{"id":"02146831-6311-44c9-a206-5f948bf3cf34","arxiv_id":"2506.18103","paper_version":2,"verdict":"REJECT","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"high","formal_verification":"none","parameter_count":2,"one_line_summary":"A paper on self-referential 'hiccup' sequences claims asymptotic densities and explicit Beatty formulas, but the main theorem and several formulas are contradicted by small parameter choices.","lead":"Scientists often study sequences where the next number depends on whether the step number has already appeared; this paper examines a three-parameter family of such sequences. It claims a general density formula, exact Beatty-sequence formulas for two subfamilies, and ties to lattice and tree patterns, but several of these claims fail on simple examples.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 2.1 is false: S(1,2,1) has a(n)=n and density 1, not r0=(1+√5)/2; Step 2 drops the index restriction j<k in the 'Key property', making H_n and N(n)+O(1) differ by n.","rationale":"The paper's central claim is Theorem 2.1: for every S(x,y,z) with y>z>0, a(n)/n converges to the positive root of r^2−zr−(y−z)=0. This result anchors the abstract and the whole of Section 2, so it is the load-bearing statement. The proof's Step 2 is the hinge: it equates H_n, the number of indices k≤n that are prior values, with the counting function N(n)+O(1) via the 'Key property' that k∈A_{k−1} iff k is in the image. The equivalence holds only with the index restriction ∃j<k, which is the very definition of A_{k−1}; the proof then switches to unrestricted image membership, and the two counts differ by the values k whose first occurrence sits at index j≥k. The sequence S(1,2,1) realizes this gap at full strength: a(k)=k, so every positive integer k is an image value at index j=k, yet no index is ever a prior value. Hence H_n=0, N(n)=n, and the density is 1, while Theorem 2.1 predicts (1+√5)/2. I verified the counterexample term by term and by induction; it satisfies Definition 1.1 with y>z>0, and no stated hypothesis excludes x=1 or z=1. The failure is not isolated: S(1,y,1)=(1,2,3,...) for all y≥2, and Proposition 3.1's Beatty formula for S(1,3,2) gives a(2)=4 instead of 3, exposing the Section 3.1 claim that the offset is determined by ⌊r_A−γ⌋=x, which only uses a(1). The paper's own remark restricts the morphic eigenvalue proof to y>1 and z>1, so the z=1 range of Theorem 2.1 is unsupported there as well. I also flag Section 5.3, where Theorem 5.3 is presented with its algebraic proof 'omitted for brevity' and only a geometric outline; this self-admitted gap is secondary to the headline failure but reinforces the high-risk assessment. The reader's weakest-assumption diagnosis matches mine, so I see no reason to change the REJECT verdict. Some secondary material (the y=0 periodic analysis of Section 4 and the x=Z+1 proofs of Theorems 3.2 and 3.4) appears sound and could be salvaged, but the central asymptotics claim does not stand.","tokens_in":10833,"tokens_out":20695,"duration_ms":168177,"concrete_test":"Simulate S(1,2,1) for n=1 to 20 using Definition 1.1: a(1)=1, and for k≥2 add y=2 if k∈{a(1),...,a(k−1)}, else add z=1. Induction shows the set of prior values is always {1,...,k−1}, so no index is ever a hit; record a(k)=k, H_n=0, and N(n)=n. The Step 2 identity H_n=N(n)+O(1) fails by n, and a(n)/n=1 for all n, contradicting the predicted r0=(1+√5)/2. The same run extended to S(1,3,2) checks Proposition 3.1 at n=2, where the published formula overshoots the actual value a(2)=3.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's headline result, Theorem 2.1, is false as stated. For S(1,2,1) (x=1, y=2, z=1, so y>z>0), a(1)=1 and by induction {a(1),...,a(k−1)}={1,...,k−1} for every k≥2, so k is never a prior value: every step is a miss, a(k)=a(k−1)+1=k, and a(n)=n. The density is 1, not the root of r^2−zr−(y−z)=0, which for this triple is (1+√5)/2. Indeed S(1,y,1)=(1,2,3,...) for every y≥2, so the theorem fails on a whole subfamily. The proof breaks in Step 2 of Section 2: the 'Key property' is stated as k∈A_{k−1} iff k is in the image, 'i.e., ∃j<k with a(j)=k'. With the index restriction this is a tautology; the next line silently drops the restriction and writes H_n=|{2≤k≤n: k∈im(a)}|=N(n)+O(1). For S(1,2,1), H_n=0 while N(n)=n, so the identity fails by n, and the derived system r=z+(y−z)/r is disconnected from the actual sequence. This is exactly the gap the paper's own remark acknowledges by restricting the morphic eigenvalue proof to y>1, z>1: the z=1 range of Theorem 2.1 is both unsupported and false. The same flaw invalidates Proposition 3.1's offset formulas, which are said to be determined by a(1): for S(1,3,2), the announced Beatty form gives a(2)=4, while the definition gives a(2)=3.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies the three-parameter family of self-referential sequences S(x,y,z) defined by a(1)=x and, for k>1, a(k)=a(k-1)+y if k belongs to {a(1),...,a(k-1)} (a hit) and a(k)=a(k-1)+z otherwise (a miss). The paper claims that for y>z>0 the ratio a(n)/n converges to the positive root of r^2 - z r - (y-z)=0 (Theorem 2.1); that the subfamilies S(x,Z+1,Z) and S(x,Z,Z+1) are non-homogeneous Beatty sequences for every starting value x (Propositions 3.1 and 3.3); that the case y=0, z>=2 leads to ultimately periodic increments and a linear recurrence (Theorem 4.1); that several discriminant-zero cases have explicit closed forms with triangular, square, and hexagonal lattice interpretations (Section 5); and that there are formal connections to meta-Fibonacci recurrences (Section 6). The paper also contains an OEIS compendium and explicitly positions itself as complementary to the morphic-sequence framework of Fokkink and Joshi.","tokens_in":11335,"tokens_out":11256,"duration_ms":105430,"significance":"If the central results were correct, the paper would provide a useful analytic complement to the morphic characterization of hiccup sequences and a unified explanation for many OEIS entries. The author is to be credited for recognizing the connection to Fokkink and Joshi and for assembling a systematic OEIS catalogue. However, the main asymptotic theorem and the general Beatty propositions are false as stated, and the closed-form claims in Section 5 rest on unproved assertions. The special case x=Z+1 in Section 3 is plausible and may contain a kernel of a correct theorem, but the manuscript as it stands does not deliver reliable new results.","major_comments":[{"comment":"Theorem 2.1 is false for z=1. For S(1,2,1), an immediate induction gives a(n)=n, so a(n)/n tends to 1, whereas the claimed limit is the positive root of r^2 - r - 1 = 0, namely (1+sqrt(5))/2. In fact S(1,y,1) is the identity sequence for every y>=2. The error is in Step 2 of the proof: the 'Key property' states that k in A_{k-1} iff k is in the image of the sequence, i.e., iff there exists j<k with a(j)=k. That equivalence is false when every preimage of k occurs at an index at least k. In the counterexample, each k is in the image because a(k)=k, but it is never a hit, so H_n=0 while N(n)=n; the asserted identity H_n=N(n)+O(1) is off by n. The proof therefore does not establish the stated range y>z>0. The paper's own remark restricting the morphic eigenvalue argument to y>1 and z>1 is consistent with this failure, and the theorem would at minimum need a condition such as z>=2 or an additional argument controlling the first preimage of k.","section":"Section 2, Theorem 2.1"},{"comment":"Proposition 3.1 gives false formulas for the stated starting values. For Z=2 and x=1, the proposition says a(n)=floor(r_A n - 2(r_A-1)/(r_A+1)) with r_A=1+sqrt(2). The definition of S(1,3,2) gives a(1)=1 and a(2)=3, since 2 is not in {1}. The proposed formula gives floor(r_A - 2(r_A-1)/(r_A+1))=1 but floor(2r_A - 2(r_A-1)/(r_A+1))=4, not 3. The same type of failure occurs in Proposition 3.3: for Z=2 and x=1, the formula reduces to a(n)=floor(r_B n) with r_B=(3+sqrt(5))/2, so it gives a(2)=5, whereas S(1,2,3) has a(2)=4. The 'Theoretical Justification' after Proposition 3.1 explains the offset as being determined by the initial condition a(1)=x, but infinitely many offsets satisfy floor(r_A-gamma)=x, and the paper does not verify that the chosen offset produces a sequence whose image and increments match the self-referential rule at every index.","section":"Section 3.1, Proposition 3.1"},{"comment":"The general Beatty propositions are not proved for all x, and the constructive proofs do not cover the stated generality. The paper invokes Fokkink and Joshi's Theorem 15 to assert existence of a Beatty form for x<=Z, but the explicit formulas in Propositions 3.1 and 3.3 are claimed for all x, including x>Z, where no existence theorem is cited. The special-case proofs of Theorems 3.2 and 3.4 handle only x=Z+1 and, as written, assert that a 'hit' is equivalent to the larger increment for a Beatty sequence without proving that the specific image of the sequence has the required complementarity with respect to the index set. Establishing that equivalence is the core of the argument and cannot be replaced by a reference to the Beatty-Rayleigh theorem without checking the complementary pair.","section":"Sections 3.1 and 3.2, general Beatty claims"},{"comment":"The Section 5 closed-form results are not backed by complete proofs. Theorem 5.3 explicitly says 'A full algebraic proof showing that this closed form satisfies the S(5,1,2) recurrence is technical and omitted for brevity,' and the claimed correspondence between misses and hexagonal layers is justified only by a figure and an intuitive description. In the proof of Theorem 5.2, the key equivalence is asserted with 'This equivalence can be established by showing...' but no demonstration follows. In Theorem 5.1, the set of integers avoided by f(n) is invoked as 'a known property' with no reference. Because these equivalences are load-bearing for the advertised lattice connections, the theorems are unsupported as stated.","section":"Section 5, Theorems 5.1-5.3"}],"minor_comments":[{"comment":"The notation for the limsup and liminf of a(n)/n is visually ambiguous: both are rendered with the same letter r, and the reader must infer which is which from context. A distinct notation such as r and r, or r^* and r_*, would improve readability.","section":"Section 2 and Lemma 2.2"},{"comment":"The propositions state cases for x=0, but Definition 1.1 takes x in N, and it is not stated whether N includes 0; if it does, the case x=0 should specify how a(1)=0 is handled in the self-referential rule for k=2.","section":"Section 3, Propositions 3.1 and 3.3"},{"comment":"The definition of the layer parameter m in Theorem 5.3 uses a ceiling expression with a repeated radical, but no derivation or estimate is given to show that the displayed closed form is well-defined for all n.","section":"Section 5.3"}],"recommendation":"reject","confidential_remarks":"The paper's central theorem is false for an entire parameter region, and the Beatty propositions fail on simple examples, so a major revision would require substantially restricting or correcting the main claims. I would not recommend further review until the authors either rework Theorem 2.1 with the correct hypotheses, correct or remove the general-x Beatty formulas, and replace the proof outline of Theorem 5.3 with a complete proof. The paper also relies heavily on OEIS entries and on undocumented 'known properties'; the author should supply direct proofs or precise references for each such assertion."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The reader's stress-test is right and it's the main thing you need to know: Theorem 2.1 is false as stated. For S(1,2,1), y>z>0, yet a(n)=n for every n, so the density is 1, not (1+√5)/2. The proof's 'Key property' in Step 2 drops the index restriction j<k; with that restriction it is a tautology, and without it the identity H_n = N(n)+O(1) fails by n. The same quantifier shift invalidates the Beatty offsets in Propositions 3.1 and 3.3, which are tuned to a(1) but fail at small n (e.g., S(1,3,2) gives a(2)=4 instead of 3). This is not a manufactured flaw; it is a load-bearing error in the central claim.\n\nWhat the paper does well: Section 4 on y=0 is elementary and correct, and its linear-recurrence conclusion is genuinely nice. The closed forms for S(3,1,2) and S(4,1,2) are correct and match known OEIS entries, and the paper is honest about citing Fokkink and Joshi, whose morphic framework covers the general asymptotics. The OEIS compendium is useful as a catalogue.\n\nThe soft spots, in proportion: the y=0 result is narrow and the periodic-increment proof is fine but not deep. The lattice connections for S(5,1,2) are asserted with the proof omitted, which is a serious gap for a paper that leans on those connections. The main asymptotic and Beatty claims are either false or already consequences of the cited work. The paper's own remark concedes that the morphic eigenvalue proof only works for y>1, z>1, which should have warned the author that the z=1 range needed separate treatment.\n\nWho is this for? A reader collecting OEIS identifications or exploring hiccup sequences will find the compendium and the y=0 section worth a look, but anyone relying on Theorem 2.1 will be misled. The paper needs a major correction before it is publishable, and the headline result should be repaired or clearly restricted.\n\nMy recommendation: desk reject. The central theorem is false, the Beatty formulas fail on simple examples, and the correct residue is mostly covered by the cited literature. The author could salvage a short note on the y=0 case and the corrected lattice formulas, but not in the present form.","headline":"The headline density theorem is false on a whole subfamily (S(1,2,1) is the identity), so the paper cannot stand as-is; it does contain a few correct elementary pieces.","tokens_in":11817,"tokens_out":1998,"would_cite":false,"duration_ms":21622,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11B83","68R15","05A15"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper establishes that every sequence in the three-parameter hit-or-miss family S(x,y,z) with y>z>0 grows linearly with slope r0, the positive root of r^2 - z r - (y-z)=0, and that two subfamilies are exactly Beatty sequences with…","keywords":["self-referential sequences","hit-or-miss sequences","hiccup sequences","Beatty sequences","morphic sequences","meta-Fibonacci recurrences","lattice combinatorics","asymptotic density"],"falsifier":"Compute $S(1,2,1)$: the rule gives $a(1)=1$ and, since no index is ever a hit, $a(n)=n$ for all $n$, so $a(n)/n=1$ while Theorem 2.1 predicts $(1+\\sqrt{5})/2\\approx 1.618$; this one sequence settles whether the theorem needs the extra condition that values appear before they are tested.","tokens_in":10606,"feed_emoji":"📈","tokens_out":12860,"duration_ms":113249,"temperature":0.7,"pith_summary":"The paper studies a three-parameter family of self-referential integer sequences defined by a hit-or-miss rule: start with a(1)=x, and at each later index add y if that index has already appeared as a value, otherwise add z. Its central result is that whenever y>z>0, the sequence grows linearly with slope r0, the positive root of $r^2 - z r - (y-z)=0$, so the asymptotic density is completely determined by the two increments. Two subfamilies, $S(x,Z+1,Z)$ and $S(x,Z,Z+1)$, are shown to be non-homogeneous Beatty sequences with explicit closed forms $\\lfloor r n - \\gamma \\rfloor$ for every starting value, including the quasi-homogeneous case $a(n)=\\lceil n r \\rceil$ when $x=Z+1$. Additional results cover the $y=0$ case via periodic increments and linear recurrences, discriminant-zero cases with triangular, square, and hexagonal lattice interpretations, and a bridge to meta-Fibonacci recurrences through leaf counts in forests of complete $k$-ary trees.","feed_headline":"One quadratic law governs these self-referential sequences","feed_subtitle":"Two subfamilies are exactly Beatty sequences; other cases give linear recurrences and lattice counts.","key_machinery":"The central machinery is the hit/miss counting relation combined with the counting function $N(t)=\\max\\{k:a(k)\\le t\\}$. Lemma 2.2 states a general duality: if a strictly increasing sequence grows at rate $r$ in the sense of limsup/liminf, then its counting function grows at the reciprocal rates, and this turns the exact identity $a(n)=x+z(n-1)+(y-z)H_n$ into the equilibrium equation $r=z+(y-z)/r$, whose positive solution is $r_0$. An alternative derivation notes that the same root appears as the Perron-Frobenius eigenvalue of the generating morphism's adjacency matrix. For the Beatty results, the load-bearing tool is Rayleigh-Beatty complementarity between $\\lceil n r \\rceil$ and $\\lfloor n s \\rfloor$ with $1/r+1/s=1$: the size of the increment, the larger value $y$ or the smaller value $z$, is exactly determined by whether the index lies in the sequence's image.","core_discovery":"The central discovery is that a simple self-referential rule, \"add y if the index has already appeared as a value, otherwise add z,\" generates sequences whose asymptotic and exact structure are governed by one quadratic equation. The paper proves that for $y>z>0$ the limit $\\lim_{n\\to\\infty} a(n)/n$ exists and equals $r_0$, the positive root of $r^2-z r-(y-z)=0$, by combining the exact relation $a(n)=x+z(n-1)+(y-z)H_n$, where $H_n$ counts hits, with a duality lemma relating limsup and liminf of the sequence to those of its counting function. It then proves that the families $S(x,Z+1,Z)$ and $S(x,Z,Z+1)$ are Beatty sequences with slopes $r_A=(Z+\\sqrt{Z^2+4})/2$ and $r_B=(Z+1+\\sqrt{Z^2+2Z-3})/2$, and gives explicit offsets for every starting value; for $x=Z+1$ the closed form is simply $a(n)=\\lceil n r \\rceil$. The paper also establishes that $y=0$, $z\\ge 2$ yields eventually periodic increments with period $z$ and the linear recurrence $a(k)-a(k-1)-a(k-z)+a(k-z-1)=0$, and that discriminant-zero cases correspond to lattice counting problems on triangular, square, and hexagonal grids. Finally, it shows that $S(k+1,1,k+1)=n+k\\,a_{0,k}(n)$, where $a_{0,k}(n)$ counts leaves in a forest of complete $k$-ary trees with $n$ nodes, tying the hit/miss rule to meta-Fibonacci recurrences.","pith_inferences":["The density theorem as stated needs a small extra hypothesis: every value must be attained before it is tested as an index. Without it, $S(1,2,1)$ gives $a(n)=n$ and density $1$, not the predicted golden ratio, so the theorem is best read as a statement about non-permutation cases.","The explicit offsets in the Beatty formulas suggest a uniform Rayleigh-Beatty proof for all starting values $x$ in the two families, which would make the closed forms follow from a single complementarity argument rather than case-by-case verification.","The same counting-function duality used here could be applied to related self-referential families, such as the four-parameter variant where membership is tested against a shifting window, to obtain analogues of the density root."],"forward_implications":["For every $y>z>0$, the sequence is asymptotically linear with slope $r_0$, so the counting function satisfies $N(n)\\sim n/r_0$ and the density of hits converges to $(r_0-z)/(y-z)$.","The two Beatty families give immediate closed-form computations $\\lfloor r n-\\gamma\\rfloor$ for all starting values, so any catalogued sequence in these families can be generated without simulating the recurrence.","When $y=0$ and $z\\ge 2$, the increments are eventually periodic with period $z$, giving slope $z-1$ and the linear recurrence $a(k)-a(k-1)-a(k-z)+a(k-z-1)=0$.","The discriminant-zero sequences $S(3,1,2)$, $S(4,1,2)$, and $S(5,1,2)$ provide explicit formulas for Ramsey core numbers, a square-spiral covering count, and a hexagonal lattice structure, respectively.","The identity $S(k+1,1,k+1)=n+k\\,a_{0,k}(n)$ links the hit/miss rule to the number of leaves in forests of complete $k$-ary trees, so questions about these recurrences can be translated into tree-enumeration problems."],"supporting_citations":[{"why":"Introduced the S(x,y,z) family and announced the program this paper carries out; supplies the definitional origin.","marker":"[5]"},{"why":"Independently named and studied these as hiccup sequences, proved they are all morphic, and gave the Sturmian correspondence used here to justify the Beatty forms.","marker":"[7]"},{"why":"Earlier result on a remarkable integer sequence that the morphic framework of [7] builds on; supports the theoretical context.","marker":"[4]"},{"why":"Proves the k-ary tree leaf-counting identity that underlies the meta-Fibonacci bridge in Section 6.","marker":"[10]"},{"why":"Classical source of Beatty sequences and the Rayleigh complementarity used to connect increments to hits and misses.","marker":"[3]"},{"why":"Reference for Ramsey core numbers rc(2,n), identified with S(3,1,2) in the triangular-lattice theorem.","marker":"[6]"},{"why":"Source for the spiral-covering viewpoint used in the square-lattice thumbtack theorem.","marker":"[8]"}],"fun_headline_variants":["Self-referential sequences obey a single quadratic law","One quadratic equation controls these self-referential sequences","Beatty sequences and periodic recurrences from one self-referential rule","Quadratic root governs self-referential sequences and their subfamilies","A single self-referential rule: quadratic limit, Beatty exact forms"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes that every number produced by the sequence has been produced earlier when it is checked as an index, so being in the image is the same as being a hit; if a value first appears exactly when it is tested, the central density statement can fail.","fun_headline_variants_meta":{"raw":{"variants":["Self-referential sequences obey a single quadratic law","One quadratic equation controls these self-referential sequences","Beatty sequences and periodic recurrences from one self-referential rule","Quadratic root governs self-referential sequences and their subfamilies","A single self-referential rule: quadratic limit, Beatty exact forms"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000586,"raw_usage":{"total_tokens":2871,"prompt_tokens":1183,"completion_tokens":1688,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":799,"completion_tokens_details":{"reasoning_tokens":1604}},"tokens_in":799,"tokens_out":1688,"duration_ms":12650,"temperature":1.0,"reasoning_tokens":1604,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T18:58:22.529985+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute $S(1,2,1)$: the rule gives $a(1)=1$ and, since no index is ever a hit, $a(n)=n$ for all $n$, so $a(n)/n=1$ while Theorem 2.1 predicts $(1+\\sqrt{5})/2\\approx 1.618$; this one sequence settles whether the theorem needs the extra condition that values appear before they are tested.","supporting_citations":[{"cited_title":"Numerical analogues of Aronson’s se- quence,","cited_arxiv_id":null,"evidence_quote":"Introduced the S(x,y,z) family and announced the program this paper carries out; supplies the definitional origin."},{"cited_title":"A remarkable sequence of integers,","cited_arxiv_id":null,"evidence_quote":"Earlier result on a remarkable integer sequence that the morphic framework of [7] builds on; supports the theoretical context."},{"cited_title":"The combinatorics of certain k-ary meta-Fibonacci sequences,","cited_arxiv_id":null,"evidence_quote":"Proves the k-ary tree leaf-counting identity that underlies the meta-Fibonacci bridge in Section 6."},{"cited_title":"Beatty, Problem 3173,Amer","cited_arxiv_id":null,"evidence_quote":"Classical source of Beatty sequences and the Rayleigh complementarity used to connect increments to hits and misses."},{"cited_title":"Ramsey numbers and pseudo-random graphs,","cited_arxiv_id":null,"evidence_quote":"Reference for Ramsey core numbers rc(2,n), identified with S(3,1,2) in the triangular-lattice theorem."},{"cited_title":"Spiral covering designs and lower bounds,","cited_arxiv_id":null,"evidence_quote":"Source for the spiral-covering viewpoint used in the square-lattice thumbtack theorem."}],"review_version":2}