{"id":"e65388e2-6053-441f-9bbb-d6e0681ce496","arxiv_id":"2502.15743","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":3.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The exponents of primes in the factorization of n form fractal sequences, with the exponent of 2 drawing the Levy Dragon and the odd part of n drawing the Heighway Dragon.","lead":"This paper builds a sieve that generates prime factorizations by writing down the p-adic valuation of each integer, and proves these valuation sequences are fractal. It then identifies the 2-adic valuation sequence with the turn sequence of the Levy Dragon curve, and the odd part of n with the Heighway Dragon.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Section 4.5's 'negligible deviations' claim is the load-bearing unproven step: Theorem 15 only matches the turn subsequence at multiples of 8, not the full v2 curve, so the Levy Dragon conclusion lacks a metric.","rationale":"The reader's weakest assumption matches my own: the unproved geometric convergence in Section 4.5. I checked the block structure and believe the claim is likely true under diameter normalization, so a conditional verdict is appropriate rather than rejection. The discrete results are independent and correct. Novelty is modest because A346070 and A099545 already encode the same turn sequences, but that is not a correctness objection. The paper is honest about the informality of the geometric step. Hence the verdict should remain CONDITIONAL, and the reader's judgment needs no change.","tokens_in":20468,"tokens_out":17562,"duration_ms":165673,"concrete_test":"Compute, for m=1..20, the v2 polygonal path C_{8*2^m} with unit segments and 90-degree CCW turns, and the skeleton L_m with unit segments and turn v2(8k) at each multiple of 8. Translate both to start at the origin and normalize each by its diameter. If the Hausdorff distance d_H(C_{8*2^m}/D_C, L_m/D_L) does not tend to 0 as m increases, the Levy-Dragon claim is false as stated. If it does tend to 0, the empirical concern is resolved and only a formal proof of convergence in this metric is missing.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 14 and the discrete parts of the paper are sound: the sieve generates v_p(n), and the decimation rule at indices f(p+1) plus unboundedness proves fractalness. The geometric conclusion in Section 4 is the weak point. Theorem 15 proves that the finite Levy-dragon turn sequence at iteration m equals v2(8i) for i=1..2^m-1; it does not by itself describe the curve generated by all terms of v2. Between two multiples of 8, indices 8k+1..8k+7 carry the fixed pattern 0,1,0,2,0,1,0. This block has zero net turn and bounded displacement: relative to the direction before the block it advances 4 units while making a bounded rectangular excursion of width 2. Section 4.5 asserts without proof that these bounded deviations become negligible as iterations increase. That is plausible only if one normalizes by the overall diameter and uses a set-convergence metric; the paper specifies neither. Without a formal argument, the central claim that v2 'produces the Levy Dragon' rests on an unverified limit interchange: the turn subsequence matches, but the extra loops might persist in the chosen metric or alter the limit set. The Heighway-Dragon appendix is less vulnerable, because there the turn sequence mod 4 is exactly the standard paperfolding sequence.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a table-based sieve that successively generates, for each prime p, the sequence of p-adic valuations v_p(n), and proves (Theorem 6) that the sieve outputs exactly the primes and the prime factorizations. It then defines a fractal sequence as one that is predictably self-containing and aperiodic, proves (Theorem 14) that every v_p sequence is fractal via the decimation rule selecting indices f(p+1) and via unboundedness of the terms, and claims (Section 4) that v_2, read as the number of 90-degree counterclockwise turns at each integer, generates the Lévy Dragon. The main supporting result (Theorem 15) is that the terms of v_2 at indices divisible by 8 follow exactly the algorithm for the Lévy dragon's turn sequence. An appendix relates the odd part of n modulo 4 to the Heighway Dragon turn sequence.","tokens_in":20633,"tokens_out":5789,"duration_ms":62992,"significance":"If fully established, the paper would provide a clean exposition of a known but underappreciated fact: p-adic valuation sequences are self-similar in a precise subsequence sense, and v_2's turn sequence is related to the Lévy Dragon. The discrete portions are largely correct and self-contained, and the paper is honest in not fitting parameters and in comparing against OEIS benchmarks. However, the geometric equivalence in Section 4.5 is asserted rather than proved, and the paper's own Section 5 acknowledges that the other 'fractal figures' are not formalized. The contribution is therefore a useful set of exact identities and a suggestive geometric conjecture, rather than a complete proof of the headline geometric claim.","major_comments":[{"comment":"The sentence 'the size of these deviations, relative to the overall figure, becomes negligible as the number of iterations increases' is the load-bearing step connecting Theorem 15 to the claim that the curve generated by v_2 is the Lévy Dragon, but no metric or convergence theorem is supplied. Theorem 15 proves that the finite turn sequence generated by the Lévy-dragon algorithm at iteration m equals v_2(8i) for i=1..2^m-1; it does not describe the curve generated by all terms of v_2, and the intervening block v_2(8k+1..8k+7)=0,1,0,2,0,1,0 has zero net turn but a bounded nonzero excursion. To establish the claimed limit, the paper needs a precise notion, such as Hausdorff metric on normalized curves or a path-length accounting showing that the extra T-excursions vanish as m tends to infinity, together with a proof that they do so. Without this, the paper proves a turn-subsequence identity, not geometric equivalence of the full v_2 curve.","section":"Section 4.5"},{"comment":"The geometric interpretation of v_2 is introduced informally: terms are said to represent 'the number of 90-degree turns to make at n,' but the formal curve construction is never defined, and the equivalence 'four 90-degree turns are the same as none' is used without proof. Since v_2 contains arbitrarily large terms, the curve is only defined up to congruence modulo 4, and that reduction should be stated explicitly as part of the definition of the map from sequences to curves. This clarification is needed before the equivalence with A346070 or with the Lévy Dragon can be made rigorous.","section":"Section 4.1"}],"minor_comments":[{"comment":"The sentence 'we want p + 1 at that index' should read 'we want j + 1 at that index'; the surrounding argument is otherwise clear.","section":"Section 2.4, Proposition 4"},{"comment":"There is a duplicated article in 'the the number returned by vp(ni)'; this should be corrected.","section":"Definition 7"},{"comment":"The phrase 'can be confident that v2⟨⟩ produces the Lévy Dragon' appears to be missing a 'we' and should read 'can we be confident that v2⟨⟩ produces the Lévy Dragon.'","section":"Section 4.5"},{"comment":"The proof's 'leaves no gaps' step is persuasive but informal; stating the result as an explicit induction on the binary length of the index, or equivalently as l_i = v_2(8i) for every positive integer i, would make the argument easier to verify.","section":"Theorem 15"}],"recommendation":"major_revision","confidential_remarks":"The paper is in math.GM, and its discrete results are sound. The main risk is the informal Section 4.5 convergence claim, which is load-bearing for the title's central assertion that prime factors form fractals in the geometric sense. If the authors supply a precise metric and proof for that step, or clearly reframe the paper as proving only the turn-subsequence identity and the fractalness of the valuation sequences, the manuscript could be acceptable. The Section 5 material is appropriately labeled as exploratory, but the abstract and title make the geometric claim central, so the gap cannot be ignored."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis is a likeable paper, but its headline claim is not fully proved. The discrete core is sound: the sieve does generate the p-adic valuations, and the self-containment at indices p+1 plus unboundedness gives aperiodicity under the paper's own definition of 'fractal sequence.' Theorem 15 is the most concrete new observation: the Lévy turn-insertion algorithm exactly produces v2 at multiples of 8. That is a clean, checkable identity.\n\nThe weak point is Section 4.5. The paper only matches the turn sequence at the 'spikes' (multiples of 8). Between those, v2 contains the fixed block 0,1,0,2,0,1,0, which makes a bounded rectangular excursion. The paper asserts that this excursion becomes negligible relative to the whole curve as iterations grow, but it never gives a metric or a convergence argument. Without that, Theorem 15 gives you a matching subsequence of turns, not a proof that the full v2 curve approaches the Lévy Dragon. The stress-test note is right about this.\n\nThe Heighway Dragon appendix is on firmer ground. There, the connection is an exact mod-4 identity between the odd-part sequence and the paperfolding turn sequence, and the insertion algorithms are shown to agree. That part is convincing.\n\nNovelty is modest: the paper credits Cloitre for the sieve construction and cites OEIS entries that already label these sequences fractal and already link A346070 and A099545 to dragon curves. What is new is the unified presentation and the explicit theorem about multiples of 8. No code or machine-checked proofs are shipped, but the discrete claims are easy to verify by hand.\n\nWho is this for? Someone teaching elementary number theory or looking for a visually compelling way to present p-adic valuations. It is not a paper that resolves an open problem, and the geometric section needs more rigor before the central metaphor becomes a theorem. Still, the paper is honest about its own limits and the discrete mathematics is correct.\n\nI would send it to a referee: the discrete results deserve a record, and a referee could push the author to either tighten Section 4.5 or soften the claim. I would not cite it as a proof of the Lévy-Dragon connection, but I would cite Theorem 15 if I needed that specific identity.","headline":"A clean, honest synthesis of known p-adic valuation patterns with dragon curves; the discrete math is solid, but the geometric conclusion outruns the proof.","tokens_in":21287,"tokens_out":3996,"would_cite":false,"duration_ms":39437,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11A51","28A80","11B83"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper claims that the exponents of a fixed prime in the integers form fractal sequences, and that the 2-adic valuation sequence draws the Lévy Dragon.","keywords":["p-adic valuation","fractal sequence","prime factorization sieve","Lévy Dragon","Heighway Dragon","self-similar sequences","odd part of n","division-free sieve"],"falsifier":"Compute the turtle path for the 2-adic valuation sequence at 90-degree turns over the first 8^N indices and measure its Hausdorff distance to the Lévy Dragon polyline after N iterations; if that distance does not tend to 0 as N grows, the paper's convergence claim fails even though the turn sequence matches.","tokens_in":20130,"feed_emoji":"🐉","tokens_out":7077,"duration_ms":69632,"temperature":0.7,"pith_summary":"The paper claims that the apparent chaos of prime factorizations hides a repeatable fractal structure. It builds a division-free sieve—an algorithm that repeatedly copies, concatenates, and increments a seed sequence of zeros—and proves that the row for each prime p records exactly the exponent of p in each integer n. It then proves that every such p-adic valuation sequence is a fractal sequence in a defined sense: predictably self-containing and aperiodic. Finally, it claims that reading the 2-adic valuation sequence as a number of 90-degree turns draws the Lévy Dragon, and that the odd part of n sequence draws the Heighway Dragon. A sympathetic reader would take away that prime factorization carries geometric fractal structure.","feed_headline":"Prime-factor exponents form fractals; 2's draw the Lévy Dragon","feed_subtitle":"A division-free sieve shows p-adic valuations are self-similar and encodes dragon curves in factorization.","key_machinery":"The engine is the duplicate-concatenate-increment rule. For each prime p, begin with the one-term sequence ⟨0⟩; at each stage make p−1 copies of the current sequence, concatenate them after it, and add 1 to the final term. The nth entry of the resulting sequence is the exponent of p in n, so the sieve's rows are exactly the p-adic valuations. Fractality is certified by a decimation rule that selects indices that are multiples of p+1; because p never divides p+1, the term at f(p+1) equals the term at f, giving a predictable copy of the sequence inside itself. For the Lévy Dragon, the matching object is the turn algorithm that starts with ⟨3⟩, increments every entry, inserts a new 3 between adjacent entries, and appends 3s at both ends; Theorem 15 shows this sequence is identical to the 2-adic valuation sequence at indices that are multiples of 8.","core_discovery":"The paper's central discovery is that the exponents of any fixed prime p across the positive integers—the p-adic valuation sequence—form a fractal sequence, and that for p=2 the same sequence, drawn as a path that turns v2(n) quarter-turns at step n, produces the Lévy Dragon. The sieve that reaches this conclusion starts from a single 0 and repeatedly expands it by making p−1 copies, appending them, and increasing the final term by 1; Theorem 6 shows this reproduces prime factorizations without division. Lemma 12 exhibits a simple decimation rule—take every (p+1)-st term—that locates a copy of the whole sequence inside itself, and Lemma 13 shows no periodic block generates it. Theorem 15 proves that the terms at indices divisible by 8 exactly match an algorithm for the Lévy Dragon's turn sequence, with the intervening terms forming small 'T' detours. In the appendix, the same construction is carried for the odd part of n, whose values modulo 4 match the Heighway Dragon's turns.","pith_inferences":["I infer that if the spike-shrinking claim is formalized in a Hausdorff-metric or length-ratio sense, the v2 curve converges to the Lévy Dragon exactly; the paper states the tendency but leaves the metric unspecified.","The duplicate-concatenate-increment rule is morphic in flavor, so a natural extension is a string-replacement proof that p-adic valuation sequences are automatic or morphic words, connecting them to the paperfolding literature.","The matching at multiples of 8 may yield an explicit formula: the Lévy turn sequence is the 2-adic valuation of 8n, which could simplify dragon-curve turn computations."],"forward_implications":["A student can generate prime factorizations by mechanical copying rather than division, making the exponent pattern visible.","Every prime's exponent sequence is predictably self-containing and aperiodic, so each prime contributes its own fractal layer to the natural numbers.","The 2-adic valuation sequence encodes the Lévy Dragon's turns at every index divisible by 8; the intervening 'T' shapes are the only difference from the standard construction.","The odd part of n, taken modulo 4, reproduces the Heighway Dragon's turn sequence, tying the other classical dragon curve to integer factorization.","The paper's open questions point to a family of angle-and-prime combinations that generate additional fractal figures, and to a possible L-system translation of the sieve."],"supporting_citations":[{"why":"supplies the sequence-generation method that the sieve generalizes into a division-free factorization tool.","marker":"[2]"},{"why":"provides the formal definitions of self-containing sequences and selection functions used to certify fractality.","marker":"[8]"},{"why":"is the source for the Lévy Dragon construction and the idea of curves made of parts similar to the whole.","marker":"[9]"},{"why":"gives the iterative triangle construction whose turn rule Theorem 15 matches against the 2-adic valuation at multiples of 8.","marker":"[13]"},{"why":"analyzes how vertices and edge orientations persist across iterations, from which the Lévy turn algorithm is derived.","marker":"[1]"},{"why":"documents the paper-folding origin of the Heighway Dragon and the turn sequence the appendix compares to the odd part of n.","marker":"[16]"},{"why":"observes that the odd part of n sequence is fractal, which the appendix formalizes and connects to the Heighway Dragon.","marker":"[11]"}],"fun_headline_variants":["Prime factor exponents are fractal: 2's yield Lévy Dragon","Fractals from prime exponents: the Lévy Dragon emerges","P-adic valuations self-similar: Lévy Dragon encoded","How prime exponents draw the Lévy Dragon"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The geometric conclusion rests on the unproven claim that the 'T'-shaped detours in the v2 curve become negligible relative to the whole figure as more indices are added.","fun_headline_variants_meta":{"raw":{"variants":["Prime factor exponents are fractal: 2's yield Lévy Dragon","Fractals from prime exponents: the Lévy Dragon emerges","P-adic valuations self-similar: Lévy Dragon encoded","How prime exponents draw the Lévy Dragon"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000437,"raw_usage":{"total_tokens":2155,"prompt_tokens":812,"completion_tokens":1343,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":428,"completion_tokens_details":{"reasoning_tokens":1275}},"tokens_in":428,"tokens_out":1343,"duration_ms":11406,"temperature":1.0,"reasoning_tokens":1275,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T19:27:55.785128+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the turtle path for the 2-adic valuation sequence at 90-degree turns over the first 8^N indices and measure its Hausdorff distance to the Lévy Dragon polyline after N iterations; if that distance does not tend to 0 as N grows, the paper's convergence claim fails even though the turn sequence matches.","supporting_citations":[{"cited_title":"Cloitre, [comment on A007814 ], https://oeis.org/A007814, 2003-03-06","cited_arxiv_id":null,"evidence_quote":"supplies the sequence-generation method that the sieve generalizes into a division-free factorization tool."},{"cited_title":"Kimberling, Self-containing sequences, fractal sequences, selection functions, and parasequences, J","cited_arxiv_id":null,"evidence_quote":"provides the formal definitions of self-containing sequences and selection functions used to certify fractality."},{"cited_title":"L´ evy, Plane or space curves and surfaces consisting of parts similar to the whole","cited_arxiv_id":null,"evidence_quote":"is the source for the Lévy Dragon construction and the idea of curves made of parts similar to the whole."},{"cited_title":"Riddle, L´ evy dragon, https://larryriddle.agnesscott.org/ifs/levy/levy","cited_arxiv_id":null,"evidence_quote":"gives the iterative triangle construction whose turn rule Theorem 15 matches against the 2-adic valuation at multiples of 8."},{"cited_title":"Alster, The finite number of interior component shapes of the levy dragon, Discrete Comput","cited_arxiv_id":null,"evidence_quote":"analyzes how vertices and edge orientations persist across iterations, from which the Lévy turn algorithm is derived."},{"cited_title":"Knuth [with C","cited_arxiv_id":null,"evidence_quote":"documents the paper-folding origin of the Heighway Dragon and the turn sequence the appendix compares to the odd part of n."},{"cited_title":"Mitchell, [comment on a000265], https://oeis.org/A000265, 2005-12-07","cited_arxiv_id":null,"evidence_quote":"observes that the odd part of n sequence is fractal, which the appendix formalizes and connects to the Heighway Dragon."}],"review_version":1}