{"id":"49733952-c51f-4e23-8163-0aaa5168aea7","arxiv_id":"2607.18706","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Every mod-2 characteristic map over a simplicial 3-sphere with at most 20 vertices lifts to an integral characteristic map, via a new bad-block avoidance criterion.","lead":"The paper proves that every simplicial 3-sphere with at most 20 vertices admits an integral lift of every mod-2 characteristic map, extending a previous result limited to 8 vertices. Its new 'bad-block' criterion gives a simple counting condition for when such lifts exist, and it shows the method is sharp in a natural sense.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3.1 rests on two unverified finite facts: the det-3 classification of facets and the 168-block bad-block partition of X(Z2^4); if either is wrong, the avoidance criterion and the main theorem collapse.","rationale":"The reader's weakest-assumption analysis already identified the bad-block partition and the det-3 classification as the load-bearing unverified facts. My stress-test read agrees: Theorem 3.1's conclusion follows cleanly once Proposition 2.3 is granted, and Proposition 2.3 is a direct corollary of Lemma 2.2, which is only as strong as the two finite assertions in Section 2. I checked the rest of Theorem 3.1's proof for independent gaps: the use of the upper bound theorem for f3(K), the 15-colorability from nondegeneracy, and the counting argument for m=20 all appear correct. The bound f3(K) ≤ 165 < 168 is valid if the color-class reasoning is accepted. No internal inconsistency appears in the join-complex arguments or the image-size results, though their finite computations are similarly unverified. The concern is therefore not about the mathematical structure but about reproducibility of the finite backbone. Supplying the enumeration certificate or an independent verification script would convert CONDITIONAL to ACCEPT; without it, the reader's conditional verdict is appropriate. I would not raise the concern to a rejection because the asserted facts are plausible, finite, and explicitly attributed to prior work; they are the kind of claims that a small computation can settle quickly.","tokens_in":13916,"tokens_out":15656,"duration_ms":129933,"concrete_test":"Write a small independent script (e.g., Sage/GAP) that: (1) enumerates all 840 unordered facets of X(Z2^4); (2) for each facet computes det of the 4×4 0-1 matrix and records whether |det|=3; (3) checks that this set equals B exactly; (4) computes P(τ) for every facet τ as defined in Section 2, collects the distinct sets, and verifies they form a partition of the 840 facets into 168 sets of size 5; (5) for each block Bα, finds an explicit gα ∈ GL(4,Z2) with gα(Bα)=B. If any of these checks fails, Lemma 2.2 and Proposition 2.3 — and with them Theorem 3.1 — are unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim (Theorem 3.1) is proved by bounding f3(K) < 168 and invoking Proposition 2.3, which depends entirely on Lemma 2.2 and the bad-block partition of X(Z2^4). Two finite assertions carry this load, and neither is proved in this paper:\n\n1. Lemma 2.1 asserts that the five facets in B are exactly the facets of X(Z2^4) whose 4×4 0-1 representative has determinant of absolute value 3. This is stated as a definition/known fact, not verified. If any other facet also has determinant ±3, the {0,1}-representative in Lemma 2.1 need not be unimodular, and the lift construction fails.\n\n2. Section 2 asserts that F(X(Z2^4)) admits a partition into 168 five-facet blocks, each a GL(4,Z2)-translate of the standard bad block B, citing [3, Theorem 6.1]. The paper also says the distinct sets P(τ) form this partition, but gives no proof that P(τ) are pairwise disjoint or that every facet lies in exactly one such block. If the blocks overlap or omit facets, the pigeonhole argument in Proposition 2.3 breaks.\n\nThese are finite combinatorial facts that can be checked directly, but the manuscript provides no code, certificate, or detailed enumeration. The surrounding mathematics — the upper bound theorem, the 15-colorability argument, and the f3(K) ≤ 165 bound — is internally sound. The weak point is precisely the imported finite classification/partition.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the toric lifting property for simplicial 3-spheres: given a mod 2 characteristic map λ_R over K, when does it admit an integral characteristic lift? The main result (Theorem 3.1) states that every simplicial 3-sphere with at most 20 vertices has the toric lifting property. The proof introduces a 'bad-block partition' of the 840 facets of the universal complex X(Z_2^4) into 168 five-facet blocks. Lemma 2.2 shows that if the set of facet images of a characteristic map avoids any one block, then a lift exists; Proposition 2.3 then gives liftability whenever the number of facets is <168. The vertex bound is obtained by combining the upper bound theorem for m≤19 and a 15-coloring argument for m=20, giving f_3(K)≤165<168. The paper also proves liftability for neighborly simplicial 3-spheres, flag 3-spheres with at most 25 vertices, maps with at most 13 distinct vertex images, orientable maps, and nontrivial joins. Finally, it shows that the 13-vertex bound in Lemma 3.5 is sharp in the universal-complex sense via a finite congruence obstruction, and constructs a simplicial 3-sphere mapping surjectively to the facets of X(Z_2^4).","tokens_in":14318,"tokens_out":5128,"duration_ms":44762,"significance":"If the main theorem is correct, it is a substantial extension of the previously known PL-sphere result (≤8 vertices) to all simplicial 3-spheres on ≤20 vertices, and it introduces a potentially useful bad-block avoidance criterion. The paper also gives elegant uniform treatments of neighborly spheres, flag spheres up to 25 vertices, orientable maps, and joins. The proof structure is transparent: the heavy lifting is reduced to finite combinatorial facts about X(Z_2^4). The join results (Section 4) are proved by careful counting and do not rely on unshown computations. However, the central claim depends on two finite assertions (the det-3 classification in Lemma 2.1 and the bad-block partition in Section 2) that are either quoted from the authors' earlier work or asserted without proof or machine-checkable certificate. The paper also contains a 'direct determinant computation' in Lemma 3.5 over 616 facets and a computational obstruction output in Proposition 5.2, neither of which is reproducible from the manuscript as written. These gaps do not appear to be conceptual, but they are load-bearing and should be fixed before the paper is accepted.","major_comments":[{"comment":"The statement that the five facets in B are exactly the facets of X(Z_2^4) whose {0,1}-representative determinant has absolute value 3 is load-bearing: Lemma 2.1, Lemma 2.2, Proposition 2.3, and Theorem 3.1 all depend on it. It is asserted without proof or citation. Since there are only 840 facets, this is a finite check, but the paper should either give a short argument (e.g., using the GL(4,Z_2) action and the explicit forms of determinant-±3 facets) or include a reproducible verification (e.g., code or a table). As it stands, the reader cannot independently confirm this classification from the manuscript.","section":"Section 2, Lemma 2.1"},{"comment":"The partition F=⊔_{α=1}^{168} B_α, with each B_α a GL(4,Z_2)-translate of B, is quoted from [3, Theorem 6.1]. In addition, the paper asserts that 'the distinct sets P(τ), as τ runs over the facets, form the bad-block partition'. This latter claim — pairwise disjointness and covering — is not proved or given an exact theorem reference. Proposition 2.3's pigeonhole argument requires that the blocks be a true partition. Please either prove directly that the sets P(τ) partition F, or state precisely which result in [3] implies this and how.","section":"Section 2, bad-block partition"},{"comment":"The proof says 'a direct determinant computation shows' that every facet of X[S] gives determinant ±1 under the explicit 4×13 integer matrix L. This is a finite verification over the 616 facets of X[V\\{a_5,a_{14}}], and it underpins Theorem 3.6 and Corollary 3.7. As written, this is an unverifiable computation. The authors should provide the verification: either the code/script used, a table of the determinants, or a symmetry argument reducing the 616 cases to a manageable number. Without this, the image-size theorem rests on an unsupported claim.","section":"Section 3, Lemma 3.5"},{"comment":"The proof of non-liftability of the 14-vertex inclusion depends on Algorithm 1, but the survivor counts after each assigned vertex are stated without the program or a certificate. While this is a sharpness result rather than the main theorem, it is still a computational proof. The authors should provide the code or a detailed reproducible trace so that the UNSAT claim can be independently checked. At minimum, specify the exact version of the algorithm and all data structures so that a reader could rerun the computation.","section":"Section 5, Proposition 5.2"}],"minor_comments":[{"comment":"In the proof, 'fλ_R' appears to be a typo; it should be the map cλ_R (or λ_R). The sentence 'Since K is neighborly, fλ_R is injective on vertices' should read 'cλ_R is injective on vertices'.","section":"Section 3, Corollary 3.2"},{"comment":"The assertion 'every 4×4 0-1 matrix has determinant of absolute value at most 3' is used without proof or reference. A one-sentence justification (e.g., by the Hadamard bound or by noting the maximum determinant of a 4×4 0-1 matrix is 3) would be helpful.","section":"Section 2, Lemma 2.1 proof"},{"comment":"In the construction of D_N(v), the 'sign orbits' C_N(v)/{±1} may be confusing when N is even and a vector could coincide with its negative modulo N. Clarify that the algorithm chooses representatives in a fixed way; the proof of Lemma 5.1 handles this, but a remark at the definition would improve readability.","section":"Section 5, Algorithm 1"},{"comment":"In the case-by-case count, phrases such as 'we do not group the remaining case' are clear in context, but a brief explanation of why the ungrouped cases still contribute at most the stated number would aid the reader.","section":"Section 4, Lemma 4.2"}],"recommendation":"major_revision","confidential_remarks":"The central result is likely correct, but the manuscript as submitted is not self-contained in exactly the places that matter: Lemma 2.1, the P(τ)-partition claim, and the finite computations in Lemma 3.5 and Proposition 5.2. The reliance on [3] for the bad-block partition is legitimate if the precise statement is included, but the paper should explicitly state what is being imported. The authors should be asked to supply proofs or electronic certificates for the finite checks. If those are provided, the paper would be a strong contribution. I would not reject on the current evidence, but I cannot accept without verification of these load-bearing finite facts."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: the main theorem is real and the proof structure is sound, but the load-bearing piece is a finite combinatorial fact imported from the authors' own earlier paper, and the manuscript doesn't give the reader a way to check it. I'd send it to a referee, but I'd ask for the finite checks to be made reproducible.\n\nWhat's new: the 20-vertex bound for all simplicial 3-spheres is a genuine advance over the previous 8-vertex result for PL spheres. The proof is clean: the upper bound theorem plus a 15-colorability argument gives f3(K) ≤ 165 < 168, and the bad-block avoidance criterion says any characteristic map whose facet images miss one of 168 blocks lifts. That criterion is a useful tool in its own right. The corollaries for neighborly spheres, flag spheres with at most 25 vertices, join complexes, and image-size at most 13 all extend the literature. The sharpness results—no lift for any 14-vertex induced subcomplex of X(Z2^4), and a facet-surjective map from a 3-sphere—are substantive.\n\nThe soft spot is verification. The partition of F(X(Z2^4)) into 168 five-facet blocks and the classification of facets with determinant ±3 are stated as facts from [3], the authors' earlier work. If either is wrong, Lemma 2.2 and Theorem 3.1 fall. I don't think they are wrong—the citation is legitimate—but a result this dependent on a specific finite partition should give the referee a way to see it: a short enumeration, a certificate, or a link to code. The same goes for the 'direct determinant computation' in Lemma 3.5 and the UNSAT run in Proposition 5.2; the table of surviving assignments helps, but an independent artifact would settle it. These are fixable, not fundamental.\n\nWho this is for: people working on toric lifting, small covers, and universal complexes. The avoidance criterion might get reused. The paper does not resolve the general lifting problem, but it makes a solid dent.\n\nRecommendation: accept for peer review. Ask the authors to include the finite verification for the bad-block partition, the det-±3 classification, and the UNSAT computation—preferably as a small script or an appendix with the enumeration. The mathematics is coherent; the missing piece is reproducibility.","headline":"Solid advance in toric lifting with a clean 20-vertex theorem, but the load-bearing bad-block facts are quoted rather than independently verified.","tokens_in":14808,"tokens_out":3116,"would_cite":true,"duration_ms":52876,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["57S12","14M25","52B05","05E45"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every simplicial 3-sphere with at most 20 vertices has the toric lifting property: mod-2 characteristic maps lift to integral ones.","keywords":["toric lifting property","characteristic maps","simplicial 3-spheres","universal complex","bad-block partition","lifting problem","mod 2 characteristic maps"],"falsifier":"A brute-force enumeration of the 840 facets of $X(Z_2^4)$ checking two facts would settle the load-bearing classification: (i) the five listed facets of block $B$ are exactly those with 0-1 determinant $\\pm 3$, and (ii) the 168 translates of $B$ under $GL(4,Z_2)$ partition the facet set. If either fails, Lemma 2.2 is false. Separately, running Algorithm 1 with $N = 8$ on the 14-vertex induced subcomplex $X[V \\setminus \\{a_{15}\\}]$ and getting SAT would refute Proposition 5.2.","tokens_in":13813,"feed_emoji":"🔺","tokens_out":11831,"duration_ms":81266,"temperature":0.7,"texified_at":"2026-08-05T21:32:05.577829+00:00","pith_summary":"The paper proves a vertex-count threshold for a basic lifting problem in toric topology: every simplicial 3-sphere with at most 20 vertices has the toric lifting property, meaning every mod-2 characteristic map (a labeling of vertices by vectors in $Z_2^4$ that makes each tetrahedron a basis) lifts to an integral characteristic map (a labeling by integer vectors making each tetrahedron unimodular). The proof reduces liftability to a count of tetrahedra. The $840$ facets of the universal complex $X(Z_2^4)$ split into $168$ 'bad blocks' of five facets each, and a map lifts as soon as its tetrahedron images miss one whole block. For a 20-vertex 3-sphere the number of tetrahedra is at most $165$, so a miss is forced. The same criterion yields liftability for all neighborly 3-spheres, all flag 3-spheres on at most 25 vertices, all maps with at most 13 distinct vertex images (including every orientable map), and every nontrivial join.","texify_model":"deepseek-v4-flash","texify_usage":{"total_tokens":7514,"prompt_tokens":968,"completion_tokens":6546,"prompt_tokens_details":{"cached_tokens":0},"prompt_cache_hit_tokens":0,"prompt_cache_miss_tokens":968,"completion_tokens_details":{"reasoning_tokens":5618}},"feed_headline":"All 3-spheres with at most 20 vertices get integral lifts","feed_subtitle":"A counting criterion on the universal complex extends the known lifting result from 8 to 20 vertices.","key_machinery":"The central object is the universal complex $X(Z_2^4)$: its 15 vertices are the nonzero vectors of $Z_2^4$ and its 840 facets are the quadruples forming bases. Its facet set admits a partition into 168 'bad blocks' of five facets each, each a $GL(4,Z_2)$-translate of the standard block $B = \\{ \\{a_3,a_5,a_9,a_{14}\\}, \\{a_3,a_6,a_{10},a_{13}\\}, \\{a_5,a_6,a_{11},a_{12}\\}, \\{a_7,a_9,a_{10},a_{12}\\}, \\{a_7,a_{11},a_{13},a_{14}\\} \\}$ — exactly the facets whose {0,1}-representatives have determinant of absolute value 3. The workhorse is the bad-block avoidance criterion (Lemma 2.2): a mod-2 characteristic map lifts whenever its facet images are disjoint from some block, which reduces lifting to the counting question of whether the number of d","core_discovery":"The central claim is that the obstruction to lifting mod-2 characteristic data over a simplicial 3-sphere is purely combinatorial and finite: it lives in the universal complex $X(Z_2^4)$, whose 840 basis-facets partition into 168 five-facet 'bad blocks' under $GL(4,Z_2)$. The paper proves (Lemma 2.2) that if a mod-2 characteristic map's facet images avoid at least one bad block, then an integral lift exists. Since a simplicial 3-sphere on $m \\leq 20$ vertices has at most 165 tetrahedra — by the upper bound theorem plus a 15-colorability argument that caps edges at 185 — the set of facet images has size $< 168$ and must miss a block, yielding Theorem 3.1. The same criterion drives the image-size ($\\leq 13$) a","pith_inferences":["The counting slack (165 < 168) leaves room for a sharper argument: a more refined 15-coloring bound on the 1-skeleton of a 21-vertex 3-sphere could push the vertex threshold beyond 20 without new ideas.","The sharpness result for 14-element subsets suggests that the true obstruction to lifting is the size of the image of the characteristic map, not the vertex count of the sphere; a general 'image-size lifting' principle may hold for other dimensions.","Algorithm 1 offers a practical, machine-checkable UNSAT obstruction test; applying it directly to a 21-vertex construction would test whether the 20-vertex bound is sharp."],"forward_implications":["Every simplicial 3-sphere with at most 20 vertices has the toric lifting property, extending the previously known 8-vertex bound for PL 3-spheres (Theorem 3.1).","Every neighborly simplicial 3-sphere — where every pair of vertices spans an edge — has the toric lifting property (Corollary 3.2).","Every flag simplicial 3-sphere with at most 25 vertices has the toric lifting property (Theorem 3.4).","Every mod-2 characteristic map with at most 13 distinct vertex images lifts; in particular every orientable mod-2 map lifts (Theorem 3.6, Corollary 3.7).","Every simplicial 3-sphere that is a nontrivial join of lower-dimensional simplicial spheres has the toric lifting property (Theorem 4.3)."],"fun_headline_variants":["Avoid one bad block, lift every 3-sphere","All 3-spheres up to 20 vertices get integral lifts","Counting blocks: every 3-sphere ≤20 vertices lifts","Toric lifting proven for all 3-spheres under 21 vertices","No lifting obstruction in 3-spheres up to 20 vertices"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The whole argument depends on the quoted claim, borrowed from the authors' earlier work without re-derivation, that the 840 facets of $X(Z_2^4)$ partition into 168 five-facet bad blocks and that the only facets whose 0-1 determinants have absolute value 3 are the five in the standard block $B$; if that classification is wrong, the avoidance criterion and Theorem 3.1 collapse.","fun_headline_variants_meta":{"raw":{"variants":["Avoid one bad block, lift every 3-sphere","All 3-spheres up to 20 vertices get integral lifts","Counting blocks: every 3-sphere ≤20 vertices lifts","Toric lifting proven for all 3-spheres under 21 vertices","No lifting obstruction in 3-spheres up to 20 vertices"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000254,"raw_usage":{"total_tokens":1354,"prompt_tokens":641,"completion_tokens":713,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":385,"completion_tokens_details":{"reasoning_tokens":624}},"tokens_in":385,"tokens_out":713,"duration_ms":8416,"temperature":1.0,"reasoning_tokens":624,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T14:36:08.343514+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A brute-force enumeration of the 840 facets of $X(Z_2^4)$ checking two facts would settle the load-bearing classification: (i) the five listed facets of block $B$ are exactly those with 0-1 determinant $\\pm 3$, and (ii) the 168 translates of $B$ under $GL(4,Z_2)$ partition the facet set. If either fails, Lemma 2.2 is false. Separately, running Algorithm 1 with $N = 8$ on the 14-vertex induced subcomplex $X[V \\setminus \\{a_{15}\\}]$ and getting SAT would refute Proposition 5.2.","supporting_citations":[],"review_version":1}