{"id":"0c8ceedc-b626-48c1-b938-14e269edd924","arxiv_id":"2412.10378","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper presents a flawed characterization of the lazy expansion of 1 in base beta in (1,2), with the main theorem not supported by the preceding lemmas.","lead":"An undergraduate dissertation claims to characterize the lazy expansion of 1 in non-integer bases between 1 and 2, answering an open problem from Erdős and Komornik. The proof has critical gaps and relies on a misstated version of the known uniqueness theorem.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 6.1(b) is false: for beta=3/2 the lazy expansion of 1 begins 0,1,0, and the claimed lexicographic inequality fails at k=2.","rationale":"The central claim of the paper is Theorem 6.1, a characterization of lazy expansions of 1. A single counterexample refutes part (b), so the claimed solution of the Erdős-Komornik problem is false as stated. The counterexample is internal: it uses the paper's Definition 4.2 of lazy expansion and basic geometric sums. The reader's weakest assumption about Lemma 6.1 points to a real proof defect, and the derivation of Theorem 6.1(b) from Lemma 6.2 is indeed absent, but the theorem is not merely under-proved; it is wrong. I would therefore keep the rejection. My agreement is partial because the reader located the key flaw in Lemma 6.1's inequality, whereas the decisive problem is the false statement of Theorem 6.1(b).","tokens_in":20839,"tokens_out":21459,"duration_ms":162441,"concrete_test":"Run the lazy algorithm from Definition 4.2 with beta = 3/2, x = 1 and compute the first three digits a_1, a_2, a_3. If they are (0,1,0), then check the lexicographic comparison for k=2: the first entry of (1 - a_{2+i}) is 1, while a_1 = 0, so the condition in Theorem 6.1(b) fails. This settles the falsehood.","verdict_should_be":"REJECT","load_bearing_attack":"According to Theorem 6.1(b), whenever beta in (1, phi] and (a_i) is the lazy expansion of 1, one must have (1 - a_{k+i}) < (a_i) for every k with a_k = 1. This is false. Take beta = 3/2 and x = 1. Using Definition 4.2: at i=1, sum_{j>1} beta^{-j} = 1/(beta(beta-1)) = 4/3 >= 1, so a_1 = 0. At i=2, sum_{j>2} beta^{-j} = 1/(beta^2(beta-1)) = 8/9 < 1, so a_2 = 1. At i=3, a_1/beta + a_2/beta^2 + sum_{j>3} beta^{-j} = 4/9 + 16/27 > 1, so a_3 = 0. For k=2, a_k = 1. The sequence (1 - a_{2+i}) starts with 1 - a_3 = 1, while (a_i) starts with a_1 = 0, so (1 - a_{2+i}) is lexicographically greater than (a_i), not smaller. Hence Theorem 6.1(b) fails for beta = 3/2. This direct counterexample uses the paper's own definitions and does not depend on the disputed inequality in Lemma 6.1.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies beta-expansions of real numbers in non-integer bases, develops algorithms for generating such expansions, and states a characterization of the lazy expansion of 1 in Theorem 6.1, which it claims solves the first open problem in the cited Erdős-Komornik paper. The manuscript also contains a chapter on replacing words in beta-expansions and a theorem on unique expansions of 1 (Theorem 5.1). The central claims are supported by proofs of several lemmas, but the main characterization theorem is false, and the supporting arguments contain invalid inferences.","tokens_in":21216,"tokens_out":11183,"duration_ms":83774,"significance":"If Theorem 6.1 were correct, it would resolve a known open problem in beta-expansions and would be a noteworthy contribution. The paper also provides a readable exposition of standard material on greedy and lazy expansions and illustrates non-uniqueness through word replacement. These expository parts are not original research contributions, and the core result fails: Theorem 6.1(b) is directly contradicted by a concrete example, and the proofs of Lemma 6.1 and Lemma 6.2 contain load-bearing errors. The paper's main claim is therefore not established.","major_comments":[{"comment":"This statement is false. For beta = 3/2, which lies in (1, phi], Definition 4.2 gives the lazy expansion of 1 as a = (0,1,0,...): at i=1, the tail sum sum_{j>1} beta^{-j} = 1/(beta(beta-1)) = 4/3 >= 1, so a_1=0; at i=2, the tail sum is 8/9 < 1, so a_2=1; at i=3, a_1/beta + a_2/beta^2 + sum_{j>3} beta^{-j} = 4/9 + 16/27 > 1, so a_3=0. Taking k=2 (with a_2=1), the sequence (1-a_{k+i}) begins with 1-a_3 = 1, whereas (a_i) begins with a_1 = 0. Hence (1-a_{k+i}) is lexicographically greater than (a_i), contradicting the claimed inequality. The forward direction of Theorem 6.1(b) is therefore false.","section":"Theorem 6.1(b)"},{"comment":"The proof of the forward direction asserts that 1 - (B) > 0 because, for beta in ((1+sqrt5)/2, 2), one has 1 - 1/(beta-1) > 0. This inequality is false: beta-1 lies between (sqrt5-1)/2 and 1, so 1/(beta-1) > 1, making 1 - 1/(beta-1) negative. The derived upper bound (B) < 1, and the contradiction x < 1, do not follow. In addition, the argument only treats the case k=1, since it uses 'a_1 = 1'; even a corrected inequality would not establish the claim for every k with a_k=1.","section":"Lemma 6.1, proof"},{"comment":"The converse direction of Theorem 5.1(b) is unsupported. Lemma 5.4(b) shows that the condition (1-a_{k+i}) < (a_i) whenever a_k=1 implies that (a_i) is lazy, not that the expansion is unique. The statement is in fact false: for beta = phi = (1+sqrt5)/2, the lazy expansion of 1 is (0,1,1,1,...); this sequence satisfies (1-a_{k+i}) < (a_i) for every k with a_k=1, yet the greedy expansion (1,1,0,0,...) is a different expansion of 1. Thus uniqueness does not follow from the stated condition.","section":"Theorem 5.1(b)"},{"comment":"Lemma 6.2 does not establish Theorem 6.1(c). The proof considers only k=1, showing that (a_{1+i}) cannot equal or be lexicographically less than (a_i); it does not show that (a_{k+i}) > (a_i) for arbitrary k with a_k=0. The lemma statement also has a notational error, using '(a_{k+1})' where the context requires '(a_{k+i})'. Consequently, Theorem 6.1(c) does not follow from the cited lemmas.","section":"Lemma 6.2 and Theorem 6.1(c)"}],"minor_comments":[{"comment":"The title contains the typo 'charaterisation' and the abstract uses 'nominator' for 'numerator'; these should be corrected.","section":"Title/Abstract"},{"comment":"The introduction states that beta-expansions are for beta in (0,1), but the entire paper works with beta in (1,2); this is a significant notational inconsistency that should be fixed.","section":"Introduction, p. 6"},{"comment":"The proofs refer to 'Lemma 10.1', 'Lemma 10.3', and 'Lemma 10.4', but the numbering in the manuscript is Lemma 5.1, Lemma 5.3, and Lemma 5.4; the cross-references are inconsistent.","section":"Chapter 5"},{"comment":"The lemma statement writes '(1-a_{k+1})' but the proof and Theorem 6.1 require '(1-a_{k+i})'; this notational slip should be corrected.","section":"Lemma 6.1 statement"},{"comment":"For the word 0001, the polynomial is miscomputed: 1/beta^4 = 1/beta + 1/beta^2 + 1/beta^3 gives beta^3 + beta^2 + beta - 1 = 0, not beta^3 - beta^2 - beta - 1 = 0 as written.","section":"Section 2.2, words of length 4"},{"comment":"The solution of beta^2 - beta - 1 = 0 is beta = (1 + sqrt5)/2 or beta = (1 - sqrt5)/2, not 'beta = +/- (1+sqrt5)/2' as written; the negative root is not of the stated form.","section":"Lemma 2.1 proof"}],"recommendation":"reject","confidential_remarks":"The manuscript is an undergraduate dissertation and is candid about its origin, but the mathematical content does not meet research publication standards. The central Theorem 6.1 is false, as shown by a direct counterexample to part (b), and the proof of Lemma 6.1 rests on an elementary false inequality. These are load-bearing errors that cannot be repaired by local revision within the manuscript's current scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the paper's main result is false, and the proof contains a simple arithmetic error that is easy to catch.\n\nFor beta=3/2, which lies in (1, phi], the paper's own Definition 4.2 gives the lazy expansion of 1 as 0,1,0,... . Take k=2; a_k=1. The sequence (1-a_{2+i}) begins with 1 (since a_3=0), while (a_i) begins with 0. So (1-a_{2+i}) > (a_i), directly violating the necessary condition claimed in Theorem 6.1(b). This is not a subtle edge case: it uses the paper's own algorithm and a base in the stated interval.\n\nWhat is good: the first three chapters are a reasonably clear undergraduate account of beta-expansions. The algorithms for generating expansions, the word-substitution examples for the golden ratio, and the uncountability argument for typical sequences are elementary and mostly correct. If the paper were a survey of these basics, it would be acceptable.\n\nThe soft spots are load-bearing. Theorem 5.1(b) misstates the Erdős-Komornik uniqueness criterion: it drops the greedy-side condition (a_{k+i}) < (a_i) for a_k=0, which is needed alongside the lazy-side condition. Lemma 6.1's forward direction depends on claiming 1 - 1/(beta-1) > 0 for beta in (phi,2), but that quantity is negative on the whole interval. Lemma 6.2's proof does not logically establish the strict inequality it claims; the < case does not force all zero coefficients as the proof asserts. None of these are typos; they all point to the same conclusion: the characterization of the lazy expansion of 1 is not established, and the counterexample shows the stated form is false.\n\nI would not cite this paper for the open problem. The expository parts could be useful to a student wanting a gentle introduction to beta-expansions, but the final chapter should be ignored. A serious referee would waste time verifying a claim already disproved by a direct example. Recommend desk rejection, with a note pointing out the counterexample and the misstated Theorem 5.1(b).","headline":"The paper's main characterization is false: for beta=3/2 the lazy expansion of 1 violates Theorem 6.1(b), and the proof relies on an invalid inequality.","tokens_in":21668,"tokens_out":5323,"would_cite":false,"duration_ms":44842,"reading_group":"no","serious_thinker":"no","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11A63"],"pacs":[],"model":"deepseek-v4-flash","headline":"For non-integer bases, laziness of expansions of 1 is characterized by lexicographic tail inequalities, split at the golden ratio.","keywords":["beta-expansions","non-integer bases","lazy expansion","greedy expansion","lexicographic characterization","golden ratio","expansions of 1","unique expansions"],"falsifier":"Take $\\beta=3/2$ (which lies between $\\varphi$ and 2), compute the lazy expansion of 1 by the lazy algorithm, and test whether $(a_{k+i})>(a_i)$ holds at every $k$ with $a_k=0$; a single counterexample would falsify part (c). A faster check targets the proof: the bound $1-1/(\\beta-1)>0$ used in Lemma 6.1 evaluates to $1-2=-1$ at $\\beta=3/2$, so the lemma's contradiction step needs a different argument for this base.","tokens_in":20632,"feed_emoji":"🔢","tokens_out":9815,"duration_ms":79879,"temperature":0.7,"pith_summary":"Expansions of real numbers in non-integer bases $\\beta \\in (1,2)$ are usually not unique, so each number has distinguished greedy and lazy expansions. This paper attacks the lazy expansion of the number 1, the first open problem left in the classic characterization of unique expansions. Its main theorem says that a coefficient sequence $(a_i)$ with $1=\\sum a_i\\beta^{-i}$ is lazy exactly when certain lexicographic inequalities hold between shifted tails $(a_{k+i})$ (or their complements) and $(a_i)$; which inequality is right depends on whether $\\beta$ is at most or greater than the golden ratio $\\varphi=(1+\\sqrt5)/2$. If correct, this gives a complete, testable description of lazy expansions of 1 for every base in $(1,2)$.","feed_headline":"Golden ratio splits the rule for lazy expansions of 1","feed_subtitle":"A lexicographic test decides laziness of beta-expansions of 1, split at the golden ratio.","key_machinery":"The workhorse object is the lexicographic order on binary coefficient sequences, together with the tail-sum characterization of laziness: an expansion $\\sum a_i\\beta^{-i}$ is lazy iff $\\sum_{i\\ge1}(1-a_{k+i})\\beta^{-i}<1$ for every $k$ with $a_k=1$. The proof of the new theorem uses this equivalence to convert laziness into a comparison of tails, then builds an inductive upper bound in Lemma 6.1 that forces the forbidden lexicographic cases to make the sum fall below 1; the golden ratio appears as the base above which that upper bound changes sign behaviour.","core_discovery":"The central discovery is a lexicographic characterization of laziness for the expansion of 1. Write $1=\\sum_{i=1}^\\infty a_i\\beta^{-i}$ with $a_i\\in\\{0,1\\}$. Theorem 6.1 states: (a) if $(1-a_{k+i})<(a_i)$ whenever $a_k=1$, then $(a_i)$ is lazy; (b) if $\\beta\\in(1,\\varphi]$ and $(a_i)$ is lazy, then the same complement condition holds; (c) if $\\beta\\in(\\varphi,2)$ and $(a_i)$ is lazy, then instead $(a_{k+i})>(a_i)$ whenever $a_k=0$. In words, the position of the shifted coefficient tail relative to the original sequence in the lexicographic order completely determines whether the expansion is the smallest one, and the golden ratio marks where the criterion switches from complements to direct shifts.","pith_inferences":["Editorial inference: the same lexicographic pair of conditions may also characterize lazy expansions of arbitrary $x\\in[0,1/(\\beta-1)]$, after replacing the target 1 in the tail inequalities by $x$.","Editorial inference: because laziness is defined as the lexicographically smallest expansion, the theorem suggests a prefix-check algorithm that could certify laziness from finitely many tail comparisons once a quantitative bound on the switching position is known."],"forward_implications":["For every base $\\beta\\in(1,2)$, a candidate sequence for 1 can be certified as lazy or not by comparing shifted tails lexicographically with the original sequence.","The base interval splits at $\\varphi=(1+\\sqrt5)/2$: at or below $\\varphi$ laziness is captured by the complement condition on positions where $a_k=1$; above $\\varphi$ by the direct condition on positions where $a_k=0$.","The theorem resolves the first open problem of the cited 1990 paper on unique expansions, if its inequalities hold.","It also gives a symbolic, order-theoretic test that does not require computing the whole expansion numerically."],"supporting_citations":[{"why":"Supplies the characterization lemmas for greedy and unique expansions, the definition of lazy expansions, and the open problem that Theorem 6.1 addresses.","marker":"[1]"}],"fun_headline_variants":["Golden ratio decides when 1's beta-expansions go lazy","A lexicographic test for lazy expansions of 1","Golden ratio splits lazy-expansion rule for 1","Lazy 1-expansions: golden ratio marks the switch","Golden ratio threshold for lazy beta-expansions of 1"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The forward direction of the characterization depends on the inequality $1 - 1/(\\beta-1) > 0$ for every $\\beta\\in(\\varphi,2)$; if that bound fails for some base in the interval, the contradiction in Lemma 6.1 does not go through.","fun_headline_variants_meta":{"raw":{"variants":["Golden ratio decides when 1's beta-expansions go lazy","A lexicographic test for lazy expansions of 1","Golden ratio splits lazy-expansion rule for 1","Lazy 1-expansions: golden ratio marks the switch","Golden ratio threshold for lazy beta-expansions of 1"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000698,"raw_usage":{"total_tokens":3177,"prompt_tokens":989,"completion_tokens":2188,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":605,"completion_tokens_details":{"reasoning_tokens":2103}},"tokens_in":605,"tokens_out":2188,"duration_ms":14648,"temperature":1.0,"reasoning_tokens":2103,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T11:39:21.375536+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take $\\beta=3/2$ (which lies between $\\varphi$ and 2), compute the lazy expansion of 1 by the lazy algorithm, and test whether $(a_{k+i})>(a_i)$ holds at every $k$ with $a_k=0$; a single counterexample would falsify part (c). A faster check targets the proof: the bound $1-1/(\\beta-1)>0$ used in Lemma 6.1 evaluates to $1-2=-1$ at $\\beta=3/2$, so the lemma's contradiction step needs a different argument for this base.","supporting_citations":[{"cited_title":"Characterization of the unique expansions1 = ∑∞ 𝑖=1𝑞−𝑛𝑖 and related problems","cited_arxiv_id":null,"evidence_quote":"Supplies the characterization lemmas for greedy and unique expansions, the definition of lazy expansions, and the open problem that Theorem 6.1 addresses."}],"review_version":1}