing a point $r\\in S$ at distance at least $d-L$ from any prescribed word. To prevent the fitting sets from becoming too large, Lemma 12 bounds the number of ball centers fitting a weight-$m$ template via a generalized sunflower lemma: after encoding each word at distance $k$ from $w$ by an auxiliary set system of size $2k$, the ordinary sunflower lemma forces a sunflower of words with respect to $w$, provided the number of petals $p$ exceeds the alphabet size $n$. The quantitative sunflower bound $\\mathrm{Sun}(p,k)\\le(Cp\\log k)^k$ then yields the polynomial factors and the $8n^{2/3}$ in the exponent.","core_discovery":"The central claim is Theorem 1: if $S\\subset[n]^\\infty$ is d-maximal and contains no 1-ball, then $|S|\\le d^2(n+8n^{2/3})^d$. Phrased as a dichotomy, every d-maximal set is either of size at most this quantity or contains a non-trivial Hamming ball. The constant is not the point; the qualitative content is that the only way to be large is to contain a ball, and the leading term $(n+o(n))^d$ is best possible, witnessed by the d-dimensional cube $[n]^d\\times\\{0\\}$. The paper proves the same dichotomy in a stronger form for sets that avoid larger balls: for a d-maximal set with no $(L+1)$-ball, the collection $S^{(\\ell)}$ of centers of $\\ell$-balls in it has size at most $C d^{2L+2}(n+8n^{2/3})^d$, which in turn implies that the number of isomorphism types of d-maximal sets is finite. For the binary alphabet the paper separates growth bases: it constructs d-maximal sets of size $\\binom{\\lfloor 3d/2\\rfloor}{d}\\ge(2.59+o(1))^d$ and proves that every ball-free binary d-maximal set has size at most $(4-10^{-10})^d$.","pith_inferences":["Beyond the paper, the only alphabet-size-sensitive step is the word-sunflower translation; replacing that translation by another structure theorem would generalize the dichotomy to other metric spaces.","The gap between the binary construction's growth base $2.59$ and the upper bound's base $4-10^{-10}$ is probably not intrinsic; a finite search for small d could map the true range.","The exponent $2/3$ in the lower bound comes from the discrepancy-style rounding step, so a different rounding method could raise the proven minimum size, possibly toward the $2^d$ growth of the known constructions.","Theorem 4's bounded skeleton of ball centers suggests that the right next object to classify is not the whole d-maximal set but the finite set of centers whose balls cover it."],"forward_implications":["Every d-maximal set in $[n]^\\infty$ with no 1-ball has at most $d^2(n+8n^{2/3})^d$ points, so lifting-stable large examples of diameter d necessarily contain a ball.","The d-dimensional cube $[n]^d\\times\\{0\\}$ is d-maximal and has $n^d$ elements, so the main bound is tight up to the $o(n)$ term.","For the binary alphabet, ball-free d-maximal sets can be exponentially large with growth base $2.59+o(1)$, but no such set reaches base $4-10^{-10}$; the true growth base lies in between.","Every d-maximal set has at least $(1/2)(d/\\log 2d)^{2/3}$ elements, a polynomial bound far below the exponential constructions.","For fixed d and alphabet size n, there are only finitely many d-maximal sets up to coordinate and alphabet permutations, so complete classification by finite search is possible in principle."],"supporting_citations":[{"why":"Establishes the sunflower lemma that the proof generalizes from set systems to words; all later sunflower bounds refine this statement.","marker":"[7]"},{"why":"Supplies the improved quantitative bound Sun(p,k) ≤ (C p log k)^k that makes the template bounds concrete.","marker":"[2]"},{"why":"Improves the sunflower bound used in Lemma 10 to the stated form.","marker":"[4]"},{"why":"Raises the lifting-stability question and provides the constructions that motivate the lower-bound problem.","marker":"[5]"},{"why":"Gives the extremal diameter-d subsets of the binary cube that are used in the proof of the binary upper bound.","marker":"[8]"},{"why":"Supplies Azuma's inequality and Hamming-ball entropy estimates used in the lower-bound and binary arguments.","marker":"[1]"}],"fun_headline_variants":["Big diameter-maximal sets must hold a Hamming ball","Small unless ball: tight size bound for maximal sets","No ball, small set: Hamming cube maximality dichotomy","Maximal sets in cubes: tight bound or ball inside","Diameter-maximal sets: finite types, tight size limit"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing step is the generalized sunflower lemma: many words at a fixed distance from a base word must contain p words whose differing coordinates are pairwise disjoint, and the proof of that step needs p to be larger than the alphabet size.","fun_headline_variants_meta":{"raw":{"variants":["Big diameter-maximal sets must hold a Hamming ball","Small unless ball: tight size bound for maximal sets","No ball, small set: Hamming cube maximality dichotomy","Maximal sets in cubes: tight bound or ball inside","Diameter-maximal sets: finite types, tight size limit"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000276,"raw_usage":{"total_tokens":1646,"prompt_tokens":944,"completion_tokens":702,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":560,"completion_tokens_details":{"reasoning_tokens":621}},"tokens_in":560,"tokens_out":702,"duration_ms":7558,"temperature":1.0,"reasoning_tokens":621,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T17:27:50.150636+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A single construction settles the dichotomy: a d-maximal set in $[n]^\\infty$ with no 1-ball and size above $d^2(n+8n^{2/3})^d$ would refute Theorem 1; since the critical Lemma 11 is the bridge from set-sunflowers to word-sunflowers, a smaller counterexample would be a family of more than $\\mathrm{Sun}(p,2k)$ words at distance $k$ from a base word that contains no p-sunflower of words for some $p>n$.","supporting_citations":[{"cited_title":"Erd˝ os and R","cited_arxiv_id":null,"evidence_quote":"Establishes the sunflower lemma that the proof generalizes from set systems to words; all later sunflower bounds refine this statement."},{"cited_title":"Improved bounds for the sunflower lemma","cited_arxiv_id":null,"evidence_quote":"Supplies the improved quantitative bound Sun(p,k) ≤ (C p log k)^k that makes the template bounds concrete."},{"cited_title":"Note on Sunflowers","cited_arxiv_id":"2009.09327","evidence_quote":"Improves the sunflower bound used in Lemma 10 to the stated form."},{"cited_title":"Facets in the Vietoris—Rips complexes of hypercubes","cited_arxiv_id":null,"evidence_quote":"Raises the lifting-stability question and provides the constructions that motivate the lower-bound problem."},{"cited_title":"Kleitman","cited_arxiv_id":null,"evidence_quote":"Gives the extremal diameter-d subsets of the binary cube that are used in the proof of the binary upper bound."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies Azuma's inequality and Hamming-ball entropy estimates used in the lower-bound and binary arguments."}],"review_version":1}