{"id":"73bf1170-9075-4bfd-bd58-ff72d9806340","arxiv_id":"1908.08470","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Sub-images of the Hofstadter butterfly obey exact Farey-sum rules, with nesting scaling factors and chain endpoints determined by four integers (p, q, M, N).","lead":"This paper derives exact algebraic rules for the smaller copies of the Hofstadter butterfly nested inside the main plot, including the precise scaling factors as the copies shrink. A generalist might read it because it converts a famous fractal's observed self-similarity into a few simple formulas.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central claim depends on the unproven premise that bounding gaps remain open from φ_L to φ_R; the paper itself flags this in Section 5 as empirical, so equations (16) and (54) are not established for all bands.","rationale":"The reader identified the open-gap extrapolation as the weakest assumption, and my reading of the paper confirms this is the single load-bearing premise. I checked the algebraic derivations in Sections 2–4 (renormalisation of M and N, the 2x2 unimodular matrix recursion, the eigenvalue calculation, and the scaling ratio) and found no internal inconsistency. The formulas follow exactly once one assumes an open sub-image exists. The concern is not that the mathematics is wrong but that its domain of validity is asserted rather than proved: the paper explicitly acknowledges in Section 5 that the results of [7] do not guarantee the extrapolation to φ′=1, and that gap closure is the possible failure mode. The paper then classifies sub-images empirically, with no theorem or exhaustive numerical evidence. Because every headline result (right-edge location, Farey-sum centre, exact scaling factors, semi-infinite chains) is downstream of this extrapolation, the universal claim is only as strong as the empirical open/closed classification. The proposed concrete test would settle the issue for representative cases, including the examples in the figures and Table 1. If any tested gap closes strictly inside the interval, the universal form of the claim would need to be weakened or rejected; if all tested gaps remain open, the conditional accept would be supported. Since the reader already recommended a conditional verdict, my stress-test does not change that verdict, but it sharpens the condition: the paper should be accepted conditional on the open-gap premise being verified by systematic numerics or subsequent proof.","tokens_in":13665,"tokens_out":5575,"duration_ms":52325,"concrete_test":"Compute the Harper spectrum along the blue sub-image interval: for φ = 1/(3−φ′) with φ′ = j/10 (j=0,...,9) and also φ′ = 1−10^{-k} for k=1,...,6, diagonalize the q×q Harper matrix (or use a transfer-matrix gap-edge calculation) and record the widths of the gaps immediately above and below the band that forms the left edge at φ=1/3. If either gap width vanishes for any φ′<1, the open-sub-image premise fails for this example. Repeat the same check for the green sub-image (φ=1/(4−φ′)) and for each chain entry in Table 1. This directly tests whether equation (16) can be crossed from φ_L to φ_R.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central construction takes equation (15) from the exact renormalisation group of [7] and extrapolates it to φ′=1 to define the right edge of every sub-image via equation (16). This extrapolation is only valid if the band forming the left edge remains bounded by the same two gaps for all φ′ in [0,1]. The paper explicitly concedes this is not guaranteed: Section 5 states that 'the results in [7] do not guarantee that extrapolation to φ′ = 1 is possible', and that 'one way in which the procedure could fail is if the gaps ... close up as |φ′| increases.' The subsequent classification into open and closed sub-images is stated as an empirical finding ('We find...', 'we find empirically...'), not a theorem. Since the right-edge formula (16), the centre formula (18), the Farey-neighbour relations (19), the nesting scaling factors (33)-(36), and the chain formula (54) all depend on the existence of an open sub-image with well-defined Hall integers M,N throughout the interval, a single counterexample with a gap closing inside (0,1) would invalidate the universal form of the claim. The paper's own examples are consistent, but they do not establish the premise for all bands.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper develops a renormalisation-group description of the sub-images of the Hofstadter butterfly. The authors show that the edges of a sub-image are related by simple rational transformations (equations 15–18), that the edge and centre fluxes form Farey neighbours, that recursive nesting of sub-images can be represented by products of 2×2 unimodular matrices whose eigenvalues give exact irrational scaling factors (equations 33–36), and that chains of sub-images follow a linear formula (equation 54) ending in a gap-closure edge on one side and an accumulation point on the other. The main results are derived from the exact renormalisation group of [7] together with a newly derived recursion for the Hall-conductance integers (equation 12), and are illustrated with several numerical examples.","tokens_in":13908,"tokens_out":12569,"duration_ms":113750,"significance":"If the underlying premise holds, this is an elegant and parameter-free explanation of the exact self-similar structure of the Hofstadter butterfly: the scaling factors are derived from the renormalisation group rather than fitted, and the Farey-sum centre rule (equation 18) explains earlier empirical observations. The recursive rule for Hall integers (equation 12) is a useful new ingredient. The paper is also careful to present worked examples that reproduce the stated flux values. The main weakness is that the universality of the results rests on an empirically asserted property of gap persistence, which is not proven and is explicitly acknowledged in Section 5 as a potential failure mode.","major_comments":[{"comment":"The derivation of the right-edge formula (16), and consequently the centre formula (18), the nesting scaling factors (33)–(36), and the chain formula (54), all require that the two gaps bounding a band remain open as the renormalised flux is extrapolated from φ′=0 to φ′=1. As the authors state, 'the results in [7] do not guarantee that extrapolation to φ′ = 1 is possible', and gap closure in the interior of (0,1) would invalidate these equations for that band. The paper treats the persistence of the gaps as an empirical finding, but this property is load-bearing for the universal form of the claims. Please either provide a proof that the bounding gaps stay open for the bands used in the constructions, or state the results explicitly as conditional on an open-sub-image assumption and adjust the Summary accordingly.","section":"Section 5 (after equation 54)"},{"comment":"The classification of sub-images as 'open' or 'closed', and the assertion 'we find empirically that this occurs when |φ′| = 1', are not accompanied by any numerical evidence or a description of how the classification was performed. Since a single counterexample with a gap closing inside (0,1) would invalidate the formulas for that band, the paper should include a systematic numerical test (for example, over all bands up to some maximal denominator q) or a theoretical argument establishing that gap closure can only occur at the endpoint φ′=±1. Without such evidence, the claim that every rational band is an edge of at least one open sub-image remains unverified.","section":"Section 5"}],"minor_comments":[{"comment":"In Table 1, the rows for C_{1/2→0,1+} and C_{0→1/2,1+,2−} appear to be indexed inconsistently with the caption: with j=0 giving the closed edge, the first row gives p_0/q_0=1 rather than 1/2, and the second row gives p_0/q_0=1 rather than 0. Please check the indexing or the labels.","section":"Table 1"},{"comment":"In the first sentence of the paragraph after equation (45), the phrase 'where we the parameters of the initial sub-images' is missing a verb; it should read 'where we use the parameters' or similar.","section":"Section 4.5"},{"comment":"The description 'we follow a nested sequence of sub-images of the red sub-image of figure 1, with φ_L = 1/3' is ambiguous, because φ_L=1/3 is the internal left edge of the nested sub-image, not the global left edge of the red sub-image (which is 3/8). Please clarify to avoid confusion.","section":"Section 4.4, first paragraph"},{"comment":"The derivation of the Hall-integer recursion (12) uses multiplicative filling fractions, which implicitly requires that the cluster of bands considered is isolated by gaps. It would be helpful to note this condition explicitly, since it ties directly to the open-sub-image assumption discussed in Section 5.","section":"Section 2.2"}],"recommendation":"major_revision","confidential_remarks":"The paper relies heavily on the authors' earlier exact-renormalisation-group paper [7]; this is not itself a problem, but the referee should be aware that the central risk is the unproven gap-persistence assumption. The scope and style are appropriate for J. Phys. A. The main revision should focus on making the open-sub-image condition explicit and providing a systematic numerical or analytic justification for the classification."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper takes empirical observations about self-similar sub-images in the Hofstadter butterfly and turns them into exact algebraic rules, cleanly. The genuinely new pieces are the recursion for the Hall integers (equation 12), the derivation of the Farey-sum center formula from the exact renormalization group, and the exact scaling factors for nested sub-images (equations 33–36). I checked the worked examples, and the arithmetic reproduces the stated flux values. No free parameters, no fitting. That is real progress.\n\nThe paper does well to show that the nesting and chain structure follows from simple 2x2 unimodular matrices and rational arithmetic. The scaling factor depending only on q_L + M is a surprising and nice simplification. The authors also honestly connect to their earlier empirical work in [5] and [10], so the novelty is framed fairly.\n\nThe soft spot is the one the authors themselves flag in Section 5: the right-edge formula (16) relies on extrapolating the renormalized flux to phi' = 1, which the RG results from [7] do not guarantee. If a bounding gap closes before phi' reaches 1, the formula fails for that band. The paper states this and then classifies open vs closed sub-images empirically. So the universal claim that every band has an open sub-image in one direction is not a theorem; it is an observed pattern. The stress-test note is accurate on this point. I would have liked a proof, or at least a sharper conjecture about when gaps close, but I do not think the absence is disqualifying. The examples are consistent, and the empirical classification is clearly labeled. The central argument holds up conditional on that gap-open premise.\n\nOne small thing: the paper would be easier to verify with more detail on how the open/closed classification was done numerically, but the formulas are explicit enough to re-implement.\n\nThis paper is for people working on the Harper model, quantum Hall integers, or fractal structure of spectra. It deserves a serious referee. I would send it out, asking the referee to focus on the gap-closure premise and on whether equation (54) for chains is fully supported beyond the examples. Accept with revisions after that check; not a desk reject.","headline":"Clean derivation of exact self-similarity rules for Hofstadter sub-images, conditional on an honest empirical open-gap premise.","tokens_in":14420,"tokens_out":2151,"would_cite":true,"duration_ms":21136,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Every open sub-image of the Hofstadter butterfly is fixed by a four-integer rule, and its nested descendants shrink by an exact quadratic-irrational factor.","keywords":["Hofstadter butterfly","Harper equation","renormalization group","Farey tree","self-similar spectrum","Hall conductance integers","gap labelling","sub-image scaling"],"falsifier":"Compute the Harper spectrum along a dense set of flux values between a candidate left edge and the predicted right edge, for example between $\\phi = 1/3$ and $\\phi = 2/5$ for the centre band with $M = 1$, $N = -2$, and look for an intermediate flux at which either bounding gap closes; if such a flux exists, the open-sub-image criterion and the formulas built on it fail for that band, and if it never occurs, the premise survives.","tokens_in":13429,"feed_emoji":"🦋","tokens_out":7793,"duration_ms":74648,"temperature":0.7,"pith_summary":"This paper claims that the self-similar sub-structure of the Hofstadter butterfly, the fractal plot of the spectrum of a Bloch electron in a periodic potential and uniform magnetic field, is governed exactly by rational arithmetic and a renormalisation-group flow, with no fitting parameters. Given any spectral band at a rational flux $\\phi_L = p_L/q_L$, the flux at the opposite edge of the sub-image it generates is $\\phi_R = (p_L + M)/(q_L - N)$, where $M$ and $N$ are integer labels of the band, and the centre of the sub-image is the Farey sum of the two edge fractions. The paper shows that recursive nesting of sub-images is equivalent to multiplying $2\\times2$ unimodular matrices, so that the horizontal sizes of successive sub-images shrink by an exact factor $1/\\lambda_*^2$, where $\\lambda_*$ is an eigenvalue of a matrix determined by four integers. It also shows that adjacent sub-images can be joined into semi-infinite chains whose one end is a gap-closing point and whose other end is a rational accumulation point. If correct, these results reduce a visually complex fractal to a small set of number-theoretic rules and explain previously empirical Farey-tree observations.","feed_headline":"Exact ratios govern Hofstadter butterfly's nests","feed_subtitle":"A four-integer rule fixes each sub-image's edges and center; no numerical fitting is needed.","key_machinery":"The central object is the exact renormalisation-group mapping for the Harper equation, written as $\\phi' = (q_0\\phi - p_0)/(N_0\\phi + M_0)$, with integers satisfying $1 = q_0M_0 + p_0N_0$; it is a Möbius transformation that converts the spectrum near a rational flux into a sub-spectrum at a shifted flux. The paper packages the iteration of this map as multiplication of unimodular $2\\times2$ matrices, $B^L_{j+1} = C_L B^L_j$, and the eigenvalues of $C_L$ yield the scaling factor of equation (33). For chains, the opposite-edge formula iterates by simple addition, giving $\\phi_j = (p_0 + jM_0)/(q_0 - jN_0)$. The machinery works by turning geometric self-similarity into an algebraic recursion on four integers, so all quantitative predictions, edge positions, centres, scaling factors, and accumulation points, follow exactly from the recursion rather than from numerical fitting.","core_discovery":"The central discovery is that every open sub-image of the Hofstadter butterfly is labelled by four integers $(p_L, q_L, M, N)$: a rational flux edge $p_L/q_L$ and the Hall-conductance integers of the edge band. From those four integers, the other edge follows from $\\phi_R = (p_L + M)/(q_L - N)$ (equation (16)), and the centre follows from the Farey sum $\\phi_c = (p_L + p_R)/(q_L + q_R)$ (equation (18)); the three rationals are Farey neighbours, with unit-determinant relations. Nested sub-images are generated by repeated application of the renormalisation transformation, which acts on the label quadruple as multiplication by a unimodular $2\\times2$ matrix. The eigenvalues of that matrix give an exact asymptotic scaling factor $\\lambda_*$, so that successive nested butterflies shrink horizontally by $1/\\lambda_*^2$, while the numerators and denominators of the edge fluxes grow like $\\lambda_*$. Adjacent sub-images sharing a vertical edge form chains whose successive flux positions follow the additive sequence $\\phi_j = (p_0 + jM_0)/(q_0 - jN_0)$; the chains are infinite in one direction and terminate at an accumulation point, while the other end is where a spectral gap closes. The same formalism yields exact expressions for accumulation points, which are quadratic irrationals.","pith_inferences":["A consequence the paper does not draw: because $\\lambda_*$ depends only on $\\tilde q_L + \\tilde M$, the same nesting ratios should occur in any model that shares Harper's gap-labelling integers, so the scaling law is likely universal within the family of Harper-like Hamiltonians.","The paper leaves open a proof of when gaps close; a rigorous classification of open versus closed edges would convert the chain construction from an empirical into a fully proven statement.","The same unimodular-matrix recursion could be used to compute the asymptotic growth of band counts in nested sequences, connecting the butterfly's fractal dimension to the distribution of integer sums $\\tilde q_L + \\tilde M$; the paper does not compute this.","One testable extension is to check whether the exact edge formula (16) holds for the symmetry-related counterpart of the butterfly in the perturbed-Bloch-band picture, where $\\phi$ is inverted; the paper fixes one picture but predicts the same arithmetic."],"forward_implications":["For any open sub-image, knowing the four integers $(p_L, M, q_L, N)$ determines the opposite edge and the centre exactly, so predictions can be made without solving for the whole spectrum.","Successive generations of nested sub-images have horizontal widths shrinking by the exact factor $1/\\lambda_*^2$, where $\\lambda_*$ depends only on $\\tilde q_L + \\tilde M$; no numerical fit is needed.","The recursive construction converges to irrational accumulation points, and in the self-similar case those points are quadratic irrationals given by the fixed-point formula.","Chains of sub-images are semi-infinite: they terminate at one end because a spectral gap closes, and extend to an accumulation point at $\\phi = -M_0/N_0$ at the other end.","The Farey-neighbour relations between the edge and centre fractions follow from the rational-arithmetic rules, explaining the number-theoretic patterns seen empirically."],"supporting_citations":[{"why":"Defines the butterfly spectrum and first notes the repeated sub-images that are the paper's subject.","marker":"[1]"},{"why":"Supplies the exact renormalisation-group map for the flux that all edge, nesting, and scaling formulae are built on.","marker":"[7]"},{"why":"Documents the empirically observed Farey-tree structure of sub-images that the exact rules re-derive.","marker":"[5]"},{"why":"Provides empirical evidence for self-similar nesting that the scaling-factor calculation explains.","marker":"[10]"},{"why":"Shows each band carries a quantised Hall conductance integer M, the label used to specify sub-images.","marker":"[18]"},{"why":"Supplies the Hall-conductance integral used to derive how M and N renormalise under iteration.","marker":"[22]"},{"why":"Gives the gap-labelling form N phi + M for filling fractions, needed for the integer recursions.","marker":"[23]"}],"fun_headline_variants":["Four integers dictate Hofstadter's fractal nests","Exact scaling uncovered in butterfly's sub-images","Farey sums pin down butterfly nests and chains","Unimodular matrices govern fractal spectra","Four integers label every butterfly nest"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The construction works only if the two spectral gaps bounding the chosen band remain open all the way from the left edge to the right edge of the sub-image; the paper verifies this by numerical inspection rather than proving it.","fun_headline_variants_meta":{"raw":{"variants":["Four integers dictate Hofstadter's fractal nests","Exact scaling uncovered in butterfly's sub-images","Farey sums pin down butterfly nests and chains","Unimodular matrices govern fractal spectra","Four integers label every butterfly nest"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001019,"raw_usage":{"total_tokens":4287,"prompt_tokens":918,"completion_tokens":3369,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":534,"completion_tokens_details":{"reasoning_tokens":3301}},"tokens_in":534,"tokens_out":3369,"duration_ms":22946,"temperature":1.0,"reasoning_tokens":3301,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:38:28.102026+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the Harper spectrum along a dense set of flux values between a candidate left edge and the predicted right edge, for example between $\\phi = 1/3$ and $\\phi = 2/5$ for the centre band with $M = 1$, $N = -2$, and look for an intermediate flux at which either bounding gap closes; if such a flux exists, the open-sub-image criterion and the formulas built on it fail for that band, and if it never occurs, the premise survives.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the butterfly spectrum and first notes the repeated sub-images that are the paper's subject."},{"cited_title":"Phys., A, 20, 4337-4354","cited_arxiv_id":null,"evidence_quote":"Supplies the exact renormalisation-group map for the flux that all edge, nesting, and scaling formulae are built on."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Documents the empirically observed Farey-tree structure of sub-images that the exact rules re-derive."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides empirical evidence for self-similar nesting that the scaling-factor calculation explains."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows each band carries a quantised Hall conductance integer M, the label used to specify sub-images."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Hall-conductance integral used to derive how M and N renormalise under iteration."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the gap-labelling form N phi + M for filling fractions, needed for the integer recursions."}],"review_version":1}