{"id":"ca338026-3683-4242-93ea-ce60e69604f0","arxiv_id":"2411.16283","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Infinite-type cluster algebras have G-fans that are never complete, and rank 3 local behavior falls into six types that correlate with global fan shapes.","lead":"This paper proves that the g-vector fans attached to infinite-type cluster algebras are always incomplete, leaving gaps that no new cones fill. It also sorts the rank 3 infinite-type cases into six local patterns and links them to experimentally observed global shapes.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 3.1's N=0 cases for Types 4-2 and 4-3 are asserted but not proved: the sign-pattern tables and recursions are ill-defined for N=0, leaving a gap in Theorem 4.2.","rationale":"The reader's conditional verdict identified reliance on the sign-pattern classification of [GN22] and the exclusion of N=0 cases as the weakest assumption. My stress-test sharpens this into a specific internal gap: the paper claims to include N=0, but the sign-pattern tables in Types 4-2-2 and 4-3-2 become self-contradictory for N=0 (the listed t-values repeat), and the recursion used to compute tilde g_{N+2} invokes the undefined vector tilde g_0. The N=0 families are nonempty (e.g., c0=-1, d0=3 with a=3,b=2) and are exactly the cases the paper says were wrongly excluded in [GN22]. Since Proposition 3.1 is the load-bearing input for Theorem 4.2, this is a genuine proof gap. However, the gap is likely repairable: spot checks with the mutation formulas for one N=0 Type 4-2 example gave tilde g_m = (g_m,0) for m >= 2, consistent with the claimed conclusion. Thus the appropriate verdict remains conditional: the paper's central claim is plausible and no counterexample has been found, but the N=0 cases must be verified or reproved before Theorem 4.2 is fully established. I did not find a more serious objection in the geometry of the incompleteness argument itself, assuming Proposition 3.1 holds.","tokens_in":25917,"tokens_out":29996,"duration_ms":258049,"concrete_test":"Implement the mutation formulas (2.1)-(2.3) in exact integer arithmetic and, for fixed (a,b) with ab >= 5 and c0 < 0 < d0 satisfying (3.41) with N=0 (e.g., a=3,b=2,c0=-1,d0=3) and (3.59) with N=0, compute tilde g_m for m up to about 30. Verify the claimed closed forms: for Type 4-2, the third component is 0 for all m >= 2; for Type 4-3, the third component equals c0 * alpha_m + (d0 + b*c0) * beta_m for all m >= 2. If both hold, the N=0 gap is a proof defect rather than a counterexample; if either fails, Proposition 3.1 is false and Theorem 4.2 is unproved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 4.2 rests on Proposition 3.1, whose proof for Types 4-2 and 4-3 (c0 < 0 < d0) splits by the index N in (3.41)/(3.59). The paper explicitly includes N=0 ('should have been included'), but the proof for N=0 is not a valid specialization. In Type 4-2-2 the sign-pattern table (3.51) lists t = -2,-1,0,1,N,N+1,N+2,N+3; for N=0 these t-values overlap (0 and 1 appear twice), so no sign pattern is actually asserted. The recursion used, tilde g_{N+2} = -tilde g_N + b tilde g_{N+1} + d_{N+1} e3, requires N >= 2 because the sequence g_m in (2.26) is only defined for m >= 1; for N=0 it refers to the undefined vector tilde g_0. Type 4-3-2 has the same defect via (3.68) and (3.73). Thus the N=0 subcases, corresponding to integer families such as c0=-1, d0 >= b for Type 4-2, are not proved. If Proposition 3.1 fails on one of these families, the finite-limit claim and therefore the incompleteness proof collapse.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies G-fans (g-vector fans) of cluster patterns whose initial exchange matrices are of infinite type. In the first part, it reduces the study of alternating mutations for two indices (i,j) with |b_ij b_ji| ≥ 4 to the rank 3 case, classifies the asymptotic behavior of g-vectors into six types (Types 1, 2, 3, 4-1, 4-2, 4-3) with subcases, and then proves as an application that the G-fan of an infinite-type cluster pattern is incomplete (Theorem 4.2). The second part presents experimental observations on rank 3 G-fans of totally-infinite type, assigning to each vertex a local type and proposing prototypical global patterns (the wing, pinwheel, tunnel, gates, and their degenerations), many of which are explicitly computed. The paper is clearly written, contains extensive worked examples, and explicitly labels the global-pattern census as experimental.","tokens_in":1482,"tokens_out":1557,"duration_ms":93562,"significance":"If the proof of Proposition 3.1 is completed, Theorem 4.2 is a natural and valuable counterpart to the known completeness theorem for finite type, and the rank 3 local classification is a useful quantitative contribution. The paper should be credited for making the central derivation explicit (Section 3), for including hand-checked examples for each type, and for honestly marking the global-pattern classification as experimental rather than as a theorem. The explicit examples of global patterns (e.g., the tunnel, the outside/inside gates) are likely to be useful to the cluster-algebra community. However, the main theorem is load-bearing on Proposition 3.1, and that proposition has a proof gap in the N=0 subcases, which is discussed below.","major_comments":[{"comment":"The proof of Proposition 3.1 does not cover the N=0 subcases in Type 4-2-2 and Type 4-3-2. In Type 4-2-2, the sign-pattern table (3.51) lists t-values -2,-1,0,1,N,N+1,N+2,N+3; for N=0 these values overlap (0 and 1 appear twice), so no sign pattern is actually asserted. Moreover, the recursion used to derive \\tilde{g}_{N+2} just before (3.56) involves \\tilde{g}_N and \\tilde{g}_{N+1}; for N=0, \\tilde{g}_0 is undefined because the sequences g_m and \\tilde{g}_m in (2.26) and (3.36) start at m=1. The same defect occurs in Type 4-3-2 at (3.68) and (3.73), where \\tilde{g}'_{N+2} requires \\tilde{g}'_0. These N=0 cases are not vacuous: for example, (c0,d0)=(-1,b) satisfies the Type 4-2 condition (3.41) with N=0. Since Proposition 3.1 is the entire basis for Theorem 4.2 (and for the affine crack argument when ab=4), this gap leaves the incompleteness theorem unproved for these families. The author explicitly notes that N=0 was excluded in [GN22] and should have been included, but the present manuscript does not supply the missing argument. A dedicated treatment of N=0 (by direct computation or by a careful limiting argument) is required.","section":"Section 3.4.2, Type 4-2-2 and Section 3.4.3, Type 4-3-2"}],"minor_comments":[{"comment":"In the proof of Lemma 6.4(2), the subscripts in (6.14) and the following inequality appear to be misprinted: 'p1p'_2' should likely be 'p1p'_1' (the second inequality in (6.14) involves the pair (1,2) for vertex v3, so the product is p1p'_1, not p1p'_2).","section":"Section 6.3, Lemma 6.4(2)"},{"comment":"In the (C-4) row, 'the wide intside gate' is a typo for 'the wide inside gate'.","section":"Table 1"},{"comment":"The phrase 'by the mutation sequence (3,1,2) in in (6.19)' contains a duplicated 'in'.","section":"Figure 14 caption"},{"comment":"The term 'the Badlands' is used without definition; a parenthetical explanation (the uncovered two-dimensional region between the limiting rays v and v') would improve readability.","section":"Section 2.4 and Section 5(4)"},{"comment":"For N=0, the right-hand side of (3.41) and the left-hand side of (3.59) involve division by U_{-1}=0. The text says to ignore the second inequality for N=0, but the phrasing could be expanded to clarify that the undefined ratio is simply not used in that case.","section":"Equations (3.41) and (3.59)"}],"recommendation":"major_revision","confidential_remarks":"The author relies heavily on [GN22], a paper coauthored by the present author, for the sign-pattern classification that underpins Proposition 3.1. The manuscript itself identifies an omission in [GN22] (the N=0 cases) but does not prove those cases here. In my view, this is a substantive gap that the author must repair in this paper, not defer to a future correction. Once the N=0 subcases are handled, the main theorem is likely sound and the experimental global-pattern census is a welcome addition, though the experimental part is not a theorem and should not be judged as one. I recommend major revision to address the N=0 gap and the minor issues above."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read it. The paper does something genuinely useful: a six-way local classification of rank-3 g-vector behavior, explicit formulas for each type, and a direct proof of the known incompleteness theorem that avoids the heavy scattering-diagram machinery. The global-pattern part (wing, pinwheel, gates, tunnel) is honestly labeled experimental, and the pictures make the phenomena concrete. That is real value.\n\nThe soft spot is Proposition 3.1, and the stress-test note lands. For Type 4-2-2 and Type 4-3-2, the paper includes N=0, correctly noting that [GN22] omitted it, but the sign-pattern tables (3.51) and (3.68) list the t-columns -2,-1,0,1,N,N+1,N+2,N+3. For N=0, the entries 0 and 1 appear twice, so no sign pattern is actually asserted there. Worse, the recursion in both cases uses tilde g_{N+2} = -tilde g_N + ...; for N=0 this needs tilde g_0, which is never defined (the sequence g_m starts at m=1). So the displayed computations for the N=0 families do not follow as written. The claim may well be true, and I expect a direct check of those integer families (e.g., c0=-1, d0 >= b) would repair it, but as it stands Theorem 4.2 has a gap for those cases.\n\nThe reliance on [GN22] is worth flagging but is not a flaw by itself: the sign-pattern classification is the key external input, and the paper openly says it is correcting the N=0 omission. Self-citation is defensible here because [Nak23] is the natural background reference. Section 6 is explicitly experimental, so I don't hold its conjectures to proof standard.\n\nVerdict: the paper deserves a serious referee, but it should not be accepted until the N=0 cases in Proposition 3.1 are actually proved or the argument is restructured to avoid them. Once that gap is filled, the paper is solid. I would not cite it in the next year until that happens.","headline":"Worth refereeing despite a genuine gap in the N=0 cases of Proposition 3.1; the local classification is solid.","tokens_in":92,"tokens_out":4879,"would_cite":false,"duration_ms":280028,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["13F60"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every cluster pattern of infinite type has an incomplete G-fan, and rank-3 local patterns predict its global geometry.","keywords":["cluster algebra","G-fan","g-vector","infinite type","totally-infinite type","rank 3 classification","Markov constant","incompleteness"],"falsifier":"Compute, for the paper's own Type 4-2 example with $(a,b,c_0,d_0)=(3,2,-100,159)$, the complete mutation sequence for $g$-vectors and check whether the third component stays identically zero for every $m\\ge 5$ as claimed; any later nonzero third component before the normalization limit, or any divergence of the normalized sequence, would falsify Proposition 3.1 and hence Theorem 4.2.","tokens_in":25698,"feed_emoji":"🌀","tokens_out":7349,"duration_ms":64592,"temperature":0.7,"pith_summary":"The paper establishes that a cluster pattern of infinite type always has an incomplete $G$-fan: some directions in $\\mathbb{R}^n$ are never covered by any $G$-cone. The proof works by looking at alternating mutations of a pair of indices whose exchange-matrix entry satisfies $|b_{ij}b_{ji}|\\ge 4$, showing the rescaled $g$-vectors converge to two finite limiting vectors, and observing that the codimension-one cones these limits span are left unfilled. In rank 3 the same analysis gives a six-type classification of the local behavior around a ray, determined by the signs and ratios of the initial off-diagonal entries. These local types, together with the Markov constant, are then matched to named global patterns of the fan, several of which are new. A direct corollary is that completeness of the $G$-fan characterizes exactly the finite-type cluster patterns.","feed_headline":"Every infinite-type cluster pattern has an incomplete G-fan","feed_subtitle":"Rank-3 local types predict the fan's global shape, from pinwheels to tunnels and gates.","key_machinery":"The central object is the $G$-fan: the fan in $\\mathbb{R}^n$ whose cones are spanned by the $g$-vectors attached to each seed of a cluster pattern. The argument is carried by the alternating-mutation sequences for two indices of infinite type, along which the $g$-vectors are expressed through Chebyshev polynomials of the second kind evaluated at $\\kappa/2=\\sqrt{ab}/2$. The sign pattern of the evolving third-row entries $(c_t,d_t)$ -- classified into six types (1, 2, 3, 4-1, 4-2, 4-3) -- decides whether the third component of the $g$-vectors is zero, a linear function of the rank-2 components, or a combination that cancels after finitely many steps; this is what makes the limits $\\tilde v$ and $\\tilde v'$ finite and computable.","core_discovery":"The paper's main theorem (Theorem 4.2) states: if the cluster pattern $\\Sigma(B)$ is of infinite type, the $G$-fan $\\Delta(B)$ is incomplete. The proof reduces to rank 3, where the initial exchange matrix has an infinite-type pair and two further entries $c_0,d_0$. Under alternating mutations in that pair, the normalized forward and backward $g$-vectors converge to vectors $\\tilde v$ and $\\tilde v'$ (Proposition 3.1), and the cones $\\sigma(\\tilde v,e_3,\\ldots,e_n)$, $\\sigma(\\tilde v',e_3,\\ldots,e_n)$ form a boundary that no full-dimensional $G$-cone crosses: in the affine case ($ab=4$) the two boundary vectors coincide, leaving a codimension-one crack, and in the non-affine case ($ab\\ge5$) they enclose an unfilled region because their normal vectors are irrational. Thus the $G$-fan cannot be complete.","pith_inferences":["If the proposed local-to-global dictionary is complete, the Markov constant acts as a phase parameter for the global shape: at or below 4 the fan stays in the pinwheel class, while above 4 it opens a tunnel or gates; this is testable by enumerating all rank-3 totally-infinite matrices with a small bound on entries.","The reduction to rank 3 suggests the six-type classification is the universal building block for the asymptotic behavior of any infinite-type pair in higher rank; one could seek higher-rank analogues of the crack and Badlands by tracking the third-row signs along the same alternating paths.","The equality cases in the Type 4-2 and Type 4-3 inequalities, which the paper calls finite degenerations, are natural candidates for a complete description of all degenerations; verifying that no other degenerations occur would strengthen the proposed exhaustion of global patterns."],"forward_implications":["Theorems 4.1 and 4.2 together give a dichotomy: a cluster pattern is of finite type if and only if its $G$-fan is complete.","In rank 3, the local type of each elementary vertex (1, 2, 3, 4-1, 4-2, 4-3) together with the Markov constant predicts the global pattern, yielding the named prototypes: the wing, the pinwheel, the tunnel, the outside and inside gates, and the dual gates.","For totally-infinite cyclic matrices with Markov constant at most 4, all elementary vertices are of Type 4-1 and the global pattern is the pinwheel; for larger Markov constant the tunnel appears.","The proof gives a direct route to incompleteness that uses only mutation formulas and sign-coherence, without invoking the uniqueness of consistent scattering diagrams."],"supporting_citations":[{"why":"Supplies the sign-pattern classification of $(c_t,d_t)$ for rank-3 alternating mutations, which is the load-bearing input for Proposition 3.1.","marker":"[GN22]"},{"why":"Gives the finite-type classification and the fact that an infinite-type pattern has a mutation-equivalent matrix with $|b_{ij}b_{ji}|\\ge 4$, the starting reduction.","marker":"[FZ03]"},{"why":"Provides the rank-2 g-vector formulas via Chebyshev polynomials and the completeness theorem for finite type that Theorem 4.2 complements.","marker":"[Rea14]"},{"why":"Establishes the tropical duality $G_t = D^{-1}(C_t^T)^{-1}D$ and unimodularity used to define $G$-cones.","marker":"[NZ12]"},{"why":"Provides column sign-coherence of C-matrices, ensuring each G-matrix is unimodular and the G-fan is well defined.","marker":"[GHKK18]"}],"fun_headline_variants":["Infinite-type cluster patterns always yield incomplete G-fans","Rank-3 G-fans of infinite type are never complete","Incomplete G-fans: universal for infinite-type cluster patterns","Why infinite-type G-fans always leave gaps","Local rank-3 patterns foretell global G-fan gaps"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything rests on the correctness of the listed sign patterns for the evolving off-diagonal coefficients along the alternating mutation path -- the paper itself notes the $N=0$ cases were missing from the earlier version of that list -- and if any of those sign patterns is wrong, the finite limiting vectors that create the uncovered boundary need not exist.","fun_headline_variants_meta":{"raw":{"variants":["Infinite-type cluster patterns always yield incomplete G-fans","Rank-3 G-fans of infinite type are never complete","Incomplete G-fans: universal for infinite-type cluster patterns","Why infinite-type G-fans always leave gaps","Local rank-3 patterns foretell global G-fan gaps"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000571,"raw_usage":{"total_tokens":2672,"prompt_tokens":889,"completion_tokens":1783,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":505,"completion_tokens_details":{"reasoning_tokens":1701}},"tokens_in":505,"tokens_out":1783,"duration_ms":12553,"temperature":1.0,"reasoning_tokens":1701,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T13:16:53.861080+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for the paper's own Type 4-2 example with $(a,b,c_0,d_0)=(3,2,-100,159)$, the complete mutation sequence for $g$-vectors and check whether the third component stays identically zero for every $m\\ge 5$ as claimed; any later nonzero third component before the normalization limit, or any divergence of the normalized sequence, would falsify Proposition 3.1 and hence Theorem 4.2.","supporting_citations":[],"review_version":1}