{"id":"bee511ff-2f52-4bf7-8193-eed0c1f7c48b","arxiv_id":"1908.02384","paper_version":2,"verdict":"REJECT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"The claimed congruences and automaticity results for the Thue-Morse and period-doubling continued fractions are unsupported because the key induction proposition is false.","lead":"What did the paper do: it claims two continued fractions built from the Thue-Morse and period-doubling sequences are algebraic modulo 4, making their coefficient sequences automatic. Why read it: the result would connect continued fractions, algebraic series, and finite automata, but the central proof is invalid.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the Reader's Proposition 3.2 base-case contradiction does not hold; the formulas agree with the recurrence at m=1 and m=2.","rationale":"The paper's central claim is that the Thue-Morse and period-doubling continued fractions are congruent modulo 4 to algebraic series, making their coefficient sequences 2-automatic. The proof rests on Proposition 3.2, whose explicit convergent formulas I verified against the paper's own recurrence at the m=1 and m=2 levels. The Reader's objection asserts that items 3, 5, and 7 fail at m=1 because of x^{-1} terms; this is incorrect: S_1=x, so the apparent pole cancels, and each stated value matches the directly computed P2, Q2, Q3. I also spot-checked the nontrivial induction identity used in item 3 and found it consistent modulo 4. The direct Cartier-operator proofs of Theorems 1.4 and 1.5 provide independent support for automaticity without relying on Theorem 1.3. Minor presentational typos exist (e.g., the summation index in Lemma 3.4 appears to start at k=1 rather than k=0), but they do not affect the validity of the central congruences. Since the sole stated reason for rejection does not land, I recommend accepting the paper, subject to ordinary refereeing for typographical errors.","tokens_in":21222,"tokens_out":45712,"duration_ms":403203,"concrete_test":"Recompute P2, Q2, Q3 from P0=P1=Q0=1, Q1=1-x and P_n=P_{n-1}+a_n x P_{n-2}, Q_n=Q_{n-1}+a_n x Q_{n-2}, then compare with Proposition 3.2 items 3, 5, 7 for m=1; as a second check, recompute P14 and Q14 for m=2 from the same recurrence.","verdict_should_be":"ACCEPT","load_bearing_attack":"The Reader's claimed contradiction is not a contradiction. From the paper's recurrence, P2=1-x, Q2=1-2x, Q3=1-x-x^2. Proposition 3.2 items 3 and 7 with m=1 give 1+x^{-1}S_1^2-2S^e_1 = 1-x and 1+2S_1=1+2x (congruent to 1-2x mod 4), and item 5 gives 1-x+2x^2-S_1^2+2xS^e_1 = 1-x-x^2. All match the recurrence polynomials. The apparent x^{-1} poles are absent because S_m begins with x, so x^{-1}S_m^2 is a polynomial. I also checked P14 and P15 against the m=2 cases. The induction step's equality x^{-1}T^4 ≡ x^{-1}(S_{2m-1}-x)^2 (mod 4) is valid via T^2=(S_{2m-1}-x)+2b. I therefore find no load-bearing gap in the central automaticity argument.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies two formal Stieltjes continued fractions C(x) and D(x) whose coefficients are respectively the Thue-Morse and period-doubling sequences. It claims that modulo 4 each of these series is congruent to an explicit algebraic power series (Theorems 1.1 and 1.2) and consequently, by the Denef-Lipschitz theorem, that the sequences of coefficients of their power series expansions are 2-automatic. The main technical work is Proposition 3.2, which gives explicit formulas for eight subsequences of the numerator and denominator polynomials of the convergents of C(x); these formulas are proved by an induction and then passed to the limit. The paper also computes explicit 2-kernels for the reduced coefficient sequences, proves a lower bound on the algebraic degree modulo 2 (Theorem 1.6), and derives the 2-automaticity of the corresponding Hankel determinants.","tokens_in":21457,"tokens_out":36477,"duration_ms":299126,"significance":"If the proofs are correct, this is a substantial contribution to the converse direction of the continued-fraction/automatic-sequence literature: automatic sequences are used as partial quotients, and the resulting generating series are shown to be algebraic modulo a fixed prime power. The paper gives explicit, parameter-free algebraic closed forms and carries out the Cartier-operator computations that exhibit finite kernels. The proof of Proposition 3.2 is intricate and is the load-bearing part of the argument; the period-doubling case is then handled by a contraction argument rather than by repeating the induction. The algebraic-degree result and the Hankel-determinant corollaries are natural and nontrivial additions.","major_comments":[{"comment":"The alleged base-case failure of Proposition 3.2 does not occur. With the intended reading of the exponent in item 5 as 2^{2m-1} rather than 2^{2m}-1, the m=1 values are P_2=1-x, Q_2≡1-2x, and Q_3≡1-x-x^2 modulo 4, all matching the recurrence of Lemma 3.1. The same reading is forced by the induction step in the proof of item 5. I therefore see no load-bearing gap in the derivation of congruence (1.3).","section":"§3, Proposition 3.2"}],"minor_comments":[{"comment":"The exponent notation in Proposition 3.2 is very easy to misread. In particular, item 5 should be typeset with explicit braces, e.g. 2x^{2^{2m-1}}, because the unbraced rendering can be read as 2x^{2^{2m}-1}. I recommend adding parentheses or braces to all exponents in this proposition.","section":"§3, Proposition 3.2"},{"comment":"The sentence beginning 'The constant term of Q_{2^{2m}+2}(x) being 1' appears to be a typo: the next line uses Q_{2^{2m}-2}(x), and the constant term of that denominator is indeed 1.","section":"§3, Proposition 3.3"},{"comment":"In the sentence 'Then our goal (1.4) can be written as D(x) ≡ (H_1(x)H_3(x)-1)/x', the factor H_1 should be H_2; the subsequent calculation correctly uses H_2(-x^2)H_3(-x^2).","section":"§4, proof of Theorem 1.2"},{"comment":"In the irreducibility check for y^4+y+x over F_2[x], the statement that a linear factor must be y+x or y+1 is too quick. The standard degree argument is that a root a ∈ F_2[x] would satisfy a^4+a=x, which is impossible by degree comparison; please include that justification.","section":"§5"},{"comment":"Minor language: 'The right hand side of congruence (1.3) and (1.4) are respectively' should be 'is respectively'; also, the displayed polynomials in Section 5 would benefit from consistent notation such as x^1 and x^2 instead of x1 and x2.","section":"§1"}],"recommendation":"minor_revision","confidential_remarks":"The rejection in the reader's report is not supported: the claimed base-case failure of Proposition 3.2 comes from misreading the exponent in item 5 as 2^{2m}-1 instead of 2^{2m-1}. The central congruence (1.3) is not undermined. The remaining issues are typographical and local. I recommend minor revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nYou'll want to know about this paper before believing the desk reject: the reader's central objection does not survive contact with the definitions. The claimed base-case contradiction in Proposition 3.2 is based on reading S_m(x) as starting with constant 1; actually S_m(x)=x+x^2+...+x^{2^{m-1}}, so S_1=x. That means x^{-1}S_1^2=x, not x^{-1}, and item 3 gives 1+x-2x=1-x, which is exactly P_2 from the recurrence. Item 7 gives 1+2x≡1-2x=Q_2, and item 5 gives 1-x+2x^2-x^2+2x^2=1-x-x^2=Q_3. So the m=1 case is consistent. I also spot-checked m=2 via the induction formulas; they agree with the recurrence. The proof is tedious but the machinery is coherent.\n\nWhat's genuinely new: explicit algebraic congruences for the Thue-Morse and period-doubling continued fractions modulo 4, leading to automaticity of coefficient sequences; explicit 2-kernel tables; a degree-4 minimality result; and automatic Hankel determinants. The method—guessing eight subsequence formulas and proving them by mutual induction—is a legitimate, if dense, route. The contraction trick linking C(x) and D(x) is neat. The citation pattern is appropriate and not self-inflated; the earlier Hankel work is background, not load-bearing.\n\nSoft spots, in proportion: The induction proof of Proposition 3.2 is long and written in a compressed style; a referee should check the algebra line by line, but there's no obvious gap. The base-case sentence 'Relations 1)-8) are true for m≡0 or m≡1' is sloppy, because the ranges differ per item, and for m=0 some formulas involve S_{-1}. That should be cleaned up. The proof of Theorem 1.6's irreducibility is a bit telegraphic but valid. I didn't find a fatal flaw.\n\nWho it's for: specialists in automatic sequences, finite-field continued fractions, and Hankel determinants. It answers an interesting converse question and gives concrete new examples. It deserves a real referee. My recommendation: send it out; expect a substantial but fixable report.\n\nBest.","headline":"A real new result on automaticity of continued fractions; the reader's rejection is based on a misreading of S_m, and the main induction survives.","tokens_in":22034,"tokens_out":4684,"would_cite":true,"duration_ms":42969,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11B85","11J70","11B50","11Y65","05A15"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that the Thue-Morse and period-doubling continued fractions are congruent modulo 4 to explicit algebraic power series, so their coefficient sequences modulo 4 are 2-automatic.","keywords":["automatic sequence","continued fraction","Thue-Morse sequence","period-doubling sequence","Hankel determinant","2-automatic","formal power series","congruence modulo 4"],"falsifier":"Compute the first convergents directly from the recurrence: P_0=P_1=Q_0=1, Q_1=1-x, P_2=1-x, Q_2=1-2x, Q_3=1-x-$x^{2}$. Proposition 3.2 items 3), 5), and 7) for m=1 give expressions such as -1+$x^{{-1}}$, 1+x+$2x^{2}$, and 3, which do not match these polynomials. A reader can check this disagreement in two lines and thereby determine whether the announced congruences are supported.","tokens_in":20943,"feed_emoji":"🔢","tokens_out":9675,"duration_ms":90443,"temperature":0.7,"pith_summary":"This paper studies the converse of the usual automatic-sequence/continued-fraction question: instead of asking what continued fractions automatic sequences produce, it takes two continued fractions whose partial quotients are the automatic Thue-Morse and period-doubling sequences and asks what power series they define. The main claim is that the two series C(x) and D(x) are congruent modulo 4 to explicit algebraic series over Z, so by the Denef-Lipschitz theorem their coefficient sequences modulo 4 are 2-automatic, meaning each coefficient can be generated by a finite automaton reading the index in binary. The proof is a direct calculation: the authors guess closed forms for certain subsequences of the canonical convergents P_n/Q_n, prove them by an induction involving eight coupled subsequences, then pass to the limit to identify C(x) and D(x) modulo 4. If correct, these are concrete examples of formal power series that are algebraic modulo a prime power while being generated by automatic continued fractions.","feed_headline":"Thue-Morse and period-doubling continued fractions are algebraic mod 4","feed_subtitle":"Their coefficients modulo 4 are 2-automatic, a new link between continued fractions and automatic sequences.","key_machinery":"The load-bearing object is the pair (P_n(x),Q_n(x)) of canonical convergents of the Stieltjes continued fraction, with C(x)=lim P_n/Q_n, together with eight coupled subsequences P_{2^k-1}, P_{2^k-2}, Q_{2^k-1}, Q_{2^k-2} expressed modulo 4 in terms of truncated lacunary sums S_m=sum_{j=0}^{m-1} $x^{{2^j}}$, S^e_m=sum $x^{{2^{2j}}$}, and S^o_m=sum $x^{{2^{2j+1}}$} (Proposition 3.2). The induction over these eight subsequences produces the explicit limit series; the Cartier operators Lambda_0 and Lambda_1 then convert that series into a finite 2-kernel, proving automaticity directly. The contraction theorem for Stieltjes and Jacobi continued fractions turns the Thue-Morse congruence into the period-doubling congruence, and the same explicit series feeds the degree-4 annihilator and Hankel determinant arguments.","core_discovery":"The central discovery is Theorem 1.1 and Theorem 1.2: with C(x)=sum c_n x^n the Thue-Morse Stieltjes continued fraction and D(x)=sum d_n x^n the period-doubling one, the paper proves the congruences C(x) equivalent to ($\\sqrt$(1-4x)-1)/(2x)+1+$\\sqrt$(2 $\\sqrt$(1-4x)-1) modulo 4 and D(x) equivalent to ((1+$\\sqrt$(1+4x))($\\sqrt$(2 $\\sqrt$(1-$4x^{2}$)-1)-2))/(2x) modulo 4. As a direct corollary, via the Denef-Lipschitz theorem, the reduced coefficient sequences (c_n mod 4) and (d_n mod 4) are 2-automatic, and the paper explicitly computes their 2-kernels, which have 9 states and 5 states respectively. The paper further shows that both reduced series satisfy the degree-4 polynomial S(x,y)=($xy^{2}$+y+1)^2 in Z/4Z[x,y], that no lower-degree polynomial with invertible leading coefficient in the Laurent series ring annihilates either series, and that the Hankel determinant sequences of C(x) and D(x) are 2-automatic.","pith_inferences":["One testable extension is to run the same eight-subsequence scheme on the m=3 kernels reported in the paper; if the induction base is repaired or re-indexed, the m=3 automaticity could likely be proved directly, giving the first case beyond modulo 4 of Conjecture 1.9.","The explicit 2-kernels are self-contained objects: they let a reader generate coefficients modulo 4 by iterating a two-letter transducer without computing continued fractions, so any finite prefix of (c_n mod 4) or (d_n mod 4) can be checked independently against the algebraic congruences.","If the closed forms in Proposition 3.2 are adjusted to hold for m >= 1, the same limit argument would go through unchanged, since the infinite-series identity is what the induction ultimately approximates."],"forward_implications":["The coefficient sequences modulo 4 are 2-automatic, with explicit transducers of 9 states for (c_n mod 4) and 5 states for (d_n mod 4).","Both reduced series are algebraic of degree 4 over Z/4Z[x], sharing the same annihilating polynomial (xy^2+y+1)^2 and admitting no lower-degree annihilator with invertible leading Laurent-series coefficient.","The Hankel determinant sequences of C(x) and D(x) are 2-automatic.","The contraction identity C(x)=1/(1-x-x^2 D(-x^2)) makes the period-doubling theorem a formal consequence of the Thue-Morse theorem, so the two automaticity results stand or fall together.","These examples show that automatic continued fractions can define series that are algebraic modulo a prime power, in contrast with Bugeaud's theorem for real algebraic numbers of degree at least three."],"supporting_citations":[{"why":"Supplies the Denef-Lipschitz theorem that turns algebraicity of a power series over Z_p into p-automaticity of its coefficient sequence modulo p^alpha.","marker":"[16]"},{"why":"Provides the congruence sqrt(1-4x) equivalent to 1+2 sum x^{2^k} modulo 4 and the Hankel determinant calculus used in the proof of Theorem 1.1.","marker":"[22]"},{"why":"Supplies the contraction theorem equating Stieltjes and Jacobi continued fractions, the step that converts the Thue-Morse congruence into the period-doubling congruence.","marker":"[35,30,32]"},{"why":"Gives Heilermann's formula for Hankel determinants of Stieltjes and Jacobi continued fractions, used in the proof of Theorem 1.8.","marker":"[24]"},{"why":"Provides the algebraic-iff-automatic criterion for finite fields and the 2-kernel/minimal-polynomial method used in Theorem 1.6.","marker":"[14]"},{"why":"States the closure property of automatic sequences under prefix products, used to prove automaticity of the Hankel determinant sequences.","marker":"[3]"}],"fun_headline_variants":["Thue-Morse and period-doubling continued fractions algebraic mod 4","Coefficients of Thue-Morse and period-doubling CFs are 2-automatic mod 4","Algebraic mod 4: Thue-Morse and period-doubling continued fractions","Thue-Morse and period-doubling CFs: algebraic mod 4, 2-automatic coeffs"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The explicit closed-form formulas in Proposition 3.2 for the convergent subsequences P_{2^k-1}, P_{2^k-2}, Q_{2^k-1}, Q_{2^k-2}, claimed for all m >= 1, are the load-bearing step; the base case m=1 already conflicts with the recurrence defining P_n and Q_n, and if these formulas are not true, the limit computation of C(x), the congruences (1.3) and (1.4), and the subsequent automaticity results do not follow.","fun_headline_variants_meta":{"raw":{"variants":["Thue-Morse and period-doubling continued fractions algebraic mod 4","Coefficients of Thue-Morse and period-doubling CFs are 2-automatic mod 4","Algebraic mod 4: Thue-Morse and period-doubling continued fractions","Thue-Morse and period-doubling CFs: algebraic mod 4, 2-automatic coeffs"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.002007,"raw_usage":{"total_tokens":7844,"prompt_tokens":976,"completion_tokens":6868,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":592,"completion_tokens_details":{"reasoning_tokens":6767}},"tokens_in":592,"tokens_out":6868,"duration_ms":47355,"temperature":1.0,"reasoning_tokens":6767,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:49:07.712470+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the first convergents directly from the recurrence: P_0=P_1=Q_0=1, Q_1=1-x, P_2=1-x, Q_2=1-2x, Q_3=1-x-$x^{2}$. Proposition 3.2 items 3), 5), and 7) for m=1 give expressions such as -1+$x^{{-1}}$, 1+x+$2x^{2}$, and 3, which do not match these polynomials. A reader can check this disagreement in two lines and thereby determine whether the announced congruences are supported.","supporting_citations":[{"cited_title":"An irrationality measure for regular paperfolding numbers","cited_arxiv_id":null,"evidence_quote":"Supplies the Denef-Lipschitz theorem that turns algebraicity of a power series over Z_p into p-automaticity of its coefficient sequence modulo p^alpha."},{"cited_title":"On the irration ality exponent of the regular paperfolding numbers","cited_arxiv_id":null,"evidence_quote":"Provides the congruence sqrt(1-4x) equivalent to 1+2 sum x^{2^k} modulo 4 and the Hankel determinant calculus used in the proof of Theorem 1.1."},{"cited_title":"Hankel continued fraction and its applica tions","cited_arxiv_id":null,"evidence_quote":"Gives Heilermann's formula for Hankel determinants of Stieltjes and Jacobi continued fractions, used in the proof of Theorem 1.8."},{"cited_title":"Hankel determinants, Pad´ e approximations, and irrationality exponents","cited_arxiv_id":null,"evidence_quote":"Provides the algebraic-iff-automatic criterion for finite fields and the 2-kernel/minimal-polynomial method used in Theorem 1.6."},{"cited_title":"Quasicry stal Ising chain and automata theory","cited_arxiv_id":null,"evidence_quote":"States the closure property of automatic sequences under prefix products, used to prove automaticity of the Hankel determinant sequences."}],"review_version":1}