{"id":"1d8a60cc-86e4-4351-8bad-80cf83cce0ed","arxiv_id":"2607.21348","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Penrose rhombic tilings admit a four-valued vertex frieze pattern and Godrèche–Lançon–Billard tilings admit a three-valued one, both satisfying the diamond rule bc−ad=1.","lead":"This short note puts positive whole numbers on the corners of two famous never-repeating floor tilings so that every diamond-shaped tile obeys the same multiplication rule as a Conway–Coxeter frieze. It gives the first examples of frieze patterns whose underlying grid is aperiodic rather than periodic.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"GLB labeling is under-specified: interior rhombi of a supertile can have all vertices at distance ≥2 from every supervertex, hence all label 3, violating bc−ad=1. The paper's one-sentence consistency claim does not rule this out.","rationale":"The reader's weakest_assumption identified the GLB construction's dependence on unique decomposition and distance-based labeling, which is the right area. However, the specific failure mode I see is more basic: the proposed labeling can assign label 3 to every vertex of an interior rhombus, making the diamond rule fail outright. The reader's phrasing focused on ambiguity ('no vertex is both one edge from a supervertex and an interior vertex'), but the real danger is vertices that are neither supervertices nor adjacent to supervertices. The paper's claim that consistency follows from the supertile decomposition is not supported by the stated 'three simple steps': those steps address supervertices and midpoints of superedges, but not rhombi in the interior of a supertile. The Penrose construction, by contrast, is a finite check over two tile shapes and ten orientations, and the asserted patterns do satisfy the diamond rule; that part is plausible and could be verified mechanically. The GLB theorem, however, is the more novel and less standard claim, and the informal proof leaves a concrete gap. A computational patch check would settle whether every rhombus in the GLB substitution contains a vertex labeled 1 or 2 in the required 1,2,2,3 configuration. Until that check is reported (or an analytical argument is given that all interior vertices are within distance 1 of a supervertex), acceptance of Theorem 3.1 is premature. Hence I recommend CONDITIONAL rather than ACCEPT.","tokens_in":6753,"tokens_out":16247,"duration_ms":168406,"concrete_test":"Code up the GLB substitution rule as in [15] (or use the Tilings Encyclopedia [14]) and generate a patch at least a few supertiles across. Mark the level-1 supertile vertices as 1; mark every vertex at graph distance 1 from a 1 as 2; mark all remaining vertices 3. Then iterate over all rhombi in the patch and verify bc−ad=1 with the cyclic order determined by the black-point orientation from Fig. 7. Report the number of rhombi whose four labels are (3,3,3,3) or otherwise fail. If any such rhombus occurs, Theorem 3.1 is false; if none occurs in a large enough patch, run a finite atlas check over the GLB substitution to prove it for all patches.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"In §3, the labeling rule is: supertile vertices → 1; vertices one edge away from a 1 → 2; all others → 3. The proof then claims the diamond rule holds on every rhombus because each rhombus lies in exactly one supertile. This does not follow. For a rhombus in the interior of a supertile, none of its four vertices need be a supertile vertex or adjacent to one; all four would receive label 3, so bc−ad = 3·3−3·3 = 0, violating the diamond rule. The 'three simple steps' only discuss the midpoint of a superedge; they say nothing about interior rhombi or about rhombi near a superedge away from the midpoint. The unique decomposition property ([21]) guarantees a unique grouping into supertiles but imposes no lower bound on distance from interior vertices to supertile corners. Thus Theorem 3.1 rests on an unverified local-configuration claim: every rhombus in the GLB substitution atlas must contain at least one vertex that is a supervertex or adjacent to one, arranged as 1,2,2,3. This is exactly the sort of claim that needs a patch check.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces frieze patterns supported on aperiodic rhombic tilings, i.e. positive-integer decorations of the vertices so that every rhombic tile satisfies the diamond rule bc−ad=1. It gives two constructions. For every Penrose rhombic tiling, a function r defined from A*_4 coordinates labels vertices by residues mod 5; each tile is claimed to have exactly one shallow hole, and the resulting local patterns are {1,2,2,3} or {2,3,3,4}, yielding a frieze pattern with four values (Theorem 2.1). For the Godrèche–Lançon–Billard tiling, the unique decomposition into supertiles is used: supervertices are labelled 1, vertices one edge away from a supervertex are labelled 2, and all remaining vertices are labelled 3; the paper claims this gives a frieze pattern with three values (Theorem 3.1). Corollaries assert infinitely many friezes by shifting the numerical labels.","tokens_in":7117,"tokens_out":12070,"duration_ms":126306,"significance":"If established, these are the first frieze patterns on aperiodic rhombic tilings, connecting Conway–Coxeter friezes with aperiodic order. The Penrose construction is explicit and parameter-free, and the GLB construction attempts to use the hierarchical decomposition in a genuinely new way. The paper is short and readable, and the Penrose part is plausible. However, the GLB proof is incomplete at a load-bearing point: the argument does not verify the diamond rule for rhombi in the interior of a supertile. Because the central construction is likely correct and the gap is repairable by a finite local check, the appropriate outcome is a major revision rather than acceptance as written.","major_comments":[{"comment":"The proof of Theorem 3.1 does not rule out rhombi whose four vertices all receive label 3. With the stated rule (supervertices = 1, vertices one edge from a supervertex = 2, remaining = 3), a rhombus in the interior of a supertile can have all its vertices at graph distance at least 2 from every supervertex. For such a rhombus the diamond rule gives 3·3−3·3 = 0, not 1. The unique decomposition property cited from [21] guarantees a unique grouping into supertiles but imposes no lower bound on the distance from interior vertices to supertile corners. The 'three simple steps' only discuss supervertices and the midpoint of a super-edge; they say nothing about interior rhombi or about rhombi elsewhere along a super-edge. The authors need to supply a finite verification, for example from the GLB substitution atlas, that every rhombus inside a supertile contains a supervertex or a vertex adjace","section":"§3, paragraph 'Since the GLB tiling can be grouped into supertiles...'"},{"comment":"The proof of Theorem 2.1 rests on the assertion that every tile has exactly one shallow hole and on the enumeration of possible decorations for thick and thin rhombuses. This is a finite check, but the paper only says it is easy. Because this local verification is the entire content of the Penrose theorem, the authors should list the possible residue patterns around each tile type, or give a precise citation that includes this statement. In addition, the definition of r(v)=Σ n_i mod 5 assumes that the representation v=Σ n_iπ(a_i) gives a well-defined residue; since the π(a_i) are linearly dependent in R^2, a sentence explaining uniqueness of the lift (or a reference to [18]) is needed.","section":"§2, 'One can easily check...' and Figure 4"}],"minor_comments":[{"comment":"The phrase 'n may as well be an arbitrary real number' conflicts with the definition of a frieze pattern as positive integers. Either restrict n to positive integers or explicitly extend the definition to real labels.","section":"Corollaries 2.2 and 3.2"},{"comment":"The text says 'This labelling is indicated in Figure 7', but the figure caption describes only black points and arrows. Please provide a labelled version or a table showing the label pattern for each tile type within a supertile.","section":"Figure 7 and surrounding text"},{"comment":"The notion of a frieze pattern on a rhombic tiling is clear from context, but it would help to state formally: a vertex labelling such that every tile's opposite vertex pairs (a,d) and (b,c) satisfy bc−ad=1. The diamond rule orientation should be specified once in full generality.","section":"§1, definition of frieze patterns on tilings"},{"comment":"The paragraph on the A*_4 construction is terse. The statement 'Since V∈A*_4\\A_4, we always have Σn_i not≡0 mod 5' is a known fact, but the exact page reference [18] could be expanded to a theorem statement so the reader does not have to consult the source.","section":"§2, notation"}],"recommendation":"major_revision","confidential_remarks":"The reader's accept recommendation is understandable: the ideas are appealing, the Penrose construction is essentially sound, and no counterexample was found. My concern is that the GLB proof, as written, has a genuine gap: the diamond rule is not shown for rhombi whose four vertices are all at distance at least two from every supervertex. This is load-bearing for Theorem 3.1, but it is likely fixable by a finite check of the substitution atlas. I recommend major revision and would be happy to accept once that check is supplied."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nHere's the short version: the Penrose construction in this note is real and new; the GLB construction is not proved as written. The stress-test note is right to worry.\n\nThe genuinely good part is Theorem 2.1. Using the de Bruijn index r(v) mod 5 from the A4 dualisation, the authors get a global label on every Penrose vertex, r never vanishes, every tile has exactly one shallow hole, and the local decorations are forced into the shapes 1,2,2,3 and 2,3,3,4 (or rotated). Each satisfies bc−ad=1. The \"one can easily check\" is a bit terse, but the check is small and the claim is credible. The infinite family via scaling is trivial but harmless.\n\nThe GLB part is the soft spot. The labeling rule is: supertile vertices get 1, their edge-neighbors get 2, everyone else gets 3. The proof that the diamond rule holds says that since each rhombus lies in one supertile, the rule is immediate. That is not immediate. A rhombus in the interior of a supertile can have all four vertices at distance ≥2 from every supertile vertex; all four would be labeled 3, and the diamond rule would give 9−9=0, not 1. The 'three simple steps' only discuss supervertices, midpoints of superedges, and the consistency of label 3 for vertices contained in the interior of a supertile. They never verify that an interior rhombus has a vertex labeled 1 or 2, let alone the right arrangement. Unique composition ([21]) gives a unique grouping, but no bound on interior distances to supertile corners. So Theorem 3.1 is currently unsupported. It might be true and fixable, but the argument as it stands doesn't carry it.\n\nThe reader's 7/10 soundness score is a bit generous for the GLB section; the Penrose section alone would deserve it. Significance and novelty scores are fair. Citation pattern is fine: de Bruijn and Solomyak are the right sources, and the new connection to [13] is a real lead.\n\nWho is this for? Someone working in frieze combinatorics or aperiodic tilings will care about the Penrose example. The GLB gap makes the paper a discussable preprint rather than a finished one.\n\nRecommendation: send it to peer review, but make the referee focus on the GLB labeling. The referee should demand a full check of the substitution atlas or a corrected labeling. If that gets fixed, it's a nice short contribution.","headline":"Penrose construction is new and sound; the GLB proof has a real gap and needs referee scrutiny.","tokens_in":7516,"tokens_out":8461,"would_cite":true,"duration_ms":82220,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05B45","52C23"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper establishes that frieze patterns—positive-integer vertex labellings obeying the diamond rule bc−ad=1 on every rhombus—can be placed on two famous aperiodic tilings: the Penrose rhombic tiling uses only four distinct values and the","keywords":["frieze patterns","diamond rule","Penrose rhombic tiling","Godrèche–Lançon–Billard tiling","aperiodic tilings","Conway–Coxeter friezes","supertile decomposition","substitution tilings"],"falsifier":"Enumerate all rhombi in a large patch of a GLB tiling and check the local rule: every rhombus must carry labels of the form {n, n+1, n+1, n+2} with the smallest and largest labels opposite each other; a single rhombus with the smallest and largest labels adjacent would violate the diamond rule. For the Penrose tiling, compute the residues mod 5 from pentagrid coordinates of every vertex in a finite patch; finding a vertex with residue 0, or a tile with zero or two shallow holes, would disprove the construction.","tokens_in":6684,"feed_emoji":"🔷","tokens_out":9777,"duration_ms":93993,"temperature":0.7,"pith_summary":"The paper tries to establish that frieze patterns, arrays of positive integers satisfying the diamond rule bc−ad=1 on every elementary rhombus, can live on aperiodic tilings rather than only on periodic rhombic grids. Its main theorems are explicit: every Penrose rhombic tiling carries a vertex labelling with only four values, and every Godrèche–Lançon–Billard tiling carries one with only three values. The Penrose construction comes from a dual-lattice projection and a residue modulo 5 that never vanishes, forcing each tile into one of two label decorations; the GLB construction comes from the tiling's unique decomposition into supertiles, with labels 1, 2, and 3 assigned to supertile corners, edge-neighbours, and interiors. In both cases, the identity (n+1)²−n(n+2)=1 yields infinitely many variants. A sympathetic reader would care because this opens frieze patterns, long tied to polygon triangulations and periodic geometry, as a new invariant of aperiodic order.","feed_headline":"Aperiodic tilings carry frieze patterns with just 3 or 4 labels","feed_subtitle":"The diamond rule of classical friezes holds on Penrose and GLB tilings—new territory for these number arrays.","key_machinery":"The load-bearing identity is the diamond rule itself, bc−ad=1, together with the algebraic fact (n+1)²−n(n+2)=1, which makes any rhombus decorated with values n, n+1, n+1, n+2 satisfy the rule. For the Penrose tiling, the mechanism is a residue function r(v)=Σnᵢ mod 5 on the dual lattice A₄*, with vertices never having residue 0 and every tile containing exactly one shallow hole; this forces each thick or thin rhombus into one of two allowed label patterns. For the GLB tiling, the mechanism is the unique decomposition into supertiles, which yields a global 1/2/3 classification of vertices with no ambiguity across supertile boundaries. In both cases the argument is local: once the finite list","core_discovery":"On the Penrose rhombic tiling, every vertex can be expressed as an integer combination of four dual-lattice generators, and reducing the sum of coefficients modulo 5 gives a value r(v) that is never zero, hence takes only the values 1, 2, 3, 4. Each rhombus has exactly one shallow-hole vertex, and when a tile is oriented so that this vertex plays the role of a in the diamond rule, the four labels are always of the form {n, n+1, n+2} with the shallow hole at n, for n=1 or 2; the two possible decorations per rhombus type are checked directly and satisfy bc−ad=1. On the GLB tiling, the unique supertile decomposition, guaranteed by the primitive substitution rule, gives a globally consistent lab","pith_inferences":["Since two distinct positive integers a