{"id":"66bf8c6c-42f2-4373-9202-df63f32389ca","arxiv_id":"2505.16629","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Any non-side-to-side spherical tiling by congruent triangles with an irrational angle is either a 2-layer earth map tiling, an 8-tile family, or the unique 16-tile sporadic tiling.","lead":"Classifies all tilings of the sphere by congruent triangles when at least one angle is an irrational number of degrees, showing they fall into three families: 2-layer earth map tilings, an 8-tile one-parameter family, and a unique 16-tile tiling. The proof develops a matching lemma and an irrational-angle lemma that will also structure the remaining rational-angle case.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 8's Table 5 omits the extended edge c+(k+1)a=b+ka+b, which the proof of Lemma 8 itself derives; the Matching Lemma's exhaustiveness is therefore not established.","rationale":"The paper's central claim is a complete classification. The proof strategy reduces to a finite case analysis built on three lemmas, and Lemma 8 is the only place where a finite list of all possible extended edges is produced; if that list is incomplete, the case analysis is not exhaustive. Reading the proof of Lemma 8, I found a concrete, localizable gap: in the third subcase (starting with lengths b,c), the proof concludes three types of extended edges, but the displayed Table 5 contains only two of them. The type c+(k+1)a=b+ka+b is missing, even though the proof text derives it. This is not a vague worry about enumerative proofs; it is an internal inconsistency in the lemma as stated. The reader's weakest assumption was exactly the completeness of Lemma 8, so the finding confirms that concern. It does not prove the main theorem false, because the missing type might turn out to be impossible or might lead only to already-listed tilings; nevertheless, the proof as written is incomplete and the asserted exhaustive list in Lemma 8 is not established. The appropriate verdict remains conditional: the authors must either include the missing type in Table 5 and handle it in the subsequent analysis, or prove that it cannot occur. I also note the separate overclaim in Lemma 7 flagged by the reader (the statement covers half vertices while the proof only treats full vertices), but the main theorem appears to use only the full-vertex version, so that issue is secondary. My agreement with the reader is 'agree' because the same load-bearing lemma was identified, though the evidence here is more specific: a pattern is actually missing, not merely possibly missing.","tokens_in":17552,"tokens_out":10360,"duration_ms":79779,"concrete_test":"Verify whether the omitted edge type c+(k+1)a=b+ka+b can occur in an abc-tiling with no α at any half vertex. Start with the k=0 local configuration (c+a=2b) shown as the fourth picture in Figure 15, check the half-vertex angle conditions, and determine whether it extends to a valid tiling. Then re-run the arguments in Sections 3.2 and 3.3 that cite Lemma 8, adding this type to Table 5, and see whether any new AVCs or tilings arise; if none arise, the type must be proved impossible or shown redundant.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the third case of Lemma 8 (around Figure 15), the proof explicitly concludes three possible extended-edge types: c+ka+b=b+ka+c, c+(k+1)a+c=b+ka+b, and c+(k+1)a=b+ka+b. But Table 5 lists only the first two of these; the third type (with k=0 giving c+a=2b, the 'fourth picture' in Figure 15) is absent. Thus the lemma's stated conclusion that Table 5 contains all possible extended edges when α is not at any half vertex is internally inconsistent: the proof derives a case that the table omits. This matters because the AVC tables in Section 3 and the exclusion arguments in cases such as α2β, α2βγ, α4, and β4 repeatedly invoke Lemma 8 to rule out configurations, e.g. 'by Lemma 8, we have a=2c' and 'if a=2c, in the adjacent tiles ... there is a half vertex at β or γ.' If the omitted type is geometrically realizable, the list of possible extended edges is incomplete, and the subsequent classification may miss non-side-to-side tilings outside the three claimed families. At minimum, the proof of Lemma 8 as written fails to establish the exhaustiveness on which the main theorem depends.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops tools for classifying non-side-to-side tilings of the sphere by congruent triangles, specifically the irrational angle lemma (Lemma 7) and the matching lemma (Lemma 8) enumerating possible extended edges. The main theorem asserts that, for a triangle with any angle irrational in degrees, all non-side-to-side tilings are: (1) two-layer earth map tilings with 2k≥6 tiles and their rotational modifications for even k; (2) a one-parameter family of triangles each admitting a unique 8-tile tiling; and (3) a unique triangle admitting a unique 16-tile tiling. The proof derives possible anglewise vertex combinations (AVC) from low-degree full or half vertices, then eliminates or constructs tilings case by case, with exact geometric data in Table 3.","tokens_in":17738,"tokens_out":14447,"duration_ms":120113,"significance":"If correct, this completes a natural and previously open case in the classification of monohedral spherical triangle tilings, extending the edge-to-edge classification of Ueno-Agaoka and the isosceles/right-triangle results of Dawson and Doyle. The paper's strengths include explicit geometric data, detailed local diagram arguments, and a clear reduction to finitely many AVC cases using the authors' adjacent-angle-deduction method. The main theorem is specific and falsifiable, and no fitted parameters appear. However, the completeness of the classification depends on Lemma 8's exhaustive list of extended edges, and the omission identified below means the central claim is not yet established as written; the human case analysis also needs careful verification at the points where Lemma 8 is invoked.","major_comments":[{"comment":"In the third case of the proof of Lemma 8, after analyzing the case x=a, the text explicitly concludes three types of extended edges: c+ka+b=b+ka+c, c+(k+1)a+c=b+ka+b, and c+(k+1)a=b+ka+b. Table 5 lists only the first two of these (entries 3 and 7); the third type is missing. Thus the lemma's assertion that Table 5 contains all possible extended edges is internally inconsistent with its own proof. This is load-bearing: Section 3 repeatedly invokes Lemma 8 to conclude specific side relations, e.g. in Case α^2β ('by Lemma 8, we have a=2c'), in Case α^2βγ (where b=2c is concluded), and in Cases α^4 and β^4 (where a=2c or b=2c is derived). If the missing pattern c+(k+1)a=b+ka+b (for k=0, c+a=2b) is geometrically realizable, the list of possible extended edges is incomplete and the subsequent AVC-based exclusions could miss tilings outside the three claimed families. The authors should add the missing entry and either derive a contradiction from it or show that it leads only to already-listed tilings, and then re-verify every invocation of Lemma 8.","section":"§2, Lemma 8 and Table 5"},{"comment":"Lemma 7 is stated for 'all vertices' (both full and half), but its proof uses the equation n·(α,β,γ)=2 for every vertex n, which holds only for full vertices; half vertices have angle sum 1 (π), not 2. The statement is in fact false for half vertices: in Case α^3 with f=6, the half vertex βγ=(0,1,1) does not lie on the line through m=(3,0,0) and (2f/(f+4))u=(6/5,6/5,6/5). The lemma should be restated for full vertices only, with a separate argument (e.g., doubling a half vertex to a full vertex) for the half-vertex columns AVC1 in Tables 6–8. As written, the derivation of the AVC1 columns, which are used throughout Section 3, is not justified by the stated lemma.","section":"§2, Lemma 7"},{"comment":"The abstract and the theorem state a classification of all non-side-to-side tilings by congruent triangles with any irrational angle, but the proof explicitly restricts to scalene triangles: Section 2 says 'Dawson has completely classified ... isosceles triangles. Therefore, we only need to discuss convex and scalene triangle.' The theorem should either be restricted to scalene triangles or should explicitly incorporate the isosceles classification from [7,8]; otherwise the statement is broader than what is proven. This is not a mere wording issue, because the listed three families do not include Dawson's isosceles families, and the paper does not state whether Dawson's classification contains any irrational-angle isosceles tilings outside these families.","section":"Theorem statement and §2"}],"minor_comments":[{"comment":"Lemma 8 refers to 'Table 6' but the displayed table of extended edges is numbered 'Table 5'; the cross-reference should be corrected.","section":"§2, Lemma 8"},{"comment":"The references to 'the third picture' and 'the fourth picture' in the proof of Lemma 8 are difficult to follow because the subfigures in Figure 15 are not individually labeled; please add labels or explicit descriptions.","section":"Figure 15"},{"comment":"The sentence 'All other AVC in Table 6, 7, 8 are derived similarly' is very terse, especially for the half-vertex columns AVC1; after Lemma 7 is corrected, the derivation of these columns should be spelled out or a clear reference to the doubling argument should be given.","section":"§3, Tables 6–8"},{"comment":"The sentence 'The first six cases admit no half vertex' refers to Table 7 but does not list which six cases are meant; please make this explicit.","section":"§3.2"}],"recommendation":"major_revision","confidential_remarks":"The skeptical note is correct: Lemma 8's proof derives an extended-edge type that is absent from Table 5. This is a genuine internal inconsistency in a load-bearing lemma, not a stylistic issue. The overall approach is promising and the gap appears localized, so major revision rather than rejection is appropriate. I also recommend asking the authors to reconcile Lemma 7's statement with its proof and to qualify the theorem statement regarding the scalene assumption."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The headline is: this paper is a real advance, and the specific worry in the stress-test note about Lemma 8 does not survive contact with the text. The alleged missing extended edge c+(k+1)a=b+ka+b is the same equality as entry 4 in Table 5, b+ka+b=(k+1)a+c, just written in reverse order. So the Matching Lemma's table is not missing that case. That removes the only concrete incompleteness charge I saw.\n\nWhat is genuinely new: the Irrational Angle Lemma and the Matching Lemma, and the complete classification of non-side-to-side spherical tilings by congruent triangles when one angle is irrational. The three families in the main theorem—two-layer earth map tilings, the 8-tile family, and the 16-tile sporadic—are concrete, and the geometric data are given explicitly. The proof incorporates earlier isosceles and right-triangle classifications rather than hand-waving them. The AAD technique from the pentagon papers is imported and extended sensibly.\n\nSoft spots: Lemma 7 is overclaimed. It says 'all vertices n' satisfy the line condition, but the proof works only for full vertices (the right-hand side in the linear system is 2, not 1, for half vertices). The applications only use the full-vertex version, so this is an exposition bug, not a hole in the classification. The Matching Lemma's proof is terse in places; the exclusions in Figures 13–15 are plausible but a referee will want to check that no pattern is waved off too fast. No code or data accompanies the case analysis, so independent reimplementation is a nontrivial project. That is typical for this literature.\n\nOverall: the central argument holds up as far as I can see, and the classification is a major result for discrete geometry. The paper deserves a serious referee, ideally someone who knows the Dawson–Doyle examples and the Ueno–Agaoka edge-to-edge classification. I would bring it to a tilings reading group, and I would cite it if I work on spherical tilings. Send it to peer review.","headline":"A substantial classification result; the stress-test's 'missing' extended edge in Lemma 8 is actually already in Table 5, and the paper deserves a serious referee.","tokens_in":18270,"tokens_out":4018,"would_cite":true,"duration_ms":29672,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["52C20","05B45"],"pacs":[],"model":"deepseek-v4-flash","headline":"Three families exhaust all irrational-angle sphere tilings.","keywords":["spherical tiling","triangle monotile","non-side-to-side","half vertex","irrational angle","extended edge","adjacent angle deduction","matching lemma"],"falsifier":"Run an exhaustive search for non-side-to-side tilings of the sphere by congruent scalene triangles whose angles include an irrational number of degrees, for tile counts up to the bounds permitted by the counting identities; any output not congruent to a two-layer earth map tiling, the 8-tile tiling, or the 16-tile tiling, or any extended edge not listed in Table 5, would disprove the theorem.","tokens_in":17309,"feed_emoji":"🌐","tokens_out":9485,"duration_ms":75442,"temperature":0.7,"pith_summary":"An old open thread in spherical tiling theory asks which congruent triangles can cover the sphere without meeting side-to-side, meaning some tile corners lie in the interior of another tile's side. This paper closes that thread for every triangle that has at least one angle irrational in degrees. It proves that such tilings are exactly: one-parameter families of triangles admitting two-layer earth map tilings with 2k ≥ 6 tiles (plus rotation modifications when k is even), a one-parameter family of triangles each admitting a unique 8-tile tiling, and a single triangle admitting a unique 16-tile tiling. All the angular and side-length data for these monotiles are listed explicitly. A reader who wants to know whether a given irrational triangle can tile the sphere without side-to-side contact can now read the answer off the three listed families.","feed_headline":"Three families exhaust irrational-angle non-side-to-side tilings","feed_subtitle":"Only three families remain, completing the irrational-angle branch of a century-old classification problem.","key_machinery":"The argument is carried by a small set of combinatorial instruments adapted to half vertices. The irrational angle lemma (Lemma 7) forces the angle-count vector of every vertex to lie on an affine integer line through (1,1,1) and one known vertex vector; otherwise the angle equations would have a unique rational solution, contradicting irrationality. The matching lemma (Lemma 8) then enumerates every possible extended-edge equality—patterns such as b+ka = ka+b or c+(k+1)a = b+ka+b, plus the great-circle case—under the assumption that the largest angle α never occurs at a half vertex. Around these sit the balance lemma, the adjacent angle deduction, and the counting identities for full and half vertices, which together reduce the infinite classification problem to a finite table of possible vertex combinations that can be checked one by one.","core_discovery":"The paper's main theorem asserts a complete classification. In a non-side-to-side tiling of the sphere by congruent triangles with any irrational angle, the tile and the tiling must be one of the three listed types: the two-layer earth map family with 2k ≥ 6 triangles and its even-k rotation modifications, the one-parameter 8-tile family, or the sporadic 16-tile tiling. The phrase 'non-side-to-side' means that not every triangle corner is a corner of the tiling; those exceptional points are half vertices. Since degenerate and isosceles cases had already been classified, the proof assumes a convex scalene triangle and derives the full list of possible vertex combinations, eliminating every combination that cannot be completed. The result is stated up to rotation and global flip of the sphere.","pith_inferences":["If the matching lemma is indeed exhaustive, a mechanical search over all length-word matchings along a line with a > b > c under the no-α-half-vertex rule would reproduce exactly the patterns of Table 5; finding a new pattern would mean the theorem needs revisiting.","The theorem's rigidity suggests that non-side-to-side tiling with an irrational angle forces high symmetry, which is why the only families are layered earth maps plus two small exceptional tilings.","A natural next test is to apply the same vertex-combination enumeration to rational-angle triangles at small denominators; the paper's f=36 example already indicates that new sporadic tilings will appear there, so the rational classification is expected to be richer."],"forward_implications":["If the theorem is correct, the irrational-angle branch of the classification is closed; any future example must lie in one of the three families of Table 3.","The explicit formulas in Table 3 turn the existence question into a check of angular data: a triangle with an irrational angle can tile non-side-to-side exactly when it fits one of the listed parameter ranges.","The 8-tile and 16-tile tilings are the only sporadic irrational tilings, so no further isolated examples can appear in this branch.","The matching-lemma and vertex-statistics scheme gives a finite-case template that the paper's final section proposes to extend to the remaining rational-angle case."],"supporting_citations":[{"why":"Supplies the vertex-distribution statistics and adjacent angle deduction that the paper extends to non-side-to-side tilings.","marker":"[22, 23]"},{"why":"Classifies all tilings by congruent isosceles triangles, letting the theorem restrict attention to convex scalene triangles.","marker":"[7, 8]"},{"why":"Classifies the known right-triangle non-side-to-side tilings that the new classification must subsume or exclude.","marker":"[9, 10, 11]"},{"why":"Provides the quadrilateral monotiles whose subdivision yields the 2k-tile two-layer earth map families in the theorem.","marker":"[16]"},{"why":"States the convex spherical triangle inequalities used as Lemma 5 throughout the angle and side-length arguments.","marker":"[1]"}],"fun_headline_variants":["Three families wrap up irrational tiling classification","All irrational-angle triangle tilings now classified","Non-side-to-side tilings: complete classification for irrational angles","Irrational triangle tiling puzzle fully solved"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that Lemma 8 lists every possible extended-edge pattern when the largest angle never sits on a half vertex; if that enumeration misses a pattern, a tiling outside the three families could survive the case analysis.","fun_headline_variants_meta":{"raw":{"variants":["Three families wrap up irrational tiling classification","All irrational-angle triangle tilings now classified","Non-side-to-side tilings: complete classification for irrational angles","Irrational triangle tiling puzzle fully solved"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00016,"raw_usage":{"total_tokens":1168,"prompt_tokens":817,"completion_tokens":351,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":433,"completion_tokens_details":{"reasoning_tokens":291}},"tokens_in":433,"tokens_out":351,"duration_ms":3297,"temperature":1.0,"reasoning_tokens":291,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T14:57:07.472030+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run an exhaustive search for non-side-to-side tilings of the sphere by congruent scalene triangles whose angles include an irrational number of degrees, for tile counts up to the bounds permitted by the counting identities; any output not congruent to a two-layer earth map tiling, the 8-tile tiling, or the 16-tile tiling, or any extended edge not listed in Table 5, would disprove the theorem.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the quadrilateral monotiles whose subdivision yields the 2k-tile two-layer earth map families in the theorem."}],"review_version":1}