{"id":"dc02ac3e-e187-4104-9dbf-ffde7b8cbb37","arxiv_id":"2505.23364","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":2,"one_line_summary":"Generic random groups at extremely low density have a unique entropy-minimizing normalized weight, which is nearly uniform.","lead":"A math paper proves that most random groups built with very few long relators have a unique weighted metric minimizing volume entropy, and that the optimal weights are close to uniform. It introduces a small-cancellation condition that guarantees this uniqueness.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Prop 4.8's lower bound on |u|_w uses a factor-of-2 error: condition (iii) yields N|u|/(16m), not N|u|/(8m), so the proof of Theorem C as written is invalid; the gap is repairable.","rationale":"The paper's central rigidity result (Theorems A and B) rests on the translation-apparent condition and the external Cantrell-Tanaka rigidity theorem; I do not see a flaw in that core argument. However, the paper also advertises Theorem C, and its proof depends on Proposition 4.8. There the third even-distribution condition is misread: a lower bound on each pair count #_{a±}(u) is summed as though it applied separately to every letter, producing a bound twice as strong as what actually follows. This invalidates the proof of Proposition 4.8 as written, and therefore the proof of Theorem C, although the error appears repairable without changing the main theorems. The reader's identified external-theorem dependence is a legitimate concern but is a conditional reliance on published work rather than an internal inconsistency; the factor-of-2 error is a concrete internal gap that should be fixed before acceptance. I recommend CONDITIONAL rather than ACCEPT: the paper's main rigidity claims seem sound, but the manuscript must be revised to correct the Proposition 4.8 estimate and to check that the stated hypotheses still suffice under the corrected bound.","tokens_in":28002,"tokens_out":51719,"duration_ms":548883,"concrete_test":"Re-derive the estimate of q(b) in Proposition 4.8 using condition (iii) literally: replace the exponent Nl/(8m) by Nl/(16m) and redo the final inequality. The needed sufficient condition becomes 8mjl < M0^{N(l/(32m)−2)}. Then verify whether the paper's displayed hypothesis 8mjl < (2m−1)^{l/(16m)−2} implies this, using M0^N ≥ (2m−1)^{2m} and l > 32m. If the implication fails for some m or l, Proposition 4.8 and hence Theorem C require a revised hypothesis; if it holds, the stated results stand but the proof must be amended.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Proposition 4.8 estimates q(b) by claiming that the third even-distribution condition gives |u|_w > N|u|/(8m) for every u in R_{λ,s}. Condition (iii) of Definition 4.1 says only that #_{a±}(u) > |u|/(8m) for each pair a ∈ A. Summing over the m pairs gives |u|_w = Σ_a #_{a±}(u) w(a) > (|u|/(8m)) Σ_{a∈A} w(a) = N|u|/(16m), because N = Σ_{s∈S} w(s) = 2 Σ_{a∈A} w(a). The factor of 2 changes the subsequent bound on q(b): the valid estimate is q(b) ≤ 2mj / b^{Nl/(16m)}, not 2mj / b^{Nl/(8m)}. The proof of Proposition 4.8 then rewrites the final inequality using this wrong denominator, so the displayed derivation of p(b) < 0 is invalid. This is load-bearing for Theorem C because Proposition 4.9 and the uniform entropy approximation h(F_m,w) − 2/l ≤ h(G,w) both invoke Proposition 4.8. The issue is localized and likely repairable: since M0^N = e^{h(F_m,w/N)} ≥ e^{h(F_m,w*)} = (2m−1)^{2m}, the corrected sufficient condition 8mjl < M0^{N(l/(32m)−2)} is implied by the paper's hypothesis 8mjl < (2m−1)^{l/(16m)−2} for large l. But the written proof needs correction.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the rigidity of the volume entropy for weighted word metrics on hyperbolic groups. It introduces the notion of a λ-translation-apparent presentation, which combines the C′(λ) small cancellation condition with three 'even distribution' conditions on relators. The main results are: Theorem A, asserting that a torsion-free hyperbolic group admitting such a presentation has a unique normalized weight minimizing the volume entropy; Theorem B, asserting that for each m ≥ 2 there is an explicit small density d_m such that a generic random group on m letters at density d < d_m admits a 1/16-translation-apparent presentation; and Theorem C, asserting that such random groups have entropy arbitrarily close to that of the free group for every weight, that the unique minimizer is arbitrarily close to the uniform weight, and that the minimum entropy is arbitrarily close to that of the free group. The proofs combine small cancellation theory, a counting method of Myers for weighted subword avoidance, a Chernoff bound for Markov chains, and the recent convexity and length-spectrum results of Cantrell–Tanaka. The paper also proves a stability result for minimizers when uniqueness holds.","tokens_in":28374,"tokens_out":26799,"duration_ms":238921,"significance":"The paper makes a valuable contribution to the study of volume entropy rigidity for weighted word metrics. The notion of a translation-apparent presentation is a new and potentially useful tool, and the explicit genericity threshold in Theorem B gives a concrete family of hyperbolic groups where the minimizer is unique and almost uniform. The proof of Theorem C relies on a detailed counting argument with Myers' generating functions, and the paper provides both explicit constants and a self-contained appendix. If the technical gaps identified below are repaired, the results will constitute a solid advance. The main external dependency, the Cantrell–Tanaka length-spectrum rigidity theorem, is clearly stated and used in a standard way. The paper is well organized and the exposition is generally clear, though several proof details need correction.","major_comments":[{"comment":"The lower bound on |u|_w uses an incorrect interpretation of the third even distribution condition. Definition 4.1(iii) states that for each pair a ∈ A, #_{a±}(u) > |u|/(8m). It does not imply #_s(u) > |u|/(8m) for each individual s ∈ S. Summing the pair contributions gives |u|_w = Σ_a #_{a±}(u) w(a) > (|u|/(8m)) Σ_{a∈A} w(a) = N|u|/(16m), not N|u|/(8m). Consequently, the estimate q(b) ≤ 2mj/b^{Nl/(8m)} should read q(b) ≤ 2mj/b^{Nl/(16m)}, and the subsequent derivation of p(b) < 0 is invalid. This error propagates to Proposition 4.9 and Theorem C. A repair is to replace the sufficient condition by 8mjl < M_0^{N(l/(32m)-2)} and to choose ℓ in Theorem 5.13 accordingly; however, the hypothesis as stated, 8mjl < (2m−1)^{l/(16m)-2}, does not imply the stronger corrected condition, so the statement of Proposition 4.8 must be amended.","section":"Section 4.2, Proposition 4.8"},{"comment":"The proof applies the second even distribution condition to the word u3, whose length is only known to be less than ⌈4λ|r|⌉, whereas condition (ii) is formulated for subwords of length exactly ⌈4λ|r|⌉. The argument should first extend u3 to a subword of r of that length, which is possible because u3 is a suffix of a cyclic permutation of r, and then apply condition (ii). As written, this step is unjustified. Since Proposition 4.3 underpins the proof of Theorem A via Corollary 4.4, this detail needs to be fixed.","section":"Section 4.1, Proposition 4.3"}],"minor_comments":[{"comment":"In the proof, the line 'the distribution of uj is the uniform distribution on reduced words of length ⌈ℓ/16⌉' should read '⌈ℓ/4⌉'.","section":"Lemma 5.7"},{"comment":"The cited formula for the number of cyclically reduced words, (2m−1)^ℓ + m + (−1)^ℓ(m−1) ≥ (2m−1)^ℓ, appears to be incorrect: for m=2, ℓ=2 it would give 15, exceeding the total number of reduced words of length 2. Only the exponential rate matters for the subsequent argument, so the proof can be repaired by using the asymptotic count c_m(2m−1)^ℓ, but the displayed inequality should be corrected.","section":"Lemma 5.10"},{"comment":"The bound j ≤ |R_ℓ^*| ≤ ℓ(2m−1)^{dℓ} omits the factor coming from symmetrization; each relator gives at most 2ℓ cyclic permutations and inverses, so the correct upper bound is |R_ℓ^*| ≤ 2ℓ(2m−1)^{dℓ}. The argument is asymptotic, so this does not change the main conclusion, but the displayed inequality should be fixed.","section":"Theorem 5.13"},{"comment":"There are numerous typos and spacing errors in the text, such as 'W e', 'a', and 'A T' in the title, which should be corrected in a revision.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The reader's report recommended acceptance with moderate confidence, but the factor-of-2 error in Proposition 4.8 invalidates the proof of Theorem C as written. The error is localized and repairable, so the paper is suitable for major revision rather than rejection. The author should also check the counting formula in Lemma 5.10 and the symmetrization bound in Theorem 5.13. The paper's reliance on the Cantrell–Tanaka theorem is acceptable but should be clearly flagged as an external premise."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this paper deserves a referee, but not a desk reject. The new idea is translation-apparent presentations, and the proof that generic low-density random groups admit them is clever. Theorems A, B, C are meaningful advances. The main issue I found is a factor-of-2 mistake in the proof of Proposition 4.8. The third even-distribution condition yields |u|_w > N|u|/(16m), not N|u|/(8m) as written. This propagates to the bound on q(b), so the displayed derivation of p(b)<0 is wrong. The gap is repairable: with the corrected exponent, the paper's hypothesis 8mjl < (2m−1)^{l/(16m)−2} still implies the needed inequality for large l, so the theorem statements survive. But the proof needs editing before publication. That's the main thing. The rest of the paper is in good shape. The convexity argument in Prop 3.6 is clean, and the use of Cantrell-Tanaka's length-spectrum rigidity to get strict uniqueness from translation lengths is a natural and well-executed step. The surface-group application (Theorem 3.13) is a nice bonus. The probabilistic estimates in Section 5 are careful; the Markov-chain Chernoff bound is applied correctly, and the explicit density d_m is a welcome touch even if it is tiny. The reliance on Cantrell-Tanaka (Theorem 2.3) is heavy but legitimate: it's a published external theorem, and the paper cites it accurately. The density threshold d_m → 0 as m → ∞ is a limitation but the paper says so openly, and the few-relator d=0 case is still covered. Who is this for? Geometric group theorists working on entropy rigidity, random groups, and growth. The paper is readable and well structured. It deserves a serious referee, with the expectation that Prop 4.8 will be fixed. Recommendation: send it out, but ask the referee to check the constants in Prop 4.8 and Theorem C. I'd probably want to re-check the derivation of the corrected bound in the revision.","headline":"A genuinely new rigidity theorem for random groups at low densities, with a small but fixable factor-of-2 error in Proposition 4.8.","tokens_in":28911,"tokens_out":7907,"would_cite":true,"duration_ms":68584,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20F67","20F65","20P05"],"pacs":[],"model":"deepseek-v4-flash","headline":"For a torsion-free hyperbolic group with a translation-apparent presentation, the volume entropy has a unique normalized minimizing weight, and such presentations are generic for low-density random groups.","keywords":["volume entropy","weighted word metrics","hyperbolic groups","random groups","small cancellation","rigidity","length spectrum","translation-apparent presentation"],"falsifier":"For m=2 at a density such as d=d_2/2, sample random presentations with relator length ℓ up to roughly $10^{4}$ and test whether the symmetrized presentation satisfies the three even-distribution conditions of Definition 4.1 with λ=1/16; the paper predicts an exponentially small failure probability in ℓ, so observing a failure rate that stays bounded away from zero as ℓ grows would refute the genericity claim of Theorem B.","tokens_in":27767,"feed_emoji":"🎲","tokens_out":11843,"duration_ms":117706,"temperature":0.7,"pith_summary":"The paper proves that, for a large class of hyperbolic groups, assigning positive weights to the generators (normalized to sum to 1) produces exactly one weighted word metric that minimizes the volume entropy, the exponential growth rate of balls. The enabling object is a 'translation-apparent presentation': a small-cancellation presentation in which each generator appears evenly inside every relator at three separate scales. The paper shows that in such a presentation every power of a generator is a geodesic with respect to every weight, so the stable translation length of a generator equals its weight; distinct weights therefore yield non-comparable length spectra, and an external rigidity theorem converts this into strict convexity and uniqueness of the minimizer. It then proves that these presentations are generic in Gromov's density model of random groups at any density below an explicit threshold d_m, and that for such generic groups the unique minimizer is almost the uniform weight and the minimal entropy is almost the free group's entropy. The result matters because volume entropy is the natural normalizable invariant for comparing metrics on a group, and this is a setting where rigidity can be located precisely and the minimizer described explicitly.","feed_headline":"Most random groups pick a single best weighting of generators","feed_subtitle":"The unique winning weight sits near uniform and matches the free group’s growth rate.","key_machinery":"The load-bearing object is the λ-translation-apparent presentation: a symmetrized, cyclically reduced presentation satisfying the C'(λ) small-cancellation condition together with three even-distribution conditions — no relator contains a subword s^n with n≥λ|r|, no subword of length ⌈4λ|r|⌉ contains half of any generator's occurrences in the relator, and every subword of length ⌈λ|r|⌉ contains at least 1/(8m) of its letters equal to any given generator. The small-cancellation side, through the lemma of [14], shows that any word representing the identity must contain a long piece of a relator; the even-distribution side ensures that such a piece cannot have small weight, so any word avoiding long relator subwords (in particular any power of a generator) is the unique geodesic for every weight. This yields the identity ℓ_{d_w}(s)=w(s) for each generator, the direct link from weights to the length spectrum. The third even-distribution condition also feeds the weighted subword-avoidance generating functions of [28], which produce the free-group entropy bounds in Proposition 4.8 and hence Theorem C.","core_discovery":"The central claim is Theorem A (Section 4.5): if a torsion-free non-elementary hyperbolic group (G,S) admits a λ-translation-apparent presentation, then the volume entropy is strictly convex on the simplex of normalized weights on S, and there is a unique normalized weight minimizing it. The proof shows that in such a presentation every power of a generator s is a geodesic with respect to every weight w, so the stable translation length of s equals w(s); hence distinct weights give distinct length spectra, and the length-spectrum rigidity theorem cited as [6] upgrades this to non-rough-isometry of the two weighted metrics, which the convexity-up-to-rough-isometry result rules out for two minimizers. The genericity claim is Theorem B (Section 5.11): for each m≥2 and each density 0≤d