REVIEW 2 major objections 4 minor 34 references
The repetition threshold for binary rich words
T0 review · 2 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read The least possible repetition exponent for infinite binary rich words is $2+\sqrt{2}/2$.
desk verdict The structure theorem is a real contribution, but the final step from Theorem 1 to the claimed repetition threshold skips the entire lower-bound argument. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The argument rests on a structure theorem and a morphism calculus. The theorem classifies infinite binary rich words that avoid $14/5$-powers: every such word has, for every $n$, a suffix of the form $f(h^n(w_n))$ or $f(g(h^n(w_n)))$, where $f$, $g$, and $h$ are fixed substitutions (for example, $f$ maps $0 \mapsto 0$, $1 \mapsto 01$, $2 \mapsto 011$) and $w_n$ is a cube-free rich ternary word containing a $0$. The calculus identifies the two extremal words $f(h^\omega(0))$ and $f(g(h^\omega(0)))$ as complementary symmetric Rote words, meaning their first-difference sequences are Sturmian; this allows the exponent of their longest repetitions to be computed from the continued fraction of a slope, yielding the value $2+\sqrt{2}/2$.
What would settle it
Exhibit an infinite binary rich word whose critical exponent is strictly below $2+\sqrt{2}/2$, or find a cube-free rich ternary word $w$ containing a $0$ for which $f(h^n(w))$ or $f(g(h^n(w)))$ has critical exponent below $2+\sqrt{2}/2$; the paper's claim predicts neither exists.
Extended reading notes
Core claim
The paper's central claim is that the repetition threshold for binary rich words is $2+\sqrt{2}/2$. The main structural result, Theorem 1, states that if an infinite binary rich word avoids repetitions of exponent at least $14/5$, then for every $n \ge 1$ some suffix has the form $f(h^n(w_n))$ or $f(g(h^n(w_n)))$, where $w_n$ is a cube-free rich word over $\{0,1,2\}$ containing a $0$, and $f$, $g$, and $h$ are explicit morphisms. The paper then proves that the two canonical words $f(h^\omega(0))$ and $f(g(h^\omega(0)))$, both complementary symmetric Rote words, have critical exponent exactly $2+\sqrt{2}/2$; from this and Theorem 1 it concludes that no rich binary word can have a smaller critical exponent, so the threshold is exactly $2+\sqrt{2}/2$.
Load-bearing premise
The final inference from the structure theorem to the lower bound relies on the unstated premise that every suffix of the allowed forms attains critical exponent at least $2+\sqrt{2}/2$; the paper computes the critical exponent explicitly only for its two constructed extremal words, not for all words satisfying the structure theorem.
Editorial extensions
If this is right
- The binary repetition threshold for rich words is now known exactly: $2+\sqrt{2}/2$ is both a lower bound and an attained value.
- Every infinite binary rich word contains arbitrarily long factors with exponent at least $2+\sqrt{2}/2$, so near-threshold rich words are necessarily repetition-heavy.
- The structure theorem gives a normal form for $14/5$-free binary rich words, analogous to classical structure theorems for overlap-free words, which may aid future algorithmic studies of these words.
- With the binary case resolved, the same threshold question remains open for alphabets of size three and larger, where no conjectured value is currently available.
Reading between the lines
- If the unstated monotonicity is correct, then every $14/5$-free binary rich word should have critical exponent exactly $2+\sqrt{2}/2$; the paper exhibits two examples but does not rule out other words with the same value.
- The morphism calculus suggests that the family of $14/5$-free binary rich words may be generated by iterating $h$ on arbitrary rich seeds, which would make the class amenable to enumeration and could yield upper bounds for larger alphabets through continued fractions of associated slopes.
- A direct computational test of the structure theorem would be to generate all cube-free rich ternary words containing a $0$, apply the morphisms $f$ and $f\circ g$, and check whether any resulting word has critical exponent below $2+\sqrt{2}/2$; the paper's argument predicts none.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies the repetition threshold for infinite binary rich words. It introduces three morphisms f, g, h on a ternary alphabet and proves a structure theorem (Theorem 1): every infinite binary rich word that avoids 14/5-powers has, for each n ≥ 1, a suffix of the form f(h^n(w_n)) or f(g(h^n(w_n))) for some infinite ternary word w_n. The proof combines a forbidden-factor analysis with induction, using several computer-assisted backtracking checks. In Section 3 the authors show that the two words f(h^ω(0)) and f(g(h^ω(0))) are rich and have critical exponent 2+√2/2, using a connection to complementary symmetric Rote words and Sturmian words. From Theorem 1 and these two examples they conclude (Theorem 2) that the repetition threshold for binary rich words is exactly 2+√2/2, resolving a conjecture of Baranwal and Shallit.
Significance. If the conclusion is fully established, the paper resolves the binary case of Vesti's problem and confirms the Baranwal–Shallit conjecture. The structure theorem is a strong and interesting dichotomy, and the use of Sturmian/Rote theory to compute critical exponents of the extremal words is elegant and mostly self-contained. However, the final deduction of Theorem 2 from Theorem 1 is missing a load-bearing lower-bound argument. The paper also depends on several undocumented backtracking computations. These issues do not undermine the plausibility of the main result, but they prevent the paper, in its current form, from being a complete proof.
major comments (2)
- [Section 3, final paragraph] The sentence "Since f(h^ω(0)) and f(g(h^ω(0))) both have critical exponent 2+√2/2, Theorem 2 now follows immediately from Theorem 1" is the entire lower-bound half of Theorem 2, and it does not follow as written. To prove RRT(2)=2+√2/2 one must show not only that some binary rich word attains this exponent, but also that every infinite binary rich word has critical exponent at least 2+√2/2. For the lower bound, suppose w is a rich word with critical exponent < 2+√2/2; then w is 14/5-free, so Theorem 1 applies and gives, for each n ≥ 1, a suffix of the form f(h^n(w_n)) or f(g(h^n(w_n))). The paper computes the critical exponent only for the two particular words obtained when w_n is h^ω(0). No lemma states or proves that an arbitrary rich cube-free w_n (as guaranteed by Lemma 7) yields a word f(h^n(w_n)) or f(g(h^n(w_n))) with critical exponent at least 2+√2/2. The missing statement is precisely a monotonicity property: for all n ≥ 1 and all rich cube-free w, if f(h^n(w)) or f(g(h^n(w))) is 14/5-free, then its critical exponent is at least 2+√2/2. Without this lemma, Theorem 1 is compatible with the existence of rich 14/5-free words of smaller critical exponent, and Theorem 2 is not proved.
- [Section 2, Lemma 12 and Observation 3/Table 1] The exclusion of the factor 212 in Lemma 12 rests on an unspecified backtracking computation: "Backtracking by computer ... one finds that the longest right extension of 212 has length 21." The paper does not provide the code, the actual extension, or a certificate that would allow the reader to verify this finite check. The same is true for Observation 3 and for the entries of Table 1, where only the lengths of the longest right extensions are reported, not the extensions themselves. These checks are load-bearing for the induction in Theorem 1, since they rule out factors that drive the structure theorem. To make the proof reproducible, please include the program (or machine-readable certificates) and state the exact search parameters, including whether the search is over 14/5-free or cube-free extensions and how the stopping condition is certified.
minor comments (4)
- [Theorem 1 statement] The phrase "For everyn ≥ 1" should read "For every n ≥ 1".
- [Theorem 17, after Lemma 16] The text says "the number of 1's in v_{i+1} ... v_{i+ℓ−1} v_ℓ" but Lemma 16 gives the number of 1's in v_i ... v_{i+ℓ−1}; the index appears to be shifted and should be corrected.
- [Lemma 16 and Theorem 17] The symbol y is used both for the infinite word Δ(x) in Lemma 16 and for the prefix y' in the repetition y^e y'; this is confusing. Consider renaming the prefix, for example to p or z'.
- [Theorem 17, non-primitive z case] The sentence "Since the critical exponent of c_α is 3+√2, the exponent of z cannot be greater than 2" is compressed. Spelling out that z = r^m with m=2 and e=2 would make the argument easier to follow and would remove an unnecessary hurdle for the reader.
Circularity Check
No significant circularity: the main derivation is self-contained, though the final inference from Theorem 1 to Theorem 2 contains a logical gap that is not a circularity.
full rationale
The paper's derivation chain is not circular. The structure theorem (Theorem 1) is proved from scratch using the morphisms f, g, h, finite backtracking checks (Observation 3, Table 1), and lemmas about richness and cube-freeness; it does not presuppose the repetition threshold. The critical exponent computation for u = f(g(h^ω(0))) in Theorem 17 is derived independently from Sturmian word theory, the Rote-word characterization, and cited external results on repetitions in Sturmian words; it is not a fitted parameter or a restatement of the conjecture. The use of Baranwal and Shallit's word f(h^ω(0)) is an external construction, not a self-citation, and the paper independently verifies richness of both words via complementary symmetric Rote words. No step reduces by construction to its own inputs, no fitted quantity is renamed as a prediction, and no load-bearing uniqueness theorem is imported from the authors' prior work. The only substantive concern is logical rather than circular: the final sentence 'Since f(h^ω(0)) and f(g(h^ω(0))) both have critical exponent 2+√2/2, Theorem 2 now follows immediately from Theorem 1' omits the lower-bound half of the repetition-threshold proof, namely showing that every infinite binary rich word covered by the structure theorem has critical exponent at least 2+√2/2. That is an omitted proof or a gap in the written argument, but it is not an equivalence between the theorem and its assumptions. Accordingly, the circularity score is 0.
Assumptions & free parameters
assumptions (6)
- domain assumption A word is rich if and only if every nonempty prefix has a palindromic suffix that occurs only once.
- domain assumption Complementary symmetric Rote words are rich.
- domain assumption The characteristic Sturmian word with slope (3-sqrt(2))/7 has critical exponent 3+sqrt(2).
- domain assumption Repetitions in Sturmian words are characterized by standard and semi-standard words.
- domain assumption The longest factor of c_alpha with period q_{k-2}+q_{k-1} has length 2(q_{k-2}+q_{k-1})+q_{k-1}-2.
- ad hoc to paper The backtracking computations confirming Observation 3, Table 1, and Lemma 12's claim about the right-extension of 212 are correct.
Cite this review
Pith. "Pith review of The repetition threshold for binary rich words." pith.science (2026). https://pith.science/paper/3Z534MTV
@misc{pith2026190803169,
author = {Pith},
title = {Pith review of: The repetition threshold for binary rich words},
year = {2026},
howpublished = {\url{https://pith.science/paper/3Z534MTV}},
note = {Machine review of arXiv:1908.03169}
}
abstract
A word of length $n$ is rich if it contains $n$ nonempty palindromic factors. An infinite word is rich if all of its finite factors are rich. Baranwal and Shallit produced an infinite binary rich word with critical exponent $2+\sqrt{2}/2$ ($\approx 2.707$) and conjectured that this was the least possible critical exponent for infinite binary rich words (i.e., that the repetition threshold for binary rich words is $2+\sqrt{2}/2$). In this article, we give a structure theorem for infinite binary rich words that avoid $14/5$-powers (i.e., repetitions with exponent at least 2.8). As a consequence, we deduce that the repetition threshold for binary rich words is $2+\sqrt{2}/2$, as conjectured by Baranwal and Shallit. This resolves an open problem of Vesti for the binary alphabet; the problem remains open for larger alphabets.
Figures
Reference graph
Works this paper leans on
- [1]
-
[2]
L. Balkov´ a, E. Pelantov´ a,ˇS. Starosta, Infinite words with finite defect, Adv. Appl. Mat h. 47 (2011), 562–574
work page 2011
-
[3]
A. Baranwal and J. Shallit, Repetitions in infinite palin drome-rich words, in: R. Mercas and D. Reidenbach (Eds.), Proc. WORDS 2019, Lecture Notes in Computer Science, V ol. 11682, Springer, (2019), 93–105
work page 2019
-
[4]
A. Baranwal and J. Shallit, Critical exponent of infinite balanced words via the Pell number system, in: R. Mercas and D. Reidenbach (Eds.), Proc. WORDS 2019, Lecture Notes in Computer Science, V ol. 11682, Springer, (2019), 80–92
work page 2019
-
[5]
A. Blondin Mass´ e, S. Brlek, S. Labb´ e, L. Vuillon, Palindromic complexity of codings of rotations, Theoret. Comput. Sci. 412 (2011), 6455–6463
work page 2011
- [6]
-
[7]
Carpi, On Dejean’s conjecture over large alphabets, T heoret
A. Carpi, On Dejean’s conjecture over large alphabets, T heoret. Comput. Sci. 385 (2007), 137–151
work page 2007
-
[8]
J. D. Currie and N. Rampersad, Dejean’s conjecture holds for n ≥ 27, RAIRO - Theor. Inform. Appl. 43 (2009), 775–778
work page 2009
Show all 34 references
-
[9]
J. D. Currie and N. Rampersad, A proof of Dejean’s conject ure, Math. Comp. 80 (2011), 1063–1070
2011
-
[10]
Damanik and D
D. Damanik and D. Lenz, The index of Sturmian sequences, European J. Combinatorics 23 (2002), 23–29
2002
-
[11]
de Luca, A
A. de Luca, A. Glen, L. Q. Zamboni, Rich, Sturmian, and tr apezoidal words, Theoret. Comput. Sci. 407 (2008), 569–573
2008
-
[12]
Dejean, Sur un th´ eor` eme de Thue, J
F. Dejean, Sur un th´ eor` eme de Thue, J. Combin. Theory Ser. A 13 (1972), 90–99
1972
-
[13]
Droubay, J
X. Droubay, J. Justin, G. Pirillo, Episturmian words an d some constructions of de Luca and Rauzy, Theoret. Comput. Sci. 255 (2001), 539–553
2001
-
[14]
Glen and J
A. Glen and J. Justin, Episturmian words: A survey, RAIR O – Theoret. Inform. Appl. 43 (2009), 403–442
2009
-
[15]
A. Glen, J. Justin, S. Widmer, L. Q. Zamboni, Palindromi c richness, European J. Combinatorics 30 (2009), 510-531
2009
-
[16]
A. Hof, O. Knill, B. Simon, Singular continuous spectru m for palindromic Schr¨ odinger operators, Comm. Math. Phys. 174 (1995), 149–159
1995
-
[17]
Justin and G
J. Justin and G. Pirillo, Fractional powers in Sturmian words, Theoret. Comput. Sci. 255 (2001), 363–376
2001
-
[18]
Karhum¨ aki and J
J. Karhum¨ aki and J. Shallit, Polynomial versus expone ntial growth in repetition-free binary words, J. Combin. Theory Ser. A 105 (2004), 335–347
2004
-
[19]
Lothaire, Algebraic Combinatorics on Words, Cambri dge, 2002
M. Lothaire, Algebraic Combinatorics on Words, Cambri dge, 2002
2002
-
[20]
Medkov´ a, E
K. Medkov´ a, E. Pelantov´ a, L. Vuillon, Derivated sequ ences of complementary symmetric Rote words. Preprint available at https://arxiv.org/abs/1812.03748
-
[21]
Mohammad-Noori and J
M. Mohammad-Noori and J. D. Currie, Dejean’s conjectur e and Sturmian words, European J. Com- bin. 28 (2007), 876–890
2007
-
[22]
Moulin-Ollagnier, Proof of Dejean’s conjecture for alphabets with 5, 6, 7, 8, 9, 10, and 11 letters, Theoret
J. Moulin-Ollagnier, Proof of Dejean’s conjecture for alphabets with 5, 6, 7, 8, 9, 10, and 11 letters, Theoret. Comput. Sci. 95 (1992), 187–205
1992
-
[23]
J. J. Pansiot, A propos d’une conjecture de F. Dejean sur les r´ ep´ etitions dans les mots, Discrete Appl. Math. 7 (1984), 297–311
1984
-
[24]
Pelantov´ a and ˇStˇ ep´ an Starosta, Languages invariant under more symmetries: Overlapping factors versus palindromic richness, Discrete Math
E. Pelantov´ a and ˇStˇ ep´ an Starosta, Languages invariant under more symmetries: Overlapping factors versus palindromic richness, Discrete Math. 313 (2013), 24 32–2445. 16 James D. Currie, Lucas Mol, Narad Rampersad
2013
-
[25]
Peltom¨ aki, Characterization of repetitions in Sturmian words: A new proof
J. Peltom¨ aki, Characterization of repetitions in Sturmian words: A new proof. Inform. Process. Lett. 115 (2015), 886–891
2015
-
[26]
Peltom¨ aki, Privileged W ords and Sturmian W ords, PhD thesis, TUCS Dissertations No
J. Peltom¨ aki, Privileged W ords and Sturmian W ords, PhD thesis, TUCS Dissertations No. 214, Au- gust 2016. Available at http://www.utupub.fi/handle/10024/124473
2016
-
[27]
Rote, Sequences with subword complexity 2n, J
G. Rote, Sequences with subword complexity 2n, J. Number Theory 46 (1994), 196–213
1994
-
[28]
Rampersad, J
N. Rampersad, J. Shallit, ´E. V andomme, Critical exponents of infinite balanced words, Theoret. Comput. Sci 777 (2019), 454–463
2019
-
[29]
Rao, Last cases of Dejean’s conjecture, Theoret
M. Rao, Last cases of Dejean’s conjecture, Theoret. Com put. Sci. 412 (2011), 3010–3018
2011
-
[30]
Restivo and G
A. Restivo and G. Rosone, Burrows-Wheeler transform an d palindromic richness, Theoret. Comput. Sci. 410 (2009), 3018–3026
2009
-
[31]
Restivo and S
A. Restivo and S. Salemi, On weakly square free words, Bu ll. European Assoc. Theoret. Comput. Sci. 21 (1983), 49–56
1983
-
[32]
Restivo and S
A. Restivo and S. Salemi, Overlap free words on two symbo ls, in: M. Nivat and D. Perrin (Eds.), Automata on Infinite W ords, Lecture Notes in Computer Science, V ol. 192, Springer, Ber lin (1985), 198–206
1985
-
[33]
V esti, Extensions of rich words, Theoret
J. V esti, Extensions of rich words, Theoret. Comput. Sc i. 548 (2014), 14–24
2014
-
[34]
V esti, Rich square-free words, Theoret
J. V esti, Rich square-free words, Theoret. Comput. Sci . 687 (2017), 48–61
2017
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.