{"id":"ed8dc329-e57d-4860-92e5-d9aa0d3f7816","arxiv_id":"2607.15676","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"For Ising spin glasses with ±1 couplings on even-branching Migdal–Kadanoff hierarchical lattices, a single exact renormalization step induces enough bond dilution to rigorously rule out spin-glass order and stiffness at all temperatures, even as the fractal dimension approaches 3 from below.","lead":"This paper proves that certain hierarchical lattice models of spin glasses—used as approximations to real materials—have no glassy ordering at any temperature. The proof works by showing that one renormalization step randomly cuts enough bonds to kill long-range order, even for lattices with fractal dimension close to 3.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'arbitrarily close to 3 from below' construction is asserted without proof and, as printed, the b formula is inverted: it fails condition (7) and gives d > 3. The headline claim needs the corrected formula plus an asymptotic derivation.","rationale":"The reader's weakest assumption identifies the same spot; I agree. I independently checked the proof of Theorem 1: the percolation recursion, the uniqueness/existence of the nontrivial fixed point via R_b(x)=s, the monotone decrease below the fixed point, and the exponential bound (24)-(27) are correct. The self-dual enumeration in Theorem 4 is also a finite calculation and not suspect. Thus the only load-bearing weakness is the asymptotic large-s construction. The formula as printed appears to put b in the denominator of the log factor; at minimum it is seriously ambiguous, and the missing derivation means the 'arbitrarily close to 3 from below' result is unproven. This is fixable, so the appropriate action is a conditional acceptance requiring correction and proof, which is already the reader's verdict.","tokens_in":11140,"tokens_out":14195,"duration_ms":124820,"concrete_test":"One check: derive the asymptotic of (7) with p = sqrt(2/(pi*s)) and b = sqrt(pi*s/2)*(log s - log log s + C); show the RHS is ~ s^{-exp(-C)}, so (7) holds iff C > log 2. Then evaluate (7) numerically for s = 10^3, 10^4, 10^6 with both the printed (divided) and corrected (multiplied) formulas; the divided formula should fail and give d > 3, while the multiplied formula should hold with d < 3.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central advertised result — MK lattices with fractal dimension arbitrarily close to 3 from below and no spin-glass order — rests on the asymptotic assertion after Eq. (9): for large even s, choose b ≈ sqrt(pi*s/2)*(log s - log log s + C) with any C > log 2, then condition (7) holds. This assertion is stated without proof; End Matter proves Theorem 1 only for fixed (s,b) satisfying (7), not the existence of such (s,b) for large s. Moreover, as printed, 'b ≈ sqrt(s*pi/2) log s - log log s + C' is ambiguous and reads as b ≈ sqrt(s*pi/2)/(log s - log log s + C). With that reading, b*p -> 0, so the right side of (7) is about (1/log s)^s, which is far below p, and (7) fails; also log b ≈ (1/2) log s, giving d -> 3 from above. The correct scaling must be b ≈ sqrt(s*pi/2)*(log s - log log s + C); then b*p ≈ log s - log log s + C and (7) reduces asymptotically to s^{-1/2} < s^{-exp(-C)}, i.e., C > log 2. Without this corrected formula and a derivation, the headline 'near three dimensions from below' claim is unsupported. Theorem 1 itself remains valid for any even (s,b) satisfying (7).","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript considers the Ising spin glass with symmetric binary couplings on Migdal–Kadanoff (MK) hierarchical lattices with even branching number. Its central observation is that one exact RG step produces a zero effective bond with probability p_{s,b} = binom(s,s/2)/2^s. Discarding the nonzero coupling values reduces the endpoint spin-glass correlation to a bond-percolation connectivity probability on the same hierarchical lattice. If p_{s,b} satisfies condition (7), p_{s,b} < [1-(1-p_{s,b})^b]^s, then the percolation recursion pushes the vacancy probability above threshold, giving an exponential decay of the endpoint correlation and hence, under the paper's definitions, no spin-glass order and no stiffness at any temperature. Theorems 1–3 formalize this for finite and zero temperature. The paper also applies the criterion to the square-lattice MK approximation (s=b) and to a self-dual hierarchical lattice with d=log 5/log 2, using two RG steps to exceed the percolation threshold. In addition, it claims that for large even s one can choose b so that the fractal dimension approaches 3 from below while condition (7) still holds, contrasting with earlier numerical estimates of a lower critical dimension near 2.52.","tokens_in":11470,"tokens_out":14950,"duration_ms":156377,"significance":"The core mechanism is elegant and self-contained: an exact RG step produces an atom at zero interaction, and the spin-glass correlation is bounded by a percolation probability. For any fixed pair (s,b) satisfying (7), the proof of Theorem 1 is valid: the percolation recursion, the fixed-point analysis, and the exponential bound in the End Matter are correct, and no fitted or numerical input is used. This already yields rigorous statements about absence of endpoint order and stiffness for a nontrivial family of hierarchical lattices, including the s=b case. If the asymptotic construction with d approaching 3 from below were properly proved, it would be a significant result, since it would show that the lower-critical behavior of MK spin glasses depends on the detailed lattice construction and not only on fractal dimension. The self-dual lattice result is also potentially interesting, but its proof currently rests on unshown enumerations. Overall, the paper's core conditional theorem is sound, but the advertised 'near three dimensions' claim is not yet established in the present text.","major_comments":[{"comment":"The construction of MK lattices with fractal dimension arbitrarily close to three from below is asserted, not proved. The displayed formula for b is ambiguous as written: it can be read as b ≈ sqrt(sπ/2)/(log s − log log s + C), in which case condition (7) fails and the dimension approaches 3 from above. The intended expression is presumably b ≈ sqrt(sπ/2) (log s − log log s + C). Even with that correction, no lemma is provided showing that for all sufficiently large even s there exists an integer b satisfying (7) with this scaling; the End Matter proof of Theorem 1 starts from a pair (s,b) that already satisfies (7). Because the 'near three dimensions from below' statement is a central advertised result, this gap must be fixed before publication.","section":"Main text, paragraph containing Eq. (9)"},{"comment":"The values q2 = 261/512 and q2 = 2255/4096 are asserted to follow by direct enumeration, but the enumeration is not displayed. Since a single RG step gives q1 = 1/2, exactly at the percolation threshold, the proof of Theorem 4 depends entirely on these unshown counts. Please provide a reproducible enumeration (table of multiplicities, generating function, or a short verification script), or restate the theorem as conditional on the enumeration.","section":"End Matter, proof of Theorem 4"},{"comment":"The statement that vanishing of the endpoint spin-glass correlation in Eq. (2) implies vanishing of the conventional spin-glass order parameter in Eq. (3) is not demonstrated. On a hierarchical lattice there is no translation invariance, so correlation of the two root spins is not automatically representative of the spatial average over all pairs. Please either prove the required bound on the sum over pairs using the scale-by-scale connectivity estimate, or explicitly restrict the no-order claim to endpoint correlations.","section":"Main text, paragraph after Eq. (2)"}],"minor_comments":[{"comment":"The text contains a duplicated word: 'This work was was supported'.","section":"Acknowledgments"},{"comment":"The recursion is written as an inequality y_n ≥ [1-(1-y_{n-1})^b]^s. Since the effective bonds after a single exact RG step are generated from disjoint sets of original couplings and are therefore independent, equality appears to hold; if the inequality is intentional, a one-sentence explanation would be helpful.","section":"End Matter, Eq. (14)"},{"comment":"The wording 'the no-order criterion becomes easier to satisfy at lower fractal dimensions' is a statement about fixed s and increasing b; it may be misread as a general characterization. Consider rephrasing to make the fixed-s dependence explicit.","section":"Main text, after Eq. (7)"}],"recommendation":"major_revision","confidential_remarks":"The core Theorem 1 appears correct and is a genuine contribution. The problem is that the abstract and introduction give equal billing to the 'near three dimensions from below' construction, which as written is an unproved asymptotic assertion with an ambiguous formula. This is fixable within the manuscript's scope by stating the intended formula precisely and adding a rigorous asymptotic lemma with integer-rounding control. The self-dual lattice theorem likewise needs its enumeration made checkable. I would not reject the paper, but I would not accept it until these load-bearing points are resolved."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Core idea is new and largely sound. A single exact RG step on even-branching MK lattices creates zero effective bonds with positive probability, and bounding E[<σAσB>^2] by the percolation connectivity is a simple and valid reduction. I checked Theorem 1: the recursion, fixed point analysis, and exponential decay all work. The square-lattice MK case and the T=0/stiffness extensions are clean. This is a real contribution.\n\nThe soft spot is the \"near three dimensions\" construction. The asymptotic assertion after Eq. (9) is stated without proof: for large even s, b ~ sqrt(pi*s/2)*(log s - log log s + C) with C > log 2 satisfies condition (7). The claim is plausible—I did the quick calculation and it works—but it is not in the paper. The End Matter only proves Theorem 1 for fixed (s,b). On top of that, the displayed formula is typeset ambiguously; one natural reading gives b ~ sqrt(pi*s/2)/(log s - log log s + C), which violates condition (7) and gives d > 3. I assume the intended product formula is what feeds Eq. (9), but as written it needs fixing. There is also a small Stirling typo: p_s,b should be sqrt(2/(pi*s)), not sqrt(2)/(pi*s). These are fixable, but they are exactly where the headline claim lives.\n\nThe self-dual lattice section is also slightly compressed. The finite enumeration giving q2 = 261/512 (and 2255/4096 at T=0) is asserted, not shown; a referee can verify it, but the authors should provide a table or code. Minor.\n\nI do not see a load-bearing flaw in the main mechanism. If the asymptotic claim collapsed, the Letter would lose the most striking result but Theorem 1 and the square-lattice proof would stand. The references and citation pattern look appropriate; the self-citation [49] is contextual and the proof is independent.\n\nWho should read it: people working on hierarchical spin glasses, rigorous RG, and percolation-based bounds. I would send it to a competent referee. With corrected display, a proof or precise reference for the asymptotic claim, and the enumeration data, it is a solid, citable paper.","headline":"A genuinely new, mostly rigorous dilution-to-percolation argument for excluding spin-glass order on MK lattices, with a fixable soft spot in the asymptotic 'near 3D' construction.","tokens_in":11956,"tokens_out":4726,"would_cite":true,"duration_ms":41489,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["82B44","82B20","82B28"],"pacs":["75.10.Nr","64.60.ah","05.50.+q"],"model":"deepseek-v4-flash","headline":"One exact renormalization step suffices to kill spin-glass order in a family of hierarchical lattices whose fractal dimension approaches three from below.","keywords":["spin glass","Migdal-Kadanoff lattice","hierarchical lattice","bond percolation","renormalization group","spin-glass order","lower critical dimension","stiffness"],"falsifier":"Check the displayed formula for b: if it is read as b = sqrt(s pi/2) / (log s - log log s + C), the resulting fractal dimension approaches 3 from above for large s, falsifying the 'from below' claim. More directly, compute p_{s,b} = binom(s, s/2)/2^s and the right-hand side of condition (7) for the proposed b with C > log 2 for large even s (e.g., s = 10^4); failure of the inequality at any s collapses the construction. Running the hierarchical percolation recursion numerically for these parameters and observing a nonzero endpoint connectivity would also contradict the theorem's conclusion.","tokens_in":10909,"feed_emoji":"🧲","tokens_out":6401,"duration_ms":45460,"temperature":0.7,"pith_summary":"The paper proves that Ising spin glasses with symmetric binary couplings on Migdal–Kadanoff (MK) hierarchical lattices have no spin-glass order and no stiffness at any temperature, provided a simple inequality holds for the branching number s and scale factor b. The mechanism is that a single exact renormalization-group step creates zero-strength bonds with positive probability; these act as holes, and when the induced hole density is above the percolation threshold, the endpoints decouple exponentially fast with distance. The same argument rigorously rules out a spin-glass phase in the MK approximation to the square lattice, and, by choosing s and b large, it constructs lattices whose fractal dimension is arbitrarily close to 3 from below yet still satisfy the no-order criterion. This sharply contrasts with numerical estimates of a lower critical dimension near 2.52 for small s and b.","feed_headline":"Spin-glass order dies on lattices that approach three dimensions","feed_subtitle":"One exact renormalization step dilutes bonds past the percolation threshold, killing order and stiffness at all temperatures.","key_machinery":"The key machinery is the one-step exact renormalization dilution: with symmetric binary couplings and even s, each branch contribution has equal magnitude and random sign, so the renormalized interaction is exactly zero whenever the signs balance, an event of probability p_{s,b}. This converts the spin-glass problem into a hierarchical bond-percolation recursion y_n >= [1 - (1 - y_{n-1})^b]^s, whose function f(x) = 1 - (1 - x^b)^s has a unique nontrivial fixed point. Condition (7) places the initial vacancy probability above that fixed point, forcing exponential decay of endpoint connectivity; the fixed-point analysis of f supplies the exponential bound.","core_discovery":"The central discovery is that, for an even branching number s, the exact renormalization of a single generating unit of the MK lattice yields an effective interaction that vanishes with probability p_{s,b} = binom(s, s/2)/2^s, independent of temperature and scale factor. Treating these zero bonds as vacancies reduces the problem to bond percolation on the hierarchical lattice. If p_{s,b} satisfies p_{s,b} < [1 - (1-p_{s,b})^b]^s, then the endpoint connectivity probability decays as C exp(-c r_n), so the spin-glass correlation tends to zero in the thermodynamic limit. The paper proves this for finite temperature, zero temperature, and for the boundary-condition stiffness, and extends the same","pith_inferences":["If the asymptotic construction holds, it suggests that the lower critical dimension is not a universal attribute of hierarchical lattices but depends on the detailed generating rule; numerical lower-critical estimates from small s and b should not be extrapolated to large s and b.","The percolation-reduction method could be pushed further: applying two or more exact RG steps before comparing to percolation may prove absence of order on lattices where the one-step criterion fails, as already demonstrated for the self-dual lattice; a systematic multi-step version might settle cases such as (s,b) = (6,4) and (8,5).","Because the criterion requires an atom at zero effective coupling, extending it to continuous distributions would need a different mechanism; a possible route is to approximate continuous couplings by discrete ones with a controlled error, though the paper does not pursue this.","A direct numerical check of condition (7) for the proposed large-s choice of b would either validate the near-three-dimensional construction or reveal that the asymptotic assertion needs correction; the displayed formula's direction of approach to 3 is currently ambiguous."],"forward_implications":["The MK approximation to the Ising spin glass on the square lattice has no spin-glass phase or stiffness, rigorously for symmetric binary couplings.","The no-order criterion applies at all temperatures including T = 0, and the absence of stiffness follows with probability approaching one exponentially fast, effectively corresponding to a stiffness exponent of -infinity rather than power-law scaling.","For large s and a suitable choice of b, there exist MK lattices with fractal dimension arbitrarily close to 3 from below that still have no spin-glass order, supporting the scaling picture that d = 3 is marginal in the large-s, large-b regime.","The same dilution mechanism proves the absence of order and stiffness on the self-dual hierarchical lattice with d = log 5 / log 2 ~ 2.32, after two RG steps push the vacancy probability past the percolation threshold.","The sufficient condition extends to asymmetric binary distributions and to any discrete coupling distribution with a positive atom at zero effective coupling after one RG step, and in the asymmetric case it also rules out ferromagnetic order."],"fun_headline_variants":["Exact RG step eliminates spin-glass order near 3D","Bond dilution kills spin-glass order on MK lattices","One exact RG step kills spin-glass order near 3D","Spin-glass order absent on lattices approaching 3D","Percolation argument rules out spin-glass near 3D"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The claimed construction of lattices with fractal dimension arbitrarily close to three from below rests on an asymptotic choice of the scale factor b that is asserted without proof and is displayed in a form that does not obviously satisfy the required inequality.","fun_headline_variants_meta":{"raw":{"variants":["Exact RG step eliminates spin-glass order near 3D","Bond dilution kills spin-glass order on MK lattices","One exact RG step kills spin-glass order near 3D","Spin-glass order absent on lattices approaching 3D","Percolation argument rules out spin-glass near 3D"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000492,"raw_usage":{"total_tokens":2247,"prompt_tokens":732,"completion_tokens":1515,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":476,"completion_tokens_details":{"reasoning_tokens":1441}},"tokens_in":476,"tokens_out":1515,"duration_ms":10875,"temperature":1.0,"reasoning_tokens":1441,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T22:37:18.502143+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Check the displayed formula for b: if it is read as b = sqrt(s pi/2) / (log s - log log s + C), the resulting fractal dimension approaches 3 from above for large s, falsifying the 'from below' claim. More directly, compute p_{s,b} = binom(s, s/2)/2^s and the right-hand side of condition (7) for the proposed b with C > log 2 for large even s (e.g., s = 10^4); failure of the inequality at any s collapses the construction. Running the hierarchical percolation recursion numerically for these parameters and observing a nonzero endpoint connectivity would also contradict the theorem's conclusion.","supporting_citations":[],"review_version":1}