{"id":"7adeb6b9-6023-48fc-9148-0eb0bbd4cf95","arxiv_id":"2507.23681","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For Sierpinski polygon graphs with r not a multiple of 4, isomorphism is classified by a dihedral group action on the defining sequence, and eventually-constant sequences yield exactly two Busemann and countably many non-Busemann horofunctions.","lead":"This paper defines a family of infinite graphs built by gluing Sierpinski polygon graphs and shows that the isomorphism type is governed by a dihedral group acting on the defining sequence. It also proves that for eventually constant sequences, the horofunction boundary consists of exactly two Busemann points plus countably many non-Busemann points, a concrete count in a non-hyperbolic setting.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3.15 rests on the informally defined k-hole geometry: the equidistance and cut-point properties of antipodal points are assumed in Definition 3.7, not proved, and drive the key cancellations in Lemma 3.12 and Proposition 3.13.","rationale":"The reader's weakest assumption identifies exactly the same load-bearing issue: the k-hole and antipodal-point geometry are introduced informally, and the equidistance and cut-point properties used in the proof of Theorem 3.15 are assumed rather than proved. My reading of the manuscript confirms this. Definition 3.6 explicitly omits the formal definition of a k-hole, and Definition 3.7 defines antipodal points through an equality that is then reused as the main computational engine in Lemma 3.12 and Proposition 3.13. Proposition 3.11's uniqueness argument for the two Busemann points also depends on the same cut-point structure. Because the missing definition is explicitly acknowledged in the text, this is an internal incompleteness rather than a mere stylistic preference. The concern is load-bearing but not obviously fatal: the claimed theorem may be true, and the computations in the paper suggest the authors have a concrete geometric picture in mind. A computational check on small r=6 examples would test the key identities directly; if they hold, the remaining issue is a proof obligation, not a counterexample. Thus I do not see grounds to move the verdict away from the reader's CONDITIONAL assessment.","tokens_in":10428,"tokens_out":14481,"duration_ms":152551,"concrete_test":"Generate Γ_m exactly for r=6, ξ=j^∞ with j=4, for m=2,3,4,5, and use BFS to compute all distances. Verify: (i) there exists a vertex p_m on the side of the m-hole opposite ξ with d(p_m,b_m)=d(p_m,\\bar b_m), and that this vertex is unique; (ii) for N<m, d(p_m,b_N)=d(p_m,eb_N) for the two cut vertices of Γ_N; and (iii) every path from p_m to a fixed vertex inside Γ_N passes through b_N or \\bar b_N. If any of these fails for a tested m, the equality chain in Lemma 3.12 breaks and Theorem 3.15 is unsupported; if all pass, the concern is reduced to a proof obligation rather than a counterexample.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Definition 3.6 explicitly leaves the k-hole undefined, stating that a formal definition 'is rather technical and not relevant for our purposes'. However, Definition 3.7 defines the antipodal point p_m by the equality d(b_m,p_m)=d(\\bar b_m,p_m), and Lemma 3.12's proof of weak geodesicity cancels terms using exactly this equality, together with the related identities d(p_N,b_N)=d(p_N,eb_N), d(p_m,b_m)=d(p_m,\\bar b_m), and d(b_m,\\bar b_m)=d(b_m,eb_m). Proposition 3.13 extends the same cancellation to shifted points via d(b_N,p_{m,t})=d(\\bar b_N,p_{m,t})+2|t|. Proposition 3.11 also relies on the cut-point assertion that removing b_m and \\bar b_m disconnects Γξ[m] and that every geodesic between the two antipodal copies of Γ_m passes through one of these vertices. This assertion is not derived from a formal definition of a hole. These are not cosmetic gaps: if the actual hole geometry yields a different distance relation at some scale, or if the asserted equidistant vertex does not exist uniquely, the cancellations in Lemma 3.12 and Proposition 3.13 become nonzero or invalid, and the claimed count of Busemann and non-Busemann points in Theorem 3.15 could change. Since the manuscript itself flags the omitted formal definition, the proof of the central boundary theorem is incomplete as written.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs infinite Sierpinski polygon graphs Γξ^(r) for r not divisible by 4, defined as strong Gromov–Hausdorff limits of recursively built finite graphs indexed by an infinite sequence ξ over r letters. It proves an isomorphism classification (Theorem 2.4): Γξ and Γη are isomorphic if and only if η is cofinal with σ(ξ) for some σ in the dihedral group D_r. The main new result (Theorem 3.15) claims that for eventually constant ξ = w j^∞, the horofunction boundary contains exactly two Busemann points and countably many non-Busemann points; the proof uses holes, cut points, antipodal sequences, and weakly-geodesic rays. A conjecture extends the description to general sequences.","tokens_in":10728,"tokens_out":5946,"duration_ms":57008,"significance":"If Theorem 3.15 is correct, the paper gives the first complete description of the horofunction boundary for this family of infinite self-similar graphs, complementing the earlier Sierpinski carpet work of D'Angeli and Donno. The isomorphism classification, Theorem 2.4, is clean, parameter-free, and appears self-contained, and the strong Gromov–Hausdorff setup is a useful framework for the later boundary analysis. The paper contains no machine-checked proofs, and its central boundary theorem rests on informally defined k-hole geometry that the authors explicitly leave unformalized; this is a genuine load-bearing gap. The use of [DD16, Proposition 4.2] for the non-Busemann criterion is legitimate external support rather than circular reasoning, since the main theorem does not reduce to an input parameter of the paper.","major_comments":[{"comment":"Definition 3.6 explicitly omits the formal definition of a k-hole, stating that it 'is rather technical and not relevant for our purposes.' Definition 3.7 then defines the antipodal point p_m by the equality d(b_m, p_m) = d(bbar_m, p_m), and Proposition 3.11 later asserts that removing the two gluing vertices b_m and bbar_m disconnects the graph and that every geodesic between antipodal copies passes through one of them. These existence, uniqueness, and cut-point properties are not proved from any formal definition. The cancellations in Lemma 3.12 (d(p_m,b_m)=d(p_m,bbar_m), d(b_m,bbar_m)=d(b_m,eb_m)) and Proposition 3.13 (d(b_N,p_{m,t})=d(bbar_N,p_{m,t})+2|t|) depend on exactly these relations, and they drive the proof that the antipodal sequence is weakly geodesic. As written, the proof of Theorem 3.15 is therefore incomplete unless the omitted definition is supplied or the geometric equalities are stated as explicit lemmas with proofs.","section":"Definitions 3.6–3.7 and Theorem 3.15"},{"comment":"The proof of Lemma 3.10 contains the unproved inequality d(b_N, ξ) < d(b_N, bbar_N) + d(bbar_N, ξ), introduced with the phrase 'one can get.' This inequality is used to show that certain sequences cannot represent horofunctions, and it underpins the trichotomy of weakly-geodesic rays stated immediately after the lemma. Since the trichotomy is used in Proposition 3.11 to conclude that there are exactly two Busemann points, a rigorous derivation of the inequality (or a precise reference) is needed.","section":"Lemma 3.10"},{"comment":"The proof of Proposition 3.11 asserts that a geodesic from γ(m_1) to γ(m_3) must pass through one of the cut points b_{m_2} or bbar_{m_2}, and then excludes bbar_{m_2} by the informal copy-counting argument: 'to reach bbar_{m_2}, one must either pass through b_{m_1} and then through f(r) copies, or pass through more than f(r)+1 copies of Γ_k.' Neither the disconnectivity statement nor the copy-counting estimate is derived from a formal definition of the hole geometry. These assertions are load-bearing for the 'exactly two Busemann points' part of Theorem 3.15, so they need to be proved or made into explicit geometric axioms.","section":"Proposition 3.11"},{"comment":"The non-Busemann part of Lemma 3.12 is delegated to '[DD16, Proposition 4.2]' with the note 'almost verbatim' and the identifications v_1 = b_N, v_2 = bbar_N, φ_y(x)=d_x(y). Proposition 3.13 makes the same delegation. The reader is not shown that the hypotheses of [DD16, Proposition 4.2] are satisfied by the sequences and cut points constructed here; in particular, the required distance equalities at all scales are exactly the unformalized k-hole relations. Please spell out the verification of the hypotheses or include the argument directly.","section":"Lemma 3.12 and Proposition 3.13"}],"minor_comments":[{"comment":"The word 'isomorphim' should be 'isomorphism'.","section":"Abstract"},{"comment":"In the odd-r case, the phrase 'the antipodal point p of a hole H in Γξ with respect to ξ is the gluing vertex of the two antipodal copies of the copy of Γm containing ξ' is hard to parse; please clarify which vertices are being glued and why the antipodal point is unique.","section":"Definition 3.7"},{"comment":"The notation T is used in the definition of P: T → Γξ before T is introduced; please define T explicitly as the unbounded subset of N parametrizing the antipodal sequence.","section":"Lemma 3.12"},{"comment":"The phrase 'if t1t2 < 0' should presumably read 'if t_1 t_2 < 0' with the product of the two shifts; please fix the typography for clarity.","section":"Proposition 3.13"},{"comment":"The statement says the boundary 'contains countably many non-Busemann points,' which is a lower bound. If the intended claim is that the non-Busemann subset is exactly countable, that stronger statement should be stated and proved explicitly.","section":"Theorem 3.15"}],"recommendation":"major_revision","confidential_remarks":"The main theorem is plausible and the manuscript is worth pursuing, but the reader's main stress-test concern lands precisely on the weakest point: the central proof depends on k-hole geometry that the authors themselves decline to formalize. This is a fixable issue and not grounds for rejection, provided the omitted definitions and the key equidistance/cut-point lemmas are supplied. I would also encourage the authors to clarify whether the count of non-Busemann points in Theorem 3.15 is exact or merely a lower bound."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing you should know: this is a genuine extension of the D'Angeli–Donno program, and the isomorphism classification (Theorem 2.4) is clean and probably correct. The horofunction count (Theorem 3.15) is the main new payoff, but the proof has a load-bearing informality: the k-hole is never formally defined, and the antipodal point used throughout is defined by a distance equality that is assumed rather than proved.\n\nThe paper does several things well. The GH limit construction from an infinite sequence over r letters is standard but carefully set up, and Lemma 2.2–2.3 give a short, believable proof that isomorphism of limit graphs is exactly dihedral action up to cofinality. That part reads well.\n\nThe soft spots are all in Section 3. Definition 3.6 explicitly says a formal definition of k-hole is 'rather technical and not relevant.' That sentence does a lot of work, because Definition 3.7 defines the antipodal point p_m by d(b_m,p_m)=d(b̄_m,p_m), and Lemma 3.12 cancels terms using exactly that identity, plus the asserted cut-point fact that every geodesic between two copies passes through one of the two gluing vertices. Proposition 3.11 uses the same geometry. The stress-test note is right that these are not cosmetic: if the actual hole geometry gives a different distance relation, the weak-geodesic check in Lemma 3.12 and the count in Theorem 3.15 could change. I would want a lemma that derives the equidistance and disconnect properties from an actual definition of the hole, for all r not divisible by 4. The symmetries of the case ξ = wj^∞ are plausible, but 'exploiting the symmetries' is an argument sketch, not a proof.\n\nThere are smaller issues: Lemma 3.10's 'one can get' inequality is underexplained, and the non-Busemann step is delegated to [DD16, Prop 4.2]. The latter is fine if the hypotheses match, but since it is the key negative result, the authors should make the application explicit.\n\nThe citation pattern is honest, and the self-overlap with [DD16] is not circular: the borrowed criterion is external and prior.\n\nWho is this for: people studying horofunction boundaries of self-similar graphs. It deserves a serious referee; I would not desk reject. The referee should require a formal definition of k-holes and a real proof of the equidistance/cut-point lemmas before acceptance. After that, I expect it to be a solid paper.","headline":"A real extension of the Sierpinski carpet program, with a clean isomorphism theorem, but the horofunction count rests on hole geometry the paper explicitly leaves informal.","tokens_in":11312,"tokens_out":2739,"would_cite":false,"duration_ms":27065,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05C12","05C63","20F65"],"pacs":[],"model":"deepseek-v4-flash","headline":"The horofunction boundary of an infinite Sierpinski polygon graph with an eventually constant defining sequence has exactly two Busemann points and countably many non-Busemann points.","keywords":["horofunction boundary","Busemann points","Sierpinski polygon graphs","Gromov–Hausdorff convergence","self-similar graphs","cut points","weakly-geodesic rays","dihedral group"],"falsifier":"For a concrete case such as $r=6$ and $\\xi = 4^\\infty$, compute at every level $m$ the distances $d(b_m, p_m)$, $d(\\bar{b}_m, p_m)$, and $d(b_m,\\bar{b}_m)$ in $\\Gamma^{(6)}_m$; if the asserted identity $d(b_m, p_m) = d(\\bar{b}_m, p_m)$ or the cut-point property fails at any $m$, the weakly-geodesic computation breaks and the count in Theorem 3.15 would change. A complementary check is to enumerate all horofunctions of this graph by the weakly-geodesic ray characterization and confirm that exactly the two described Busemann points appear.","tokens_in":10189,"feed_emoji":"🔺","tokens_out":21763,"duration_ms":177744,"temperature":0.7,"pith_summary":"This paper studies infinite Sierpinski polygon graphs, the limiting graphs obtained by recursively gluing $r$ copies of a polygon graph, for any $r$ not divisible by 4. It proves that two such limit graphs are isomorphic exactly when their defining infinite sequences agree up to a dihedral symmetry of the $r$-polygon and a finite prefix. It then analyzes the horofunction boundary, a standard compactification formed by limiting distance functions to points, and shows that for an eventually constant defining sequence the boundary contains exactly two Busemann points (limits of geodesic rays) and countably many non-Busemann points. The result generalizes earlier work on Sierpinski carpet and triangle graphs, where a similar dichotomy between Busemann and non-Busemann points was observed.","feed_headline":"Infinite Sierpinski polygon graphs have exactly two geodesic limits","feed_subtitle":"For eventually constant sequences, the horofunction boundary has two Busemann points and countably many others.","key_machinery":"The argument is carried by the self-similar geometry of holes and cut points in $\\Gamma^{(r)}_k$. The recursive gluing rule identifies vertices at offsets $f(r) = \\min\\{i : 4i > r\\}$ and $\\tilde{f}(r) = 2f(r)$, producing a central 'hole' at each scale; a $k$-hole is the subgraph surrounding that hole, and the antipodal point $p_m$ of the $m$-hole with respect to the base vertex $\\xi$ is the vertex of the opposite copy of $\\Gamma_m$ that satisfies $d(b_m, p_m) = d(\\bar{b}_m, p_m)$, where $b_m$ and $\\bar{b}_m$ are the two gluing vertices of the copy containing $\\xi$. This equidistance identity, together with the cut-point fact that any geodesic joining the two antipodal copies of $\\Gamma_m$ must pass through $b_m$ or $\\bar{b}_m$, is what makes the sequences $\\{p_m\\}$, $\\{b_m\\}$, and $\\{\\bar{b}_m\\}$ weakly-geodesic. The paper also uses the characterization of horofunctions as pointwise limits of weakly-geodesic rays (Theorem 3.3) to turn these geometric sequences into boundary points.","core_discovery":"The paper claims that the pointed Gromov–Hausdorff limits $\\Gamma^{(r)}_\\xi$ of the recursive Sierpinski polygon graphs $\\Gamma^{(r)}_k$, taken along an infinite sequence $\\xi$ over $r$ letters with $r$ not a multiple of 4, are classified up to isomorphism by the cofinality class of $\\xi$ under the action of the dihedral group $D_r$ (Theorem 2.4). For the boundary analysis, when $\\xi = w j^\\infty$ is eventually constant, the horofunction boundary $\\partial_h \\Gamma^{(r)}_\\xi$ contains exactly two Busemann points, realised by the two symmetric geodesic rays that run along the sequences of gluing vertices $\\{b_m\\}$ and $\\{\\bar{b}_m\\}$, and it contains countably many non-Busemann points, realised by the antipodal sequence $\\{p_m\\}$ of points on the central holes together with its integer shifts (Theorem 3.15). The non-Busemann points are limits of weakly-geodesic rays but not of almost-geodesic rays, so they sit strictly outside the Busemann subset.","pith_inferences":["The theorem is stated for eventually constant sequences, but the machinery of antipodal points is built for the 'grows away from $j$' condition; a natural next step, already anticipated by the paper's Conjecture 3.16, is that for sequences containing other letters infinitely often the boundary should still have exactly two Busemann points, with the number and type of non-Busemann families governed","The isomorphism classification identifies limit graphs up to cofinality under $D_r$; because the paper explicitly links the construction to Schreier graphs of self-similar groups, the horofunction boundary described here may coincide with the boundary of the corresponding Schreier graph, which would transfer the counting result to a group-theoretic setting.","The paper leaves open whether the word 'contains' in Theorem 3.15 can be strengthened to 'consists of'—that is, whether the two Busemann points and the constructed countable family exhaust the entire horofunction boundary; a positive answer would make the boundary countable, while a negative one would reveal additional non-Busemann structure."],"forward_implications":["If Theorem 2.4 is correct, deciding whether two Sierpinski polygon limit graphs are isomorphic reduces to checking whether one defining sequence can be transformed into the other by a dihedral symmetry followed by a change on finitely many initial letters.","If Theorem 3.15 is correct, then for every eventually constant sequence the geodesic-ray part of the horofunction boundary is completely understood: there are exactly two Busemann points, one for each of the two symmetric gluing-vertex sequences.","The antipodal sequences and their shifts provide an explicit, enumerable family of non-Busemann horofunctions, so the non-Busemann part of the boundary is at least countably infinite in these graphs.","The same cut-point argument shows that any geodesic ray from $\\xi$ must eventually pass through the sequence $\\{b_m\\}$ or through $\\{\\bar{b}_m\\}$, so the two Busemann points constructed are the only possible ones."],"supporting_citations":[{"why":"Introduces the horofunction boundary and Busemann points, the central objects of Section 3.","marker":"[Gro81]"},{"why":"Provides the theorem that every horofunction is a pointwise limit of a weakly-geodesic ray, used to identify boundary points from geometric sequences.","marker":"[Rie02]"},{"why":"Prior work on horofunctions of Sierpinski triangle graphs that this paper generalizes.","marker":"[DAn17]"},{"why":"Contains Proposition 4.2, whose proof is adapted to show the antipodal sequences are non-Busemann points.","marker":"[DD16]"},{"why":"Supplies the analogy with Schreier graphs of self-similar groups used to justify the limit graph construction and the cofinality lemma.","marker":"[BDN17]"},{"why":"Provides the definition of pointed Gromov–Hausdorff convergence used to construct the limit graphs.","marker":"[BBI01]"}],"fun_headline_variants":["Two Busemann points, countably many other horofunctions","Sierpinski polygon limits: dihedral cofinality is the key","Exactly two geodesic rays govern the horofunction boundary","Infinite Sierpinski graphs: two symmetric geodesic limits","Horofunctions of infinite Sierpinski polygons: two plus countable"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The equidistance and cut-point properties of the antipodal point $p_m$ for the $m$-hole are stated informally and rely on an omitted formal definition of a $k$-hole; if the formal definition produced different distances, the count of non-Busemann points in Theorem 3.15 would change.","fun_headline_variants_meta":{"raw":{"variants":["Two Busemann points, countably many other horofunctions","Sierpinski polygon limits: dihedral cofinality is the key","Exactly two geodesic rays govern the horofunction boundary","Infinite Sierpinski graphs: two symmetric geodesic limits","Horofunctions of infinite Sierpinski polygons: two plus countable"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000274,"raw_usage":{"total_tokens":1592,"prompt_tokens":851,"completion_tokens":741,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":467,"completion_tokens_details":{"reasoning_tokens":653}},"tokens_in":467,"tokens_out":741,"duration_ms":6859,"temperature":1.0,"reasoning_tokens":653,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T10:27:50.718192+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a concrete case such as $r=6$ and $\\xi = 4^\\infty$, compute at every level $m$ the distances $d(b_m, p_m)$, $d(\\bar{b}_m, p_m)$, and $d(b_m,\\bar{b}_m)$ in $\\Gamma^{(6)}_m$; if the asserted identity $d(b_m, p_m) = d(\\bar{b}_m, p_m)$ or the cut-point property fails at any $m$, the weakly-geodesic computation breaks and the count in Theorem 3.15 would change. A complementary check is to enumerate all horofunctions of this graph by the weakly-geodesic ray characterization and confirm that exactly the two described Busemann points appear.","supporting_citations":[],"review_version":1}