{"id":"82792814-29bb-4d3b-b308-fcd76020e238","arxiv_id":"2508.13420","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A complete characterization of recurrent pattern Sturmian sequences as either simple circle rotation codings or members of nearly simple Toeplitz subshifts.","lead":"Mathematicians fully classified the one-sided infinite binary sequences with the minimal possible pattern complexity: they are exactly the two known families, circle-rotation codings and near-Toeplitz sequences. The result settles a question posed in 2002 and may unlock spectral proofs for Schroedinger operators with such potentials.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 5.4's reduction to simple Toeplitz subsequences is not justified; without it the odometer case of Theorem A is open.","rationale":"The reader's weakest assumption identifies exactly the step I find load-bearing. The proof of Theorem A splits into the circle-rotation case and the odometer case. The odometer case is handled by Proposition 5.4, which assumes |B|=2 and aims to contradict this unless j_k-i_k=n_k/2, ultimately reducing to a 1-hole Toeplitz. That contradiction depends on Lemma 5.3, whose hypothesis is that x' and x'' are simple 1-hole Toeplitz sequences with a common period structure. The paper attempts to reach this hypothesis by truncating the first period and then passing to a subsequence of (n_k). Neither step is justified in the text: truncating a period structure can lose the partition property of Definition 2.12, and passing to a subsequence can turn a simple Toeplitz sequence into a non-simple one because fill letters from skipped levels are merged at the coarser level. The paper cites [14] for the truncation claim, but does not reproduce or prove the needed statement. If this reduction fails, the entire odometer branch of Theorem A is unsupported. I therefore agree with the CONDITIONAL verdict and would keep it; the concern is serious but plausibly repairable by a more careful argument or by adapting Lemma 5.3 to nearly simple sequences.","tokens_in":26088,"tokens_out":25265,"duration_ms":240593,"concrete_test":"Test the reduction on a concrete simple Toeplitz sequence: let y have period structure n_k=2^k and fill letters a_k = k mod 2 (so y(n) is determined by the 2-adic valuation of n+1). Pass to the cofinal subsequence n'_j=4^j and recompute the minimal-level partition of Definition 2.12. If y is not simple with respect to (n'_j) (because positions first filled at original levels 1 and 2 appear at the same subsequence level with different letters), then 'passing to a subsequence' in Proposition 5.4 does not preserve the simple hypothesis needed for Lemma 5.3, confirming the gap.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the |B|=2 odometer case of Proposition 5.4, after forming x'=x(i0,n0) and x''=x(j0,n0), the proof invokes Theorem 2.13 and asserts that its proof in [14] shows the nearly-simple decomposition can be taken modulo the first period, so 'by truncating the first term from (n_k), we may assume x' and x'' are simple.' This is not proved here, and it is not obvious: a cofinal subsequence of a Toeplitz period structure need not be a period structure under Definition 2.12, and even when it is, the simple-Toeplitz property is not preserved under passing to a subsequence (the fill letters of positions whose minimal level is skipped can differ). The subsequent 'by passing to a subsequence of (n_k) if necessary, that a_k=b_k for all k and alternates' makes the same problem: Lemma 5.3 requires x' and x'' to be simple with the new period structure, but the subsequence can destroy simplicity. Since the contradiction forcing j_k-i_k=n_k/2 and the final reduction to a 1-hole Toeplitz both rely on Lemma 5.3, this gap is load-bearing for the Toeplitz half of Theorem A.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper classifies recurrent binary sequences whose maximal pattern complexity attains the minimum possible growth p*_x(n)=2n. Theorem A states that a recurrent sequence is pattern Sturmian exactly when it is either a recurrent simple circle rotation coding sequence or a sequence in a nearly simple Toeplitz subshift, thereby answering Kamae and Zamboni's Problem 1. Theorem B gives a near characterization of nonrecurrent pattern Sturmian sequences as either nonrecurrent simple circle rotation codings or almost constant sequences. The technical engine is a structural result, Theorem C, showing that recurrent sequences with non-superlinear maximal pattern complexity generate minimal subshifts whose maximal equicontinuous factor is either a circle times a finite cyclic group or an odometer, together with Theorem D, which proves a lower bound for recurrent but not uniformly recurrent sequences. The circle case is handled in Proposition 5.2, and the odometer case is handled in Proposition 5.4 via Lemma 5.3.","tokens_in":26306,"tokens_out":24177,"duration_ms":257926,"significance":"If the proof can be completed, Theorems A and B would close a natural question from 2002 and would show that all previously known pattern Sturmian examples are exhaustive. The MEF-based approach is a genuine methodological innovation, and Theorem D is a new and independently useful complexity lower bound. The paper is careful to rely on external results only as stated, and several folklore facts are proved in-line, which is a strength. However, the odometer case in Proposition 5.4 contains an unproved and load-bearing reduction, and the definition of simple Toeplitz in Definition 2.12 is in tension with the way Lemma 5.3 uses alternating fill letters. These issues affect the Toeplitz half of Theorem A, so the current version is not yet publishable.","major_comments":[{"comment":"The reduction to simple Toeplitz subsequences is not proved. After forming x' and x'', the paper invokes 'the proof of Theorem 2.13 in [14]' to assert that the decomposition can be taken modulo the first period, and then says that 'by truncating the first term from (n_k)' one may assume x' and x'' are simple. This is precisely the step where a cofinal subsequence of a Toeplitz period structure can fail to be a 1-hole simple period structure: positions whose first covering level is skipped may become holes, or may split into several constant progressions, and Lemma 5.3 requires the same period structure for x' and x''. The later step 'by passing to a subsequence of (n_k) if necessary, that a_k=b_k for all k and alternates' has the same problem. Since the contradiction forcing j_k-i_k=n_k/2 and the final reduction to a 1-hole Toeplitz both rely on Lemma 5.3, this gap is load-bearing for the Toeplitz half of Theorem A. Please supply a complete proof, or restructure the argument so that no culling of periods is used.","section":"Proposition 5.4 (§5)"},{"comment":"Definition 2.12 and Lemma 5.3 appear inconsistent under the literal wording of 'simple Toeplitz.' If a residue class r modulo n_k is constant with value a_k, then every subresidue r+t n_k modulo n_{k+1} is also constant with the same value; induction then forces a_k=a_{k+1}, so the choices in Lemma 5.3 with a_c=0 and a_d=1 for c<d are impossible. If the authors intend the fill letters a_k to be allowed to alternate, the definition must be reworded, for example by restricting 'constant progressions' to the progressions selected in the defining partition, and Lemma 5.3's indexing convention must be reconciled with that wording. As written, the proof of Lemma 5.3 relies on an interpretation of simple Toeplitz that is not stated in Definition 2.12.","section":"Definition 2.12 / Lemma 5.3 (§2.2.2, §5)"}],"minor_comments":[{"comment":"In Definition 2.2, the orbit closure is written identically to the orbit; please use an overline or another notation to distinguish Orb(x) from its closure.","section":"Definition 2.1-2.2"},{"comment":"The notation p_X^*(n) is used for subshifts, but p^* was defined only for sequences; please define p_X^*(n) explicitly.","section":"Theorem 3.1"},{"comment":"The formula for the number of holes at step k is hard to parse because the notation |{a_{j,i}}_i| is not defined clearly; please state it in terms of the number of distinct selected progressions at each level.","section":"Definition 2.12"},{"comment":"The phrase 'nonsimple 1-hole Toeplitz' may confuse readers because two constant progressions are listed at each level; please explain why the hole count is still one despite the multiple fill letters.","section":"Example 3.8"},{"comment":"The abstract writes x in A^N while the body consistently uses N0; please make the indexing conventions uniform.","section":"Abstract and §2.1"}],"recommendation":"major_revision","confidential_remarks":"This is a strong paper in conception and breadth, and I would support publication if the Proposition 5.4 gap is closed and Definition 2.12 is cleaned up. I found no circularity concerns: the self-citations [4,5] are contextual, and the main arguments use external results only as named ingredients. The referee report above states the load-bearing issues."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis paper is a serious piece of work. It answers Kamae and Zamboni's 2002 question on recurrent pattern Sturmian sequences, and it does so with a technique that is genuinely new: using the cardinality of the boundary of the MEF partition to control maximal pattern complexity. Theorem 4.1 is elegant, Theorem D (recurrent but not uniformly recurrent sequences have complexity growing at least like n log n) is a nice standalone result, and the nonrecurrent case (Theorem B) is handled well. The paper is clearly written and the overall architecture is coherent.\n\nThe soft spot is Proposition 5.4, which is the odometer half of Theorem A. In the |B|=2 case, the proof reduces to a contradiction by truncating the period structure and passing to a subsequence so that the two 1-hole Toeplitz sequences x' and x'' are simple and have the same alternating letter sequence. Neither reduction is proved. The first cites the proof of Theorem 2.13 in [14]; that may be true, but it is not a consequence of the theorem as stated, and the paper gives the reader no way to check it. The second, 'by passing to a subsequence ... that a_k = b_k for all k and alternates', does not follow. An arbitrary binary sequence has an alternating subsequence only if it is not eventually constant; nothing in the preceding argument rules out a_k eventually constant. Since the contradiction that forces j_k - i_k = n_k/2 and the final reduction to |B|=1 both rely on Lemma 5.3 and the alternating assumption, this gap is load-bearing. The Toeplitz half of Theorem A is not fully established as written.\n\nI don't endorse every detail of the stress-test note: a cofinal subsequence of a period structure is still a period structure in the sense of Definition 2.12. But the main concern is real and needs to be addressed.\n\nThis gap is significant but looks fixable. The statements are probably true. It needs to be written out. Minor issues: a couple of typos (e.g., 'x,x' in the text of Prop 5.4) and the paper defers the technical condition in Theorem 2.13 to [14], which is acceptable but slows verification.\n\nWho this is for: specialists in symbolic dynamics and low-complexity subshifts, and people working on Schrödinger operators with aperiodic potentials, since the classification enables the head-on approach the introduction describes. I'd send it to a serious referee, with instructions to focus on Proposition 5.4. It should be a reading-group candidate if you want to see the MEF boundary method in action.","headline":"Strong paper that likely answers the Kamae-Zamboni question, but the odometer half of Theorem A rests on an unjustified reduction in Proposition 5.4.","tokens_in":26890,"tokens_out":9264,"would_cite":true,"duration_ms":88620,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37B10","37B05"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every recurrent pattern Sturmian sequence is one of two known types, closing a 2002 classification question.","keywords":["maximal pattern complexity","pattern Sturmian sequences","symbolic dynamics","Toeplitz subshifts","circle rotations","maximal equicontinuous factor","null systems","odometers"],"falsifier":"Take a recurrent 2-hole Toeplitz sequence with period structure $(n_k)$ whose two nonconstant residue classes satisfy $j_k-i_k\\neq n_k/2$ for some $k$, and compute its maximal pattern complexity at a suitable window: Proposition 5.4 predicts $p^*_x(3)\\ge 7$, so exhibiting such a sequence with exactly six 3-letter patterns would refute the odometer half of the classification.","tokens_in":25859,"feed_emoji":"🌀","tokens_out":6330,"duration_ms":58615,"temperature":0.7,"pith_summary":"This paper closes a classification problem left open in 2002: it proves that every recurrent sequence of minimal maximal pattern complexity—those for which $p^*_x(n)=2n$ for every $n$, called pattern Sturmian—is either a coding of an irrational circle rotation by two intervals or a member of a nearly simple Toeplitz subshift. These were the two known families of examples, so the answer says there are no others. For nonrecurrent pattern Sturmian sequences the paper gives a near characterization: they are either nonrecurrent circle-rotation codings or almost constant sequences, with no other possibilities. The proof introduces a structural method, analyzing the boundary set of a partition of the maximal equicontinuous factor of the orbit closure, and yields a broader theorem about all recurrent sequences whose maximal pattern complexity grows at most linearly.","feed_headline":"Recurrent Sturmian patterns: only two families exist","feed_subtitle":"A 2002 question by Kamae and Zamboni is answered—circle codings and nearly simple Toeplitz sequences cover every case.","key_machinery":"The load-bearing object is the maximal equicontinuous factor (MEF) of the orbit closure $X$ of $x$: since pattern Sturmian subshifts are null, $X$ is an almost 1-1 extension of a group rotation, and Theorem 3.4 represents $x$ as the coding of the orbit of a point by a partition $G=U_0\\cup U_1\\cup B$ whose common boundary $B$ is finite when complexity is non-superlinear. The proof shows the size of $B$ controls $p^*_X$: if $|B|\\ge k$ then $p^*_X(n)-kn$ is bounded below, so minimal growth $p^*=2n$ forces $|B|\\le 2$. A structural argument limits the possible MEFs to a circle times a finite cyclic group or an odometer; in the circle case $B$ must have exactly two points, yielding circle-rotation codings, while in the odometer case a 3-window combinatorial lemma forces an MEF partition with $|B|=1$, which by the known characterization of pattern Sturmian 1-hole Toeplitz sequences yields the nearly simple Toeplitz conclusion.","core_discovery":"The central claim is Theorem A: for $x\\in\\{0,1\\}^{\\mathbb{N}_0}$, if $x$ is recurrent then $x$ is pattern Sturmian if and only if $x$ is a recurrent simple circle rotation coding sequence or belongs to a nearly simple Toeplitz subshift. The converse directions were already known; the new content is that these exhaust the recurrent case. Theorem B says a nonrecurrent pattern Sturmian $x$ is either a nonrecurrent simple circle rotation coding sequence or almost constant, and Theorem C describes the larger class of recurrent non-superlinear complexity sequences as finite interleavings of circle-rotation codings sharing one irrational rotation, or as elements of $m$-hole Toeplitz subshifts. Theorem D adds that a recurrent sequence that is not uniformly recurrent must have $\\liminf p^*_x(n)/(n\\ln n)>0$, so within recurrent pattern Sturmian sequences uniform recurrence is automatic.","pith_inferences":["The boundary-counting mechanism suggests a template for larger alphabets: if minimal nonperiodic complexity for $r$-letter sequences is $rn$, the same MEF partition argument might classify those sequences, provided an analogue of the 3-window lemma exists.","Theorem D's $n\\ln n$ lower bound gives a concrete target for the open question of recurrent null subshifts that are not uniformly recurrent: such a subshift, if it exists, must sit exactly in the window between linear and $n\\ln n$ complexity.","The proof's dependence on dropping periods in the odometer case points to a possible simplification: a direct proof that period-dropping preserves the simple Toeplitz property would remove the need for the subsequence argument in Proposition 5.4."],"forward_implications":["Every recurrent pattern Sturmian sequence is uniformly recurrent; non-uniformly recurrent recurrent sequences have complexity at least of order $n\\ln n$, so they cannot be pattern Sturmian.","The previously known families are exhaustive for recurrent sequences, so any future example must be a circle coding or a nearly simple Toeplitz sequence.","Nonrecurrent pattern Sturmian sequences are almost constant or circle codings; the only remaining open subproblem is which almost constant sequences qualify.","Combined with known spectral results, the classification leaves exactly one case to be settled for Schrödinger operators with pattern Sturmian potentials: the simple circle rotation coding sequences."],"supporting_citations":[{"why":"Defines maximal pattern complexity and pattern Sturmian sequences, proves the $2n$ lower bound (Theorem 2.7), and poses Question 1.1 that the paper answers.","marker":"[20]"},{"why":"Provides Theorem 2.13, the characterization of pattern Sturmian 1-hole Toeplitz sequences that the odometer case invokes to conclude nearly simple Toeplitz.","marker":"[14]"},{"why":"Supplies the earlier claim that simple Toeplitz sequences are pattern Sturmian, one of the two known families.","marker":"[19]"},{"why":"Gives the theorem that minimal null systems are almost 1-1 extensions of their maximal equicontinuous factor, the starting point of the structural proof.","marker":"[17]"},{"why":"Co-source for the same null-system almost 1-1 extension theorem used throughout Section 3.","marker":"[21]"},{"why":"Supplies the folklore result (Theorem 3.4) that an almost 1-1 extension of a rotation is represented by a two-set partition with common boundary, the coding device underlying the MEF partition.","marker":"[12]"},{"why":"Provides the path-component structure of solenoids used in Lemma 4.4 to rule out higher-dimensional or solenoidal MEFs.","marker":"[24]"}],"fun_headline_variants":["Recurrent pattern Sturmian: only two families","Two types cover recurrent Sturmian sequences","Pattern Sturmian recurrent sequences classified","MEF proof: recurrent Sturmian patterns are two known types","Answering Kamae–Zamboni: recurrent Sturmian patterns closed"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"In the odometer case, the proof assumes that when a 2-hole Toeplitz pattern Sturmian sequence's two one-hole subsequences have matching letter sequences, one can pass to a subsequence of periods that keeps the subsequences simple Toeplitz and makes the letters alternate; if that reduction fails, the argument that the boundary has size one rather than two breaks.","fun_headline_variants_meta":{"raw":{"variants":["Recurrent pattern Sturmian: only two families","Two types cover recurrent Sturmian sequences","Pattern Sturmian recurrent sequences classified","MEF proof: recurrent Sturmian patterns are two known types","Answering Kamae–Zamboni: recurrent Sturmian patterns closed"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000158,"raw_usage":{"total_tokens":1248,"prompt_tokens":993,"completion_tokens":255,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":609,"completion_tokens_details":{"reasoning_tokens":176}},"tokens_in":609,"tokens_out":255,"duration_ms":3016,"temperature":1.0,"reasoning_tokens":176,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T17:16:44.008431+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a recurrent 2-hole Toeplitz sequence with period structure $(n_k)$ whose two nonconstant residue classes satisfy $j_k-i_k\\neq n_k/2$ for some $k$, and compute its maximal pattern complexity at a suitable window: Proposition 5.4 predicts $p^*_x(3)\\ge 7$, so exhibiting such a sequence with exactly six 3-letter patterns would refute the odometer half of the classification.","supporting_citations":[{"cited_title":"Kamae and L","cited_arxiv_id":null,"evidence_quote":"Defines maximal pattern complexity and pattern Sturmian sequences, proves the $2n$ lower bound (Theorem 2.7), and poses Question 1.1 that the paper answers."},{"cited_title":"Gjini, T","cited_arxiv_id":null,"evidence_quote":"Provides Theorem 2.13, the characterization of pattern Sturmian 1-hole Toeplitz sequences that the odometer case invokes to conclude nearly simple Toeplitz."},{"cited_title":"Kamae and L","cited_arxiv_id":null,"evidence_quote":"Supplies the earlier claim that simple Toeplitz sequences are pattern Sturmian, one of the two known families."},{"cited_title":"Huang, S","cited_arxiv_id":null,"evidence_quote":"Gives the theorem that minimal null systems are almost 1-1 extensions of their maximal equicontinuous factor, the starting point of the structural proof."},{"cited_title":"Kerr and H","cited_arxiv_id":null,"evidence_quote":"Co-source for the same null-system almost 1-1 extension theorem used throughout Section 3."},{"cited_title":"Downarowicz","cited_arxiv_id":null,"evidence_quote":"Supplies the folklore result (Theorem 3.4) that an almost 1-1 extension of a rotation is represented by a two-set partition with common boundary, the coding device underlying the MEF partition."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the path-component structure of solenoids used in Lemma 4.4 to rule out higher-dimensional or solenoidal MEFs."}],"review_version":2}