{"id":"0f05fecb-a766-4941-bb9b-bf02b599b791","arxiv_id":"1908.07133","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For two nonreducible planar graphs the computed dimension of a Khovanov-Robert foam invariant exactly matches the Tait number; for the dodecahedral graph the computation leaves dimensions 58 or 60, with evidence pointing to 58.","lead":"This paper uses a computer program to compute limits on the size of a vector space that mathematicians attach to certain planar graphs called webs. The calculations test whether a combinatorial invariant equals the number of three-colorings of a graph, a question connected to a new route to the four-color theorem.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Exactness for W2/W3 rests on an unverifiable computation: the program is only referenced via a website, with no version or hash; a bug in facet/coloring enumeration or Smith normal form would invalidate Theorem 4.2.","rationale":"I agree with the reader's weakest assumption. The internal mathematics is not the source of risk: the inequalities ell_q <= qdim J^flat <= qrank<K>_phi are correctly established, and the equality ell = Tait for W2 and W3 is derived from the table entries. The load-bearing premise is that the computer program correctly implements the evaluation formula, the facet and coloring enumeration, and the Smith normal form over F[E]. The program is only referenced through a bare author URL with no version or hash, so the numerical results in Tables 2 and 3 are not reproducible from the manuscript; a bug in any step would invalidate Theorem 4.2 despite the clean logical skeleton. The W1 'suggestion' of 58 is explicitly heuristic and the paper treats it honestly, so it should not strengthen or weaken the verdict beyond the computational concern. Since the reader already assigned a CONDITIONAL verdict based on exactly this premise, my stress-test does not move the verdict. If the code is deposited and independently reproduced, the verdict should be ACCEPT; if not, the exactness claim should remain conditional or be marked unverified.","tokens_in":15458,"tokens_out":8299,"duration_ms":89805,"concrete_test":"Deposit the exact version of the Mathematica program used for Tables 2 and 3 in arXiv or Zenodo with a SHA-256 hash, then in a clean run reproduce every row of Table 2 and the ell_q and r_q polynomials of Table 3. Independently, reimplement the Khovanov-Robert closed foam evaluation and the Smith normal form over F[E] in a second language, such as SageMath or Python, and recompute the W2 and W3 rows; exact reproduction of ell(W2) = 120, ell(W3) = 162, and the corresponding polynomials would confirm Theorem 4.2. If either step fails or the code cannot be obtained, the exactness claim for W2 and W3 remains unverified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proofs in Section 4 are mathematically sound conditional on the computed numbers: Corollary 3.1 gives dim J^flat(K) <= Tait(K), Theorem 4.1 gives ell_q(K) <= qdim J^flat(K), and if the Gram-rank computation is correct, ell(K) = Tait(K) for W2 and W3 forces equality. The entire burden is on the implementation: facet enumeration (Remark 3.6), admissible coloring enumeration, Khovanov-Robert evaluation (Eqs. 4-7), and Smith normal form over F[E] (Eqs. 23-32). The program is not included in the arXiv record; reference [3] is a bare author URL with no version or checksum, so Tables 2 and 3 cannot be independently checked from the manuscript. A second, honestly labeled heuristic is the W1 'suggestion' of 58: the lower bound reaches 58 at N_ell = 156 and stays there through N_e = 6,727 of N = 11,160 half-foams, but this is finite evidence, not a proof. The paper itself does not promote this to a theorem, so the central verified claim is the W2/W3 computation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper describes a Mathematica program that computes lower bounds on the dimension and quantum dimension of the Kronheimer-Mrowka combinatorial foam-evaluation functor J^flat for planar webs, using Khovanov-Robert's closed-foam evaluation formula on finite sets of half-foams. These lower bounds are sandwiched between the known upper bounds (dim J^flat(K) <= Tait(K) and qdim J^flat(K) <= qrank<K>_phi) to obtain exact values for two fullerene webs W2 and W3 (Theorem 4.2). For the dodecahedral web W1 the computation gives a lower bound of 58 while Tait(W1)=60, and the paper suggests, without claiming a proof, that dim J^flat(W1)=58. The paper also develops structural results (Corollary 3.1, Theorems 4.1 and 4.3) that make the computational results rigorous conditional on the correctness of the program's output.","tokens_in":15645,"tokens_out":13464,"duration_ms":129345,"significance":"If the computations are correct, the paper achieves the first nonreducible webs for which dim J^flat equals the Tait number, and the first computation of qdim J^flat equal to qrank of the phi-module for such webs. The dodecahedral web W1 becomes a concrete candidate for a negative answer to Khovanov-Robert's Question 2.1. The mathematical framework is clean and parameter-free, and the paper is explicit in distinguishing proved theorems from the heuristic suggestion for W1. The main weakness is that the central numerical claims rest on an unversioned external program that the reader cannot inspect or verify from the manuscript.","major_comments":[{"comment":"The exact entries in Tables 2 and 3, on which Theorem 4.2 and the W1 discussion rest, are outputs of a Mathematica program that is not included in the submission; reference [3] is a bare URL with no version, checksum, or file name. The manuscript describes the algorithm at a high level but provides no test cases, no output files, and no certificates for the Smith normal form computations in Eqs. (23)-(32). Because a bug in facet enumeration, coloring enumeration, closed-foam evaluation, or the Smith normal form computation would invalidate Theorem 4.2, the central claim is not independently checkable as submitted. I request that the complete program be made available in a versioned form as supplementary material, and that the authors provide verifiable certificates for the table values (for example, the matrices S, B, T for the Smith decompositions, or an independent verification script).","section":"Section 3.2, Remark 3.6, Tables 2-3"}],"minor_comments":[{"comment":"The inference from ell(K)=Tait(K) to equality of the three quantum dimensions in Eq. (33) is valid but requires the observation that ell_q(1)=ell(K)=Tait(K)=qrank<K>_phi(1); since the inequalities in Eq. (33) are coefficientwise, this forces ell_q = qdim J^flat(K) = qrank<K>_phi. Please add a sentence spelling out this argument.","section":"Section 4, proof of Theorem 4.2"},{"comment":"The sentence 'dim J^flat(W1) must be either 58 and 60' should read 'either 58 or 60'.","section":"Introduction, page 1"},{"comment":"The statement 'For the empty web K = emptyset we take S(K) = emptyset' is ambiguous: if S(emptyset) is the empty set, the recursive construction for reducible webs cannot produce any half-foams. Please clarify whether S(emptyset) is intended to be the singleton set containing the empty half-foam, and make the convention explicit.","section":"Section 3.1"},{"comment":"The reference to the computer program should be a complete citation with a specific file or URL, an access date, and a version or checksum, rather than a bare homepage URL.","section":"Reference [3]"},{"comment":"The columns N_ell, N_e, and N are defined in the text but not in the caption; please include their definitions in the caption so that the table is self-contained.","section":"Table 2"},{"comment":"In the sentence 'if use only IH cobordisms' the word 'we' is missing; please fix this typographical error.","section":"Remark 4.2"}],"recommendation":"major_revision","confidential_remarks":"The main obstacle is reproducibility: Theorem 4.2 is a computational theorem, and a paper of this kind should not rely on an unversioned external download. If the authors supply the program, its version, and verifiable outputs or certificates, I would be willing to accept the paper after checking those materials. The W1 suggestion is honestly labeled as a heuristic and does not by itself block publication, though the abstract should continue to emphasize that it is a suggestion rather than a proved result."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing to know: the paper gives the first exact evaluations of dim J^flat for two nonreducible fullerene webs, and narrows the dodecahedral web to 58 or 60 with a suggested 58. If the computations are right, that's a meaningful data point for Khovanov-Robert's Question 2.1, and the author is honest that a negative answer for W1 wouldn't break Kronheimer-Mrowka's four-color strategy.\n\nWhat's genuinely new is the data. The lower-bound method—using finite sets of half-foams and the rank of the restricted bilinear form—is known from Kronheimer-Mrowka and Khovanov-Robert; the novelty is that the author pushed it far enough to get exact dimensions for W2 and W3. That is a real computational achievement. The proof skeleton in Section 4 is clean: Corollary 3.1 and Theorem 4.1 sandwich qdim, and if the table values are correct, equality follows. The Smith normal form over F[E] is explained carefully, and the modular/quantum rank arguments check out.\n\nNow the soft spot, and it's the load-bearing one. Tables 2 and 3 come from a Mathematica program that is not in the arXiv record. Reference [3] is a bare author URL with no version, no checksum, no code listing. That means Theorem 4.2 rests entirely on an unverifiable computation. A bug in facet enumeration, coloring enumeration, or Smith normal form would change the results. This is not a minor issue; it is the entire difference between a proven theorem and a conjecture from a black box. The author should be required to ship the code (or a certificate) as a condition of publication.\n\nThe W1 'suggestion' of 58 is handled properly. It is labeled a suggestion based on saturation of the lower bound after many half-foams, not a theorem. The paper even notes that the alternative 60 would keep Question 2.1 open. That's honest. It doesn't need to be fixed; it just shouldn't be over-read.\n\nCitations look fine. No parameter fitting, no circularity. The computational lower bounds are genuinely independent of the conclusion.\n\nWho is this for? People working on foam evaluation and the Kronheimer-Mrowka program, and anyone interested in whether J^flat equals Tait for nonreducible webs. A serious referee should be assigned, because the data matter and the computation must be checked—preferably by running the code. I'd condition acceptance on code release and independent verification. Don't desk reject; do send it to review.","headline":"The first exact computations for nonreducible webs are real, but the missing code means the central theorem is only as good as an uncheckable table.","tokens_in":16234,"tokens_out":2695,"would_cite":false,"duration_ms":23822,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["57K18","05C15"],"pacs":[],"model":"deepseek-v4-flash","headline":"A computer program proves that the combinatorial foam invariant equals the Tait coloring count on two fullerene webs, and constrains the dodecahedral web to 58 or 60.","keywords":["foam evaluation","Tait colorings","webs and foams","dodecahedral web","fullerene graphs","four-color theorem","quantum dimension","Smith normal form"],"falsifier":"Recompute the ranks in Tables 2 and 3 with an independent implementation of the closed-foam evaluation formula; any difference would invalidate Theorem 4.2, and for $W_1$ a single new half-foam that raises the lower bound above 58 would refute the suggested value.","tokens_in":15191,"feed_emoji":"🕸️","tokens_out":12497,"duration_ms":116301,"temperature":0.7,"pith_summary":"The paper builds a computer program that computes lower bounds on the dimension and graded dimension of $J^\\flat(K)$, the combinatorial foam invariant associated to a planar trivalent graph, or web. It uses these bounds to test the open question of whether $\\dim J^\\flat(K)$ always equals the number of Tait colorings $\\mathrm{Tait}(K)$ of $K$. For the fullerene webs $W_2$ and $W_3$ the program proves the equality; for the dodecahedral web $W_1$ it proves $\\dim J^\\flat(W_1)\\in\\{58,60\\}$ and, because the lower bound saturates at 58 across thousands of half-foams, suggests the true value is 58 rather than the Tait count 60. If that value stands, $W_1$ is the first nonreducible web where the combinatorial invariant and the Tait count diverge, settling the open question negatively without disturbing the gauge-theoretic route to the four-color theorem.","feed_headline":"Foam invariant matches Tait on two webs, may miss dodecahedron","feed_subtitle":"Computer bounds prove equality for two fullerene webs and pin the dodecahedron to 58 or 60, with 58 suggested.","key_machinery":"The load-bearing object is the closed-foam evaluation formula implemented by the program. For a closed foam $F$, the formula expresses a symmetric polynomial $\\langle F\\rangle=\\sum_{c\\in\\mathrm{adm}(F)}P(F,c)/Q(F,c)$ over the field of two elements, summing over admissible colorings of the foam's facets, and $J^\\flat(F)$ is this polynomial evaluated at $E_1=E_2=E_3=0$. The program enumerates facets and admissible colorings, generates a finite set of half-foams with top boundary $K$, forms the matrix of pairings $(F_i,F_j)_\\varphi$ in the polynomial ring $\\mathbb{F}[E]$, and performs a Smith normal form decomposition over that ring. The rank of the pairing matrix on a finite subspace is a lower bound for $\\dim J^\\flat(K)$, while the theorem that $\\langle K\\rangle_\\varphi$ is a free module of rank $\\mathrm{Tait}(K)$ supplies the matching upper bound; when the two bounds coincide, all intermediate inequalities collapse to equalities.","core_discovery":"On the paper's own terms, the discovery is a pair of computations. Theorem 4.2 states that for the webs $W_2$ and $W_3$ shown in Figure 6, $\\dim J^\\flat(K)=\\operatorname{rank}\\langle K\\rangle_\\varphi=\\mathrm{Tait}(K)$ and $\\operatorname{qdim} J^\\flat(K)=\\operatorname{qrank}\\langle K\\rangle_\\varphi=\\ell_q(K)$, so both the dimension and the graded dimension collapse to the Tait data. For the dodecahedral web $W_1$, the program yields $\\dim J^\\flat(W_1)\\in\\{58,60\\}$; the lower bound 58 is reached after 156 half-foams and remains 58 through 6,727 of the 11,160 half-foams generated, which the paper takes as strong evidence that $\\dim J^\\flat(W_1)=58$. Since $\\mathrm{Tait}(W_1)=60$, this would make $W_1$ the first nonreducible web for which the equality question is answered negatively.","pith_inferences":["The exactness of the results for $W_2$ and $W_3$ could be made independently checkable by releasing the program's source code and a transcript of the Smith normal form computations; as presented, the tables are the only evidence, and a single bug in facet enumeration or arithmetic could change a rank.","The saturation at 58 for $W_1$ is a heuristic, not a proof; a finite certificate for $\\dim J^\\flat(W_1)=58$ would require either a spanning set of half-foams for the whole space or a proof that the pairing matrix rank cannot increase past 58, for example from a degree bound on generators.","The dodecahedral web is a pivot between two conjectures: the symmetry of $\\operatorname{qrank}\\langle K\\rangle_\\varphi$ under $q\\mapsto q^{-1}$ would force $\\dim J^\\flat(W_1)=60$, so the suggested 58 would refute that symmetry as well as the equality question.","The same computational pipeline could probe other small fullerenes; for each new web the ratio $\\ell(K)/\\mathrm{Tait}(K)$ gives a cheap indicator of where the equality question might fail first."],"forward_implications":["For $W_2$ and $W_3$, equality between $\\dim J^\\flat(K)$ and $\\mathrm{Tait}(K)$ is now established for nonreducible webs, the first such examples, and the graded version is established as well through $\\ell_q(K)$.","For $W_1$, the dimension is proven to be either 58 or 60, leaving a single binary choice; the saturation data favor 58.","If $\\dim J^\\flat(W_1)=58$, the open question asking whether $\\dim J^\\flat(K)=\\mathrm{Tait}(K)$ for every web would have a negative answer, ruling out the specific strategy of identifying $J^\\flat$ with the gauge-theoretic functor.","For the other tested fullerene webs $W_4$ through $W_7$, the method gives nontrivial lower bounds strictly below the Tait counts, so their true dimensions are constrained to short intervals, such as 178, 179, or 180 for $W_4$.","The inequalities $\\dim J^\\flat(K)\\le\\mathrm{Tait}(K)\\le\\dim J^\\sharp(K)$ are tight for $W_2$ and $W_3$, so the whole chain collapses to a single number for those webs."],"supporting_citations":[{"why":"Supplies the closed-foam evaluation formula, the theorem that the associated module is free of rank Tait(K), and the equality question under test.","marker":"[6]"},{"why":"Introduces the gauge-theoretic functor, its combinatorial replacement, and the prior lower bound of 58 for the dodecahedral web.","marker":"[7]"},{"why":"Establishes the inequality dim J^sharp(K) >= Tait(K), completing the chain used to prove equality when lower bounds match.","marker":"[9]"},{"why":"Provides the closed evaluation formula that the program implements after adaptation.","marker":"[10]"},{"why":"Supplies the fullerene enumeration used to identify and label the example webs.","marker":"[4]"},{"why":"Provides the universal construction that converts closed-foam evaluations into the vector spaces J^flat(K).","marker":"[2]"}],"fun_headline_variants":["Dodecahedral web may be first counterexample to foam-Tait equality","Computer bounds: two webs match Tait, dodecahedron suggests 58 vs 60","Foam invariant equals Tait on two webs, dodecahedron may deviate","New computer bounds hint dodecahedral web may break Tait equality"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The exact statements for $W_2$ and $W_3$ assume that the paper's computer program correctly implements the closed-foam evaluation formula and the matrix computations; the suggested value for $W_1$ additionally assumes that the saturation of the lower bound at 58 over thousands of half-foams is a reliable sign of the true dimension.","fun_headline_variants_meta":{"raw":{"variants":["Dodecahedral web may be first counterexample to foam-Tait equality","Computer bounds: two webs match Tait, dodecahedron suggests 58 vs 60","Foam invariant equals Tait on two webs, dodecahedron may deviate","New computer bounds hint dodecahedral web may break Tait equality"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00104,"raw_usage":{"total_tokens":4410,"prompt_tokens":1013,"completion_tokens":3397,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":629,"completion_tokens_details":{"reasoning_tokens":3310}},"tokens_in":629,"tokens_out":3397,"duration_ms":24566,"temperature":1.0,"reasoning_tokens":3310,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:25:18.908981+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the ranks in Tables 2 and 3 with an independent implementation of the closed-foam evaluation formula; any difference would invalidate Theorem 4.2, and for $W_1$ a single new half-foam that raises the lower bound above 58 would refute the suggested value.","supporting_citations":[{"cited_title":"Brinkmann and W","cited_arxiv_id":null,"evidence_quote":"Supplies the fullerene enumeration used to identify and label the example webs."},{"cited_title":"Blanchet, N","cited_arxiv_id":null,"evidence_quote":"Provides the universal construction that converts closed-foam evaluations into the vector spaces J^flat(K)."}],"review_version":1}