{"id":"f39f48d9-8d01-4e7c-88d1-f8d1a29b527d","arxiv_id":"1908.09350","paper_version":3,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A simplicial dollar game with a Hilbert-basis degree is introduced; chains of large degree are always winnable, and zero-degree winnability characterizes higher-dimensional forests.","lead":"This paper generalizes the chip-firing dollar game from graphs to higher-dimensional simplicial complexes, using a new vector-valued degree based on the Hilbert basis of the nonnegative Laplacian kernel. It proves that chains of sufficiently large degree are always winnable, and that all degree-zero chains are winnable exactly when the complex is a higher-dimensional forest.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified","rationale":"The paper's central claim, Theorem 18, is well-supported by a chain of results: Lemma 10 provides a strictly positive element in ker L_i, Corollary 12 identifies the perpendicular of the nonnegative kernel with the perpendicular of the full kernel, Theorem 13 identifies degree-zero chains modulo firing with the torsion of the critical group, and Lemma 17 gives a bounded decomposition for chains of nonnegative degree. I checked each step in detail. Lemma 10's star argument works because the star of an (i-1)-face is a cocycle with nonnegative coefficients on all lexicographically earlier faces; the maximality contradiction is valid. Corollary 12's basis-perturbation argument is standard and correct. Theorem 13's proof of the reverse inclusion uses the fact that if kσ ∈ im L_i, then σ is orthogonal to ker L_i, which holds by the symmetry of the Laplacian. Lemma 17's decomposition relies on the standard fact that a rational polyhedral cone Q satisfies Q = (Q∩Z^f) + Π(G) for any integral generating set G; since the integer points are exactly the first summand, the intersection with Z^f splits as the sum of integer parts plus a finite set (P1+P2)∩Z^f. The final argument of Theorem 18 is a straightforward application of these lemmas. I found no internal inconsistency or unproven step that threatens the theorem. The only issue is a typographical error in the introduction to Section 4, where a cone C is defined with L_i v ≥ 0 instead of L_i v = 0; this C is not used in any proof and the subsequent Definition 4 correctly defines the Hilbert basis for ker+ L_i. The reader's identification of Lemma 10 as the most delicate structural premise is fair; it is the key to Corollary 12 and hence to the degree-zero/torsion identification, but it is correct for finite simplicial complexes. Therefore the verdict of ACCEPT with moderate confidence remains appropriate.","tokens_in":25083,"tokens_out":52971,"duration_ms":487332,"concrete_test":"Recompute the Hilbert basis for ker+ L_1 of the hollow tetrahedron (Example 24) and verify that the sum of all Hilbert basis elements is a strictly positive element of ker+ L_1, confirming Lemma 10 for this nontrivial complex.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I traced the proof of Theorem 18 through its supporting lemmas and found no load-bearing flaw. Lemma 10's lexicographic maximality argument is sound: the star S = ∂_i^t(v1...vi) lies in ker L_i because im ∂_i^t ⊆ ker ∂_{i+1}^t, has coefficient 1 at m, and has nonnegative coefficients on all lexicographically earlier faces, so σ' = σ + (1 - σ(m))S has no nonpositive face ≤ m, contradicting maximality. Corollary 12 correctly shows the Z-span of ker+ L_i is ker L_i by perturbing a basis with large multiples of a primitive positive vector. Theorem 13's identification of degree-zero chains modulo firing with the torsion of K_i is correct, and Lemma 17's cone decomposition is justified because rational polyhedral cones split into integer points plus a bounded fundamental parallelepiped. The final step of Theorem 18, choosing ω that dominates all torsion representatives and the finite error set, is sound. The only textual inconsistency is the stray definition of C = {v : L_i v ≥ 0, v ≥ 0} before Definition 4; it is never used, and Definition 4 correctly uses the Hilbert basis of ker+ L_i. This appears to be a typographical relic and does not affect the mathematics.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a higher-dimensional analogue of the Baker--Norine dollar game on simplicial complexes, using Duval--Klivans--Martin's chip-firing theory. It defines a degree vector for i-chains via the Hilbert basis of the nonnegative kernel of the i-th up-down Laplacian. The main results are Theorem 18, stating that every i-chain of sufficiently large degree is winnable, and Corollary 34, characterizing when all (i-1)-chains of degree zero are winnable in terms of spanning i-forests and torsion-free (i-1)-homology, thereby generalizing the tree case for graphs. Additional contributions include the identification of degree-zero chains modulo firing with the torsion of the critical group (Theorem 13), a combinatorial description of the Hilbert basis in codimension one for orientable pseudomanifolds (Theorem 22), a generalization of the reduced Laplacian isomorphism (Theorem 30), and several worked examples including minimal winning degrees.","tokens_in":25301,"tokens_out":10806,"duration_ms":107952,"significance":"The paper gives a natural invariant notion of degree for simplicial chip-firing and establishes a higher-dimensional analogue of the classical winnability threshold. The proof strategy is clean: the degree is defined independently of winnability, the key technical Lemma 10 is proved directly, and Theorem 18 follows from Lemma 17 together with finiteness of the torsion of the critical group. The identification of degree-zero chains modulo firing with torsion (Theorem 13) and the pseudomanifold Hilbert-basis theorem (Theorem 22) are valuable structural results. The paper also provides explicit computations and examples showing that higher-dimensional behavior differs from graphs, which is informative. I find no circularity: external dependencies, such as Duval--Klivans--Martin's Theorem 33, are prior results by other authors, and the authors' own arguments are not used to prove the main theorems in a circular manner.","major_comments":[],"minor_comments":[{"comment":"The displayed definition \"C := {v in R^{f_i} : L_i v >= 0 and v >= 0}\" is inconsistent with Definition 4, which correctly defines the degree using the Hilbert basis of ker_+ L_i. The cone C is never used afterward, and a reader could momentarily mistake it for the intended kernel; it should be removed or replaced.","section":"Section 4, opening paragraph"},{"comment":"The phrase \"for each integer i\" should be restricted to i with 0 <= i <= dim(Delta), or the empty-face case should be handled explicitly, since the proof of Lemma 10 chooses a lexicographically smallest i-face and does not apply when Delta_i is empty.","section":"Lemma 10 and Theorem 18"},{"comment":"The term \"integrally generated\" is used without definition; because the Hilbert basis property depends on it, a one-sentence definition would help the reader.","section":"Section 2.2, Polyhedral cones"},{"comment":"The Sage computations are reported but the input data are not included; for reproducibility, the authors might include a short appendix or reference to the code used.","section":"Examples 36 and 37"},{"comment":"The phrase \"unique maximal linearly independent subset\" should be read as \"unique basis of the column space\" or \"unique maximum-cardinality linearly independent subset\"; the current wording is slightly ambiguous.","section":"Proof of Proposition 32"}],"recommendation":"minor_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know before you read it. First, the paper really does generalize the graph dollar game to all dimensions, not just relabel the old theorems. The new object is a vector-valued degree defined by dot products with the Hilbert basis of the nonnegative Laplacian kernel, and the two headline results—Theorem 18 (large enough degree implies winnable) and Corollary 34 (all degree-zero chains winnable iff the appropriate forest number is one)—are genuine extensions of the Baker–Norine facts. Second, the proofs hold up. I traced the chain behind Theorem 18: Lemma 10's lexicographic maximality argument is sound, Lemma 17's cone decomposition is standard, and the deduction from DKM's Theorem 33 is legitimate. No circularity: the degree invariant is defined without reference to winnability.\n\nThe strongest parts are the clean proof of Theorem 18 and the pseudomanifold result, Theorem 22, where the Hilbert basis is described as incidence vectors of simple directed cycles in the γ-incidence graph. That is a concrete and useful payoff. The paper also improves DKM's reduced Laplacian theorem by loosening the hypotheses, and the appendix proofs are careful. Proposition 21 for orientable pseudomanifolds is a nice complement to DKM.\n\nSoft spots are minor. The Sage/PyNormaliz computations in the examples are not attached as code or data, so those examples are not independently reproducible without re-doing the computations; this affects examples, not the main theorems. There is a stray definition of C = {v : L_i v ≥ 0, v ≥ 0} just before Definition 4 that is never used—a harmless typo. The degree is a vector in Z^{|H|}, so Theorem 18's δ is existential and nonconstructive; the paper is upfront about this and lists it under further work. None of this undermines the central argument.\n\nThe reader's ACCEPT with moderate confidence matches my own read. This is a solid, important-for-the-subfield extension, not a breakthrough, and it will be cited by people working on combinatorial divisor theory and chip-firing. I would send it to a serious referee: the main results deserve verification and the computational examples could be checked. I'd also bring it to a reading group if anyone in the group works on chip-firing, because the Hilbert-basis degree is worth understanding in its own right.","headline":"This is a genuine higher-dimensional generalization of the graph dollar game, and the main theorems—large-degree winnability and the forest/torsion-free characterization—are proven carefully from a new Hilbert-basis degree; it deserves a serious referee.","tokens_in":25819,"tokens_out":1354,"would_cite":true,"duration_ms":16263,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05E45"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every high-degree chain wins the simplicial dollar game","keywords":["chip-firing","dollar game","simplicial complex","critical group","Hilbert basis","Laplacian","winnability","spanning forest"],"falsifier":"For any fixed simplicial complex and dimension $i$, compute the realizable degrees and test whether every chain of degree at least some $\\delta$ is winnable; if no such $\\delta$ exists, Theorem 18 fails, and independently, finding a complex where $\\ker L_i$ contains no strictly positive integer vector would falsify Lemma 10, the step on which the theorem rests.","tokens_in":24878,"feed_emoji":"🪙","tokens_out":15054,"duration_ms":130587,"temperature":0.7,"pith_summary":"The dollar game on a graph asks whether an integer assignment of dollars to vertices can be transformed by lending and borrowing moves into a position with no debts. This paper lifts that game to simplicial complexes of any dimension, playing on $i$-faces with firing moves governed by the combinatorial Laplacian $L_i$. Because the total amount of money is not conserved in higher dimensions, the authors replace scalar degree with a vector-valued degree $\\deg(\\sigma)$ computed by the Hilbert basis of the nonnegative kernel of $L_i$. Their main theorem, Theorem 18, says that for every dimension $i$ there is a realizable degree $\\delta$ such that every $i$-chain with degree at least $\\delta$ is winnable, generalizing the graph statement that divisors of degree at least the genus can be won. A second result, Corollary 34, characterizes the simplicial complexes on which every degree-zero chain is winnable: exactly those whose $i$-skeleton is a spanning forest that is torsion-free in codimension one.","feed_headline":"Every high-degree chain wins the simplicial dollar game","feed_subtitle":"A vector-valued degree is the right measure; degree-zero wins mark the higher-dimensional trees.","key_machinery":"The load-bearing object is the Hilbert basis of the nonnegative kernel of the Laplacian. Concretely, $\\mathcal H_i$ is the unique minimal set such that every nonnegative integer $i$-chain in $\\ker L_i$ is a nonnegative integer combination of elements of $\\mathcal H_i$, and the degree of a chain is the vector of inner products with these basis elements. A second mechanism is Lemma 10, which guarantees a strictly positive element in $\\ker L_i$; that positivity forces the nonnegative kernel to span the whole kernel, so degree-zero chains can be identified with torsion classes of the critical group and Theorem 18 follows from a finite decomposition argument. For orientable pseudomanifolds, the Hilbert basis in codimension one is combinatorially explicit: it consists of the incidence vectors of simple directed cycles of the $\\gamma$-incidence graph.","core_discovery":"The paper establishes that the graph-theoretic dichotomy 'large enough degree wins, and zero degree wins exactly on trees' survives in higher dimensions once degree is redefined. Two $i$-chains are linearly equivalent if they differ by an element in the image of $L_i$, and the degree of a chain is the vector of dot products with the elements of the Hilbert basis $\\mathcal H_i$ of $\\ker^+ L_i$, the monoid of nonnegative integer chains in the kernel of the Laplacian. This degree is invariant under firing moves and nonnegative on effective chains, while the naive sum of coefficients is neither. Theorem 18 states that for each $i$ there exists a realizable degree $\\delta$ such that every chain of degree at least $\\delta$ is winnable; Theorem 13 identifies degree-zero classes modulo firing with the torsion subgroup of the critical group $K_i(\\Delta)$; and Corollary 34 shows that all $(i-1)$-chains of degree zero are winnable in $\\Delta$ if and only if the $i$-skeleton is a spanning $i$-forest with $\\widetilde H_{i-1}(\\Delta)$ torsion-free.","pith_inferences":["When the Hilbert basis consists of 0-1 vectors, Corollary 20 upgrades the theorem to an exact threshold: winnability at a single degree forces winnability at all larger degrees, so the minimal winning set is particularly clean for such complexes.","Because degree-zero classes are torsion of the critical group, tabulating critical-group torsion over families of complexes would automatically answer whether the zero-degree dollar game is universally winnable.","The documented failure of the greedy algorithm and q-reduction in higher dimensions suggests that deciding winnability for unfixed dimension could be harder than in graphs, though the paper only poses this as an open question.","The paper's open problem about minimal winning degrees for (d-2)-chains on the d-simplex could serve as a concrete benchmark for how the vector threshold grows with dimension."],"forward_implications":["If the theorem is right, a finite computation at one sufficiently large degree settles winnability for all larger degrees, because degree-zero classes form a finite torsion group.","The tree characterization carries over: the complexes with universally winnable degree-zero chains are exactly the spanning forests torsion-free in codimension one, so winnability at degree zero detects a homological property.","On orientable pseudomanifolds, the Hilbert basis is encoded by simple directed cycles in a gamma-incidence graph, making degree calculations combinatorial rather than algebraic.","The set of minimal winning degrees is a finite antichain of vectors, replacing the single graph genus as the threshold invariant; the paper computes one such antichain for the hollow tetrahedron."],"supporting_citations":[{"why":"Introduces the graph dollar game, divisor rank, and Riemann-Roch theorem whose degree threshold and tree characterization this paper generalizes.","marker":"[2]"},{"why":"Supplies the simplicial Laplacian and critical groups that define the higher-dimensional firing moves and the target objects $K_i(\\Delta)$.","marker":"[8]"},{"why":"Provides the theorem relating critical-group torsion to forest numbers, used to prove Corollary 34.","marker":"[9]"},{"why":"Establishes the theory of simplicial and cellular trees and forests that underlies the spanning-forest characterization.","marker":"[10]"},{"why":"Motivates the alternative nonnegative-degree sets $X_i$ considered in Section 6.1 and Proposition 39.","marker":"[6]"}],"fun_headline_variants":["Simplicial dollar game: high-degree chains always win","Higher-dimensional dollar game: degree is now a vector","Zero-degree wins mark higher-dimensional trees in dollar game","Dollar game in simplicial complexes: vector degree decides","From graphs to complexes: chip-firing winnability with new degree"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is Lemma 10, the assertion that in every dimension the Laplacian kernel contains a chain whose coefficients are all strictly positive, since without such a positive kernel element the identification of degree-zero classes with critical-group torsion, and hence the high-degree winnability theorem, would not follow.","fun_headline_variants_meta":{"raw":{"variants":["Simplicial dollar game: high-degree chains always win","Higher-dimensional dollar game: degree is now a vector","Zero-degree wins mark higher-dimensional trees in dollar game","Dollar game in simplicial complexes: vector degree decides","From graphs to complexes: chip-firing winnability with new degree"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000155,"raw_usage":{"total_tokens":1239,"prompt_tokens":994,"completion_tokens":245,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":610,"completion_tokens_details":{"reasoning_tokens":162}},"tokens_in":610,"tokens_out":245,"duration_ms":3476,"temperature":1.0,"reasoning_tokens":162,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:15:13.568905+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For any fixed simplicial complex and dimension $i$, compute the realizable degrees and test whether every chain of degree at least some $\\delta$ is winnable; if no such $\\delta$ exists, Theorem 18 fails, and independently, finding a complex where $\\ker L_i$ contains no strictly positive integer vector would falsify Lemma 10, the step on which the theorem rests.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the graph dollar game, divisor rank, and Riemann-Roch theorem whose degree threshold and tree characterization this paper generalizes."},{"cited_title":"Duval, Caroline J","cited_arxiv_id":null,"evidence_quote":"Supplies the simplicial Laplacian and critical groups that define the higher-dimensional firing moves and the target objects $K_i(\\Delta)$."},{"cited_title":"Algebraic Combin","cited_arxiv_id":null,"evidence_quote":"Provides the theorem relating critical-group torsion to forest numbers, used to prove Corollary 34."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the theory of simplicial and cellular trees and forests that underlies the spanning-forest characterization."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Motivates the alternative nonnegative-degree sets $X_i$ considered in Section 6.1 and Proposition 39."}],"review_version":1}