The Fibonacci Quilt Game

Alexandra Newlon; Steven J. Miller

REVIEW 2 major objections 5 minor 13 references

The Fibonacci Quilt Game

T0 review · 2 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Every play of the Fibonacci Quilt Game ends at an FQ-legal decomposition, and the shortest game on n takes exactly n − L(n) moves.

desk verdict New game, solid termination proof, but the shortest-game theorem overreaches: the paper's own words concede the upper bound is unproven. read the letter →

arxiv 1909.01938 v1 pith:W2HRGFOM submitted 2019-09-04 math.NT math.CO

Steven J. Miller , Alexandra Newlon This is my paper · ORCID

classification math.NTmath.CO MSC 11B3991A46

keywords FibonacciQuiltsequenceZeckendorfgameFQ-legaldecompositionterminationlengthmonovariantnon-uniqueintegerdecompositionsrecurrencerelations

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper introduces a two-player game on the Fibonacci Quilt sequence, a spiral construction that generalizes the Fibonacci numbers and, unlike the Fibonacci numbers, does not give every integer a unique legal decomposition. It tries to establish that this game is well behaved: every play, regardless of how players choose moves, terminates in finitely many moves at an FQ-legal decomposition—a sum of quilt terms obeying the spiral's adjacency rules—so the last player to move wins. It also proves that the shortest possible game on a starting number $n$ lasts exactly $n - L(n)$ moves, where $L(n)$ is the largest number of summands in any FQ-legal decomposition of $n$. Because players can create games with different numbers of moves, the winner is not fixed by $n$ alone, and simulation evidence suggests uniformly random game lengths are asymptotically Gaussian. A reader should care because these results show which features of the classical Fibonacci-number game survive when uniqueness of decomposition is lost.

What carries the argument

The mechanism that makes termination provable is an almost-monovariant: the quantity $M = \sum_{i} \sqrt{i}$, where the sum runs over the indices of the quilt terms currently on the board. Every legal move except the special rule $\{q_1\wedge q_5\to q_2\wedge q_4\}$ strictly decreases $M$; the inequalities proving this are checked case by case for all rule families. Because that exceptional rule can be used at most once, $M$ is effectively strictly decreasing over the whole game. Since the board always represents a partition of the fixed number $n$, there are only finitely many possible positions, so no position can repeat and the game must end. For the length result, the load-bearing identity is the simple accounting fact that a move lowers the number of terms by at most one, which forces at least $n-L(n)$ moves, where $L(n)$ is the maximum cardinality of an FQ-legal decomposition.

What would settle it

Search the full legal-move tree for $n=50$, where the paper notes an FQ-legal decomposition has at most four terms. The theorem predicts a shortest game of $50-4=46$ moves; finding a 46-move game would support it, while an exhaustive proof that no 46-move game exists would refute it.

Watch

Extended reading notes

Core claim

The paper's central claim is that the Fibonacci Quilt Game is a well-defined combinatorial game with exact shortest-play information. Starting from $n$ copies of the quilt term $q_1=1$, players apply the rules of Definition 1.7, which combine two terms that are forbidden from appearing together (same index, consecutive, distance three, distance four, or the special $\{q_1,q_3\}$ pair) into allowed terms recorded by the quilt's recurrences. Theorem 1.8 asserts that every maximal sequence of legal moves terminates at an FQ-legal decomposition, so the game cannot loop forever and always has a winner. Theorem 1.9 asserts that a fastest game reaches such a decomposition in exactly $n-L(n)$ moves, where $L(n)$ is the maximum number of terms in any FQ-legal decomposition of $n$; the key observation is that each move can reduce the number of terms on the board by at most one. The paper also constructs, for every $n>5$, two distinct legal games whose lengths have opposite parity, meaning either player can win, and conjectures that the length of a uniformly random game converges to a Gaussian distribution.

Load-bearing premise

The shortest-game theorem assumes that for every starting number $n$, a legal representation with the maximum possible number of terms can actually be reached from $n$ ones by legal moves that each reduce the number of terms by exactly one; the paper says no general algorithm for doing this is known.

Editorial extensions

If this is right

No infinite or looping play is possible: every game of the Fibonacci Quilt Game ends at an FQ-legal decomposition, so the 'last move wins' rule always determines a winner.
The shortest possible game length is known exactly and equals $n-L(n)$; computing the shortest game is therefore equivalent to computing $L(n)$, a static extremal quantity about legal decompositions.
Any optimal shortest game must avoid every move that leaves the term count unchanged, because the lower bound is only achieved by moves that reduce the count by exactly one.
For every $n>5$, there are legal games of odd and even length, so the winner is not determined by the starting number alone; game strategy can decide the outcome.
If the Gaussian conjecture is correct, random play has a well-defined typical length with fluctuations of order $\sqrt{n}$, making the game's duration statistically predictable even though individual games vary.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The same almost-monovariant template—one exceptional rule that can fire at most once—likely proves termination for other games built from recurrences with a finite number of exceptional low-index moves, provided those exceptions cannot recur.
The two branching rules, $\{q_4^2\to q_1+q_6\}$ or $\{q_3+q_5\}$ and $\{q_6^2\to q_2+q_8\}$ or $\{q_3+q_7\}$, are natural levers for constructing whatever winning strategy exists; checking whether always choosing one branch changes the parity of the game would be a direct test.
Extending the random-game simulations to larger $n$ and comparing fourth and sixth moments of the length distribution would sharpen the paper's Gaussian conjecture beyond the $n\le 200$ evidence reported.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The Fibonacci Quilt Game is a real extension of the Zeckendorf game to a sequence with non-unique decompositions. That is genuinely new, and the termination proof (Theorem 1.8) is the strongest part: the sum-of-square-roots monovariant is a clever adaptation, and the treatment of the exceptional Rule (2a) is careful. The paper also earns credit for being honest: it explicitly says later that no algorithm reaches the claimed lower bound for all n.

But Theorem 1.9 is stated too strongly. The lower bound, n − L(n), follows from the term-count argument and is fine. The upper bound requires showing you can actually reach a maximal-length FQ-legal decomposition in exactly that many moves. The proof simply assumes that building each summand independently costs q_i − 1 moves and that those subgames never interfere on the shared unordered list. That is not justified, and the paper later contradicts the theorem by saying an algorithm to reach the lower bound for all n is still unknown. So the equality in Theorem 1.9 is a conjecture, not a theorem. Corollary 2.6 inherits the gap. The fix is easy: restate Theorem 1.9 as a lower bound plus a conjecture, or supply a concrete construction. There is also a small misprint in the proof of Theorem 1.8: the monovariant starts at n, not sqrt(n).

Minor concern: Lemma 2.2 relies on a case tree whose figures are not included in the text I saw. The argument is plausible, but a referee would want those figures or a written case analysis.

Overall, this is a solid recreational-math paper with one load-bearing flaw that is openly acknowledged. It deserves peer review, not desk rejection, because the game definition and the termination proof are worth having in the literature. The authors just need to bring the claims in line with what is proven.

Referee Report

2 major / 5 minor

Summary. The paper defines the Fibonacci Quilt Game, a two-player game based on the Fibonacci Quilt sequence, which generalizes the Zeckendorf Game to a sequence without unique legal decompositions. The main claimed results are: Theorem 1.8, that every game terminates in a finite number of moves at an FQ-legal decomposition; Theorem 1.9, that the shortest game on n ends in exactly n - L(n) moves, where L(n) is the maximum number of terms in an FQ-legal decomposition of n; Corollary 2.5, that game lengths can vary and either player can win for n > 5; and Conjecture 1.10, that random game lengths are asymptotically Gaussian. The termination argument uses a monovariant that decreases for all moves except one exceptional rule, which is shown to be usable at most once. The shortest-game argument splits into a lower bound via the maximum term count and an upper bound by concatenating shortest games on the summands of a maximum decomposition.

Significance. If the main results were fully established, this would be a worthwhile extension of the Zeckendorf Game to a setting with non-unique decompositions and a non-PLRS recurrence, and the explicit dependence of the shortest game length on the maximum decomposition size L(n) is an interesting contrast with the original game. The paper also gives a plausible termination mechanism through an 'almost monovariant' and presents numerical evidence for a Gaussian limit conjecture. The lower bound in Theorem 1.9 is correct and the termination proof is largely sound in spirit. However, the upper bound in Theorem 1.9 is not proved, and the paper explicitly acknowledges that no algorithm reaching the lower bound is known; this is a load-bearing gap in the central claim.

major comments (2)

[Section 2.2, proof of Theorem 1.9] The lower bound in Theorem 1.9 is sound: each move reduces the number of terms by at most one, and every terminal FQ-legal decomposition has at most L(n) terms, so at least n - L(n) moves are necessary. The upper bound, however, is not established. The proof assumes that a maximum decomposition n = q_{l1} + ... + q_{lL(n)} can be reached by playing independent shortest games on each summand, taking sum_i(q_{li} - 1) moves. This ignores that the game is played on one unordered list: terms produced while building one summand can interact with terms produced while building another, and leftover 1's can combine with intermediate terms such as q2, q3, or q4 through Rules (1a), (4a), or (5). The induction hypothesis on smaller integers does not automatically transfer to subgames embedded in a larger list. The paper itself states in Section 2.2 that 'an algorithm to reach the lower bound for all n is still unknown,' and Corollary 2.6 is derived only from Theorem 1.9, so no independent construction is supplied. Theorem 1.9 should be revised to present the lower bound as proved and the upper bound as a conjecture unless a concrete strategy is provided.
[Section 2.2, Corollary 2.6] Corollary 2.6 asserts that for any n it is possible to play the game using no two-to-two rules, which is exactly the content of the upper bound in Theorem 1.9. Since the proof of Theorem 1.9's upper bound is missing, Corollary 2.6 is unsupported. In particular, the paper's own admission that no algorithm for the lower bound is known is in tension with the unqualified statement of Corollary 2.6. The corollary should either be proved directly or recast as a conjecture.

minor comments (5)

[Section 1.2, Conjecture 1.10] The conjecture says 'decomposing n into it's Zeckendorf expansion' in a setting where Zeckendorf uniqueness does not hold; it should refer to an FQ-legal decomposition, and the phrase 'converges to a Gaussian' should be made precise by stating that the distribution of game lengths, suitably normalized, converges to a Gaussian.
[Section 2.1, proof of Theorem 1.8] In the monovariant argument, the text says 'At the beginning of the game the sum of the square roots of the indices is sqrt(n)'. Since the game begins with n copies of q1, the sum is n, not sqrt(n). This does not affect the termination argument, but the displayed value should be corrected.
[Section 2.1, Lemma 2.2] In the case where q7 is present, the displayed chain after Rule (2a) writes q2 ∧ q4 ∧ q9; the intended term is q7, since the next move q4 + q7 -> q1 + q8 uses q7. Please correct this typo.
[Section 2.2] The sentence 'We have obtained this lower bound for many values of n' is confusing; presumably it means that the lower bound of n - L(n) has been observed to be attainable for many values of n, but as written it does not clearly distinguish the proved lower bound from the unproved upper bound.
[References] Reference [BEFM1] is cited as 'to appear'; please update the citation if a final version or publication data now exists.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: cited prior work supplies definitions and proof strategy, while the termination proof and the lower-bound argument are independently established; the acknowledged gap in Theorem 1.9's upper bound is a proof limitation, not a circular reduction.

full rationale

The Fibonacci Quilt sequence and its recurrences are imported from [CFHMN], and the Zeckendorf Game framework is imported from [BEFM1, BEFM2]; these self-citations provide background, definitions, and proof strategy, but they do not supply the paper's target results. The termination proof for Theorem 1.8 is self-contained: Lemma 2.2 limits the exceptional Rule (2a) to at most one use, and Lemma 2.3 verifies by direct inequalities that every other move strictly decreases the sum of square roots of indices, giving a finite game. The lower bound in Theorem 1.9 is an independent counting argument: each move can reduce the number of terms by at most one, and every terminal FQ-legal decomposition can have at most L(n) terms, so at least n - L(n) moves are needed. The upper-bound construction is asserted by concatenating independent subgames on each summand of a maximum-cardinality decomposition; this is not proved in detail, and the paper itself concedes in Section 2.2 that 'an algorithm to reach the lower bound for all n is still unknown.' That concession is an explicit limitation on the proof of Theorem 1.9's upper bound, and it should be weighed as a correctness gap, not as circularity: the assertion does not reduce to its own conclusion, to a fitted parameter, or to a self-citation chain. No derivation step in the paper is equivalent by construction to its input.

Assumptions & free parameters 0 free parameters · 2 assumptions · 0 invented entities

No free parameters are fitted, and the paper introduces no new mathematical objects beyond the game itself. The central claims rest on the cited definition of the Fibonacci Quilt sequence and standard finiteness arguments.

assumptions (2)

domain assumption The Fibonacci Quilt sequence satisfies the recurrence relations q_{n+1}=q_n+q_{n-4} for n>=6 and q_{n+1}=q_{n-1}+q_{n-2} for n>=5, with initial values q1=1 through q4=4, taken from Catral et al.
The game rules in Definition 1.7 are built directly from these recurrences and from the FQ-legality condition of Definition 1.4; the paper proves no properties of the sequence itself.
standard math On every turn, the unordered list of terms is a partition of n into finitely many parts, so there are finitely many possible game states.
Used in the termination proof to combine with the strictly decreasing monovariant; no proof is supplied because it is elementary.

how reviews work

0 comments

Cite this review

Pith. "Pith review of The Fibonacci Quilt Game." pith.science (2026). https://pith.science/paper/W2HRGFOM

@misc{pith2026190901938,
  author       = {Pith},
  title        = {Pith review of: The Fibonacci Quilt Game},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/W2HRGFOM}},
  note         = {Machine review of arXiv:1909.01938}
}

abstract

Zeckendorf proved that every positive integer can be expressed as the sum of non-consecutive Fibonacci numbers. This theorem inspired a beautiful game, the Zeckendorf Game. Two players begin with $n \ 1$'s and take turns applying rules inspired by the Fibonacci recurrence, $F_{n+1} = F_n + F_{n-1}$, until a decomposition without consecutive terms is reached; whoever makes the last move wins. We look at a game resulting from a generalization of the Fibonacci numbers, the Fibonacci Quilt sequence. These arise from the two-dimensional geometric property of tiling the plane through the Fibonacci spiral. Beginning with 1 in the center, we place integers in the squares of the spiral such that each square contains the smallest positive integer that does not have a decomposition as the sum of previous terms that do not share a wall. This sequence eventually follows two recurrence relations, allowing us to construct a variation on the Zeckendorf Game, the Fibonacci Quilt Game. While some properties of the Fibonaccis are inherited by this sequence, the nature of its recurrence leads to others, such as Zeckendorf's theorem, no longer holding; it is thus of interest to investigate the generalization of the game in this setting to see which behaviors persist. We prove, similar to the original game, that this game also always terminates in a legal decomposition, give a lower bound on game lengths, show that depending on strategies the length of the game can vary and either player could win, and give a conjecture on the length of a random game.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

13 extracted references · 13 canonical work pages

[1]

The Zeckendorf Game

P. Baird-Smith, A. Epstein, K. Flint, S. J. Miller, The Zeckendorf Game, to appear in the Proceedings of the Conference in Combinatorial and Additive Number Theory. https://arxiv.org/abs/1809.04881

work page Pith review arXiv
[2]

The Generalized Zeckendorf Game

P. Baird-Smith, A. Epstein, K. Flint, S. J. Miller, The Generalized Zeckendorf Game, preprint. https://arxiv.org/abs/1809.04883

work page Pith review arXiv
[3]

Legal Decompositions Arising from Non-positive Linear Recurrences

M. Catral, P. Ford, P. Harris, S. J. Miller, D. Nelson, Legal Decompositions Arising from Non-positive Linear Recurrences, Fibonacci Quarterly 54 (2016), no. 4, pp. 348-365. https://arxiv.org/abs/1606.09312

work page Pith review arXiv 2016
[4]

New Behavior in Legal Decompositions Arising from Non-positive Linear Recurrences

M. Catral, P. Ford, P. Harris, S. J. Miller, D. Nelson, Z. Pan, H. Xu, New Behavior in Legal Decompositions Arising from Non-positive Linear Recurrences, Fibonacci Quarterly 55 (2017), no. 3, pp. 252-275. https://arxiv.org/abs/1606.09309

work page Pith review arXiv 2017
[5]

Cordwell, M

K. Cordwell, M. Hlavacek, C. Huynh, S. J. Miller, C. Peterson and Y. N. Truong Vu, On Summand Minimality of Generalized Zeckendorf Decompositions, Research in Number Theory 4 (2018), no. 43, https://doi.org/10.1007/s40993-018-0137-7

work page doi:10.1007/s40993-018-0137-7 2018
[6]

Hoggatt, Generalized Zeckendorf theorem, Fibonacci Quarterly 10 (1972), no

V.E. Hoggatt, Generalized Zeckendorf theorem, Fibonacci Quarterly 10 (1972), no. 1 (special issue on representations), pages 89-93

work page 1972
[7]

Keller, Generalizations of Zeckendorf's theorem, Fibonacci Quarterly 10 (1972), no

T.J. Keller, Generalizations of Zeckendorf's theorem, Fibonacci Quarterly 10 (1972), no. 1 (special issue on representations), pages 95-102

work page 1972
[8]

Kolo g lu, G

M. Kolo g lu, G. Kopp, S. Miller, Y. Wang, On the Number of Summands in Zeckendorf Decompositons, Journal of Number Theory, 49 (2011), no. 2, 116-130

work page 2011

Show all 13 references

[9]

Koshy, Fibonacci and Lucas Numbers with Applications, Wiley-Interscience, New York, 2001

T. Koshy, Fibonacci and Lucas Numbers with Applications, Wiley-Interscience, New York, 2001

2001
[10]

Lekkerkerker, Voorstelling van natuurlyke getallen door een som van getallen van Fibonacci, Simon Stevin 29 (1951-1952), 190-195

C.G. Lekkerkerker, Voorstelling van natuurlyke getallen door een som van getallen van Fibonacci, Simon Stevin 29 (1951-1952), 190-195

1951
[11]

Miller, Y

S. Miller, Y. Wang, From Fibonacci Numbers to Central Limit Type Theorems, Journal of Combinatorial Theory, Series A 119 (2012), no. 7, 1398-1413

2012
[12]

Miller, Y

S. Miller, Y. Wang, Gaussian Behavior in Generalized Zeckendorf Decompositions, Combinatorial and Additive Number Theory, CANT 2011 and 2012 (Melvyn B. Nathanson, editor), Springer Proceedings in Mathematics & Statistics (2014), 159-173

2014
[13]

E. Zeckendorf, Repres\'entation des nombres naturels par une somme des nombres de Fibonacci ou de nombres de Lucas, Bulletin de la Soci\'et\'e Royaledes Sciences de Li\'ege 41 (1972), pp. 179--182

1972

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

The Zeckendorf Game

P. Baird-Smith, A. Epstein, K. Flint, S. J. Miller, The Zeckendorf Game, to appear in the Proceedings of the Conference in Combinatorial and Additive Number Theory. https://arxiv.org/abs/1809.04881

work page Pith review arXiv

[2] [2]

The Generalized Zeckendorf Game

P. Baird-Smith, A. Epstein, K. Flint, S. J. Miller, The Generalized Zeckendorf Game, preprint. https://arxiv.org/abs/1809.04883

work page Pith review arXiv

[3] [3]

Legal Decompositions Arising from Non-positive Linear Recurrences

M. Catral, P. Ford, P. Harris, S. J. Miller, D. Nelson, Legal Decompositions Arising from Non-positive Linear Recurrences, Fibonacci Quarterly 54 (2016), no. 4, pp. 348-365. https://arxiv.org/abs/1606.09312

work page Pith review arXiv 2016

[4] [4]

New Behavior in Legal Decompositions Arising from Non-positive Linear Recurrences

M. Catral, P. Ford, P. Harris, S. J. Miller, D. Nelson, Z. Pan, H. Xu, New Behavior in Legal Decompositions Arising from Non-positive Linear Recurrences, Fibonacci Quarterly 55 (2017), no. 3, pp. 252-275. https://arxiv.org/abs/1606.09309

work page Pith review arXiv 2017

[5] [5]

Cordwell, M

K. Cordwell, M. Hlavacek, C. Huynh, S. J. Miller, C. Peterson and Y. N. Truong Vu, On Summand Minimality of Generalized Zeckendorf Decompositions, Research in Number Theory 4 (2018), no. 43, https://doi.org/10.1007/s40993-018-0137-7

work page doi:10.1007/s40993-018-0137-7 2018

[6] [6]

Hoggatt, Generalized Zeckendorf theorem, Fibonacci Quarterly 10 (1972), no

V.E. Hoggatt, Generalized Zeckendorf theorem, Fibonacci Quarterly 10 (1972), no. 1 (special issue on representations), pages 89-93

work page 1972

[7] [7]

Keller, Generalizations of Zeckendorf's theorem, Fibonacci Quarterly 10 (1972), no

T.J. Keller, Generalizations of Zeckendorf's theorem, Fibonacci Quarterly 10 (1972), no. 1 (special issue on representations), pages 95-102

work page 1972

[8] [8]

Kolo g lu, G

M. Kolo g lu, G. Kopp, S. Miller, Y. Wang, On the Number of Summands in Zeckendorf Decompositons, Journal of Number Theory, 49 (2011), no. 2, 116-130

work page 2011