{"id":"7469f4e7-6f7e-4b2e-8679-5d92c366c2ed","arxiv_id":"2411.17298","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Generalized q-dimensions of measures on nonautonomous similar and affine attractors are estimated from contraction data, with exact formulas under separation or random-translation conditions.","lead":"This paper derives formulas for generalized (Renyi) q-dimensions of measures on nonautonomous fractals, where the construction rules change level by level. It gives upper and lower bounds, and exact formulas under separation or random-translation assumptions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 2.9 rests on an unproved, imported Lemma 5.1 that is false under the paper's stated 'absolutely continuous' hypothesis: unbounded translation densities make the conditional expectation infinite. The theorem needs a bounded-density hypothesis (or equivalent) and a proof of Lemma 5.1.","rationale":"The reader's weakest-assumption identification is the same as mine: Lemma 5.1 is the load-bearing imported ingredient for the main random-affine theorem. My stress-test sharpens the concern: the lemma is not merely unproved in this paper, it is false under the paper's stated absolute-continuity hypothesis because unbounded densities can make the conditional expectation infinite. This is a decisive gap for Theorem 2.9, since Proposition 5.2, the Borel-Cantelli step, and the final almost-sure lower bound all depend on that estimate. A bounded-density hypothesis would likely repair the argument, because then the conditional integral is controlled by the singular-value-function bound of Lemma 6.1; however, adding such a hypothesis changes the theorem statement and requires a proof of Lemma 5.1 that is not present. The reader also noted the false step in Proposition 2.6 about maximal cut-set words and omitted q=1 proofs; those are real but secondary, since the paper's advertised central claim for random nonautonomous affine sets is Theorem 2.9. Therefore I agree with the CONDITIONAL verdict: the central result is not verified from the text and needs both a strengthened hypothesis and a complete proof of its main lemma.","tokens_in":37794,"tokens_out":7199,"duration_ms":75079,"concrete_test":"Verify Lemma 5.1 directly in d=1 with T_{k,i}=1/2 and iid translation density f(t)=c|t|^{-1/2} on [-1,1]. Choose u,v with u wedge v=empty and condition F to fix all translations except omega_{u|1} so that the conditional difference is exactly omega_{u|1}. Compute E(|omega_{u|1}|^{-s}|F) = c * integral_{-1}^{1} |t|^{-1/2-s} dt, which is infinite for non-integral s=3/4, contradicting the finite bound C/psi_s(I). If instead the density is required to be bounded, re-run the proof of Lemma 5.1 using Falconer's Lemma 6.1 and check whether the conditional expectation bound holds uniformly over all u,v and the nonautonomous sequence.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central almost-sure lower bound for random nonautonomous affine sets (Theorem 2.9) is driven by Proposition 5.2, whose key estimate is Lemma 5.1: E(|Pi_omega(u)-Pi_omega(v)|^{-s} | F) <= C / psi_s(T_{u wedge v}). Lemma 5.1 is not proved here; it is imported from the authors' unpublished preprint [22]. More importantly, the stated hypotheses of Section 2.3 only require each translation distribution Pu to be absolutely continuous with respect to Lebesgue measure. Under that hypothesis Lemma 5.1 is false. In d=1, take T_{k,i}=1/2 for all k,i and translations iid with density f(t)=c|t|^{-1/2} on a neighborhood of 0, which is absolutely continuous but unbounded. For u,v with u wedge v = empty, condition F so that every translation except omega_{u|1} is fixed and arranged so that the deterministic part of Pi_omega(u)-Pi_omega(v) is 0. Then the conditional expectation is c * integral of |t|^{-1/2} |t|^{-s} dt. For non-integral s=3/4, this integral diverges, so the left side is infinite while the right side is C/psi_s(I)=C. Thus Lemma 5.1 fails exactly in the regime used by Theorem 2.9. No bounded-density or uniform absolute-continuity assumption appears anywhere in the paper, so the proof of Theorem 2.9 cannot be completed as stated. The theorem may be salvageable with an added bounded-density hypothesis and a self-contained proof of Lemma 5.1, but that is a substantive repair, not a typographical fix.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper defines critical values d_q^- and d_q^+ via sums of cylinder weights and singular value functions for nonautonomous iterated function systems, and claims to determine the lower and upper generalized q-dimensions of the projected measures. For general nonautonomous attractors the authors prove upper bounds (Theorem 2.1) and, under a gap separation condition, matching lower bounds (Theorem 2.2). For nonautonomous similar sets with strong separation they give the formula D_q = min{d_q^-, d} (Theorem 2.4). For nonautonomous affine sets they prove upper bounds (Theorem 2.8) and then treat two special classes: random translations (Theorem 2.9) and finitely many translations (Theorem 2.12). The central new lower-bound results for random nonautonomous affine sets rest on Lemma 5.1, which is imported from the authors' unpublished preprint [22].","tokens_in":38158,"tokens_out":6211,"duration_ms":56756,"significance":"If the main theorems are correct, the paper gives a natural parameter-free description of generalized q-dimensions for a broad class of nonautonomous fractals, extending Falconer's almost self-affine results to a nonautonomous setting; the critical values d_q^- and d_q^+ are defined by convergence of intrinsic cylinder sums, and the upper-bound arguments genuinely compare those sums with mesh sums, so there is no circular fitting of parameters. The combinatorial join-class machinery in Section 5 is a serious adaptation of Falconer's approach to nonautonomous trees. However, the lower-bound claims for the random-translation model and for the homogeneous-ratio similar case are currently not supported by valid proofs: the key Lemma 5.1 is both unproved and, under the paper's stated hypotheses, false, and the proof of Proposition 2.6 contains an incorrect cut-set assertion.","major_comments":[{"comment":"Lemma 5.1 is not proved in this paper; it is cited to the unpublished preprint [22], yet it is the engine of Proposition 5.2 and therefore of the almost-sure lower bound in Theorem 2.9. More seriously, the lemma is false under the hypotheses stated in Section 2.3, where each translation distribution P_u is only assumed to be absolutely continuous with respect to Lebesgue measure. In d=1, take T_{k,i}=1/2 for all k,i and let the translations be iid with density f(t)=c|t|^{-1/2} on a neighborhood of 0 (absolutely continuous but unbounded). For u,v with u∧v=∅, take F to be the sigma-field generated by all translations except ω_{u_1}, and arrange the fixed translations so that the deterministic part of Π_ω(u)-Π_ω(v) is 0. Then the conditional expectation equals c ∫ |t|^{-1/2}|t|^{-s} dt, which is infinite for non-integral s=3/4, while the right-hand side C/ψ_s(T_{u∧v}) is finite. Thus the asserted uniform bound fails exactly in the regime used by Theorem 2.9. A bounded-density (or uniform absolute-continuity) hypothesis would be needed, together with a self-contained proof, before Theorem 2.9 can be accepted as stated.","section":"Section 5, Lemma 5.1"},{"comment":"The proof of Proposition 2.6 asserts that for each u∈Σ*(s,r) of maximal length K_1, 'it is clear that u^{-}j∈Σ*(s,r) for all j∈{1,...,n_{K_1}}'. This is false: since c_{u^{-}j}=c_{u^{-}}c_{K_1,j}, the condition c_{u^{-}j}≤r is equivalent to c_{K_1,j}≤r/c_{u^{-}}, which need not hold for every child j of u^{-}, even though c_{u}=c_{u^{-}}c_{K_1,w_{K_1}}≤r. For example, if r/c_{u^{-}}<c_{K_1,j}<1 for some j, then u^{-}j∉Σ*(s,r). Because this incorrect assertion is used to derive the simplified expressions (2.15)-(2.18), the proof of Proposition 2.6 is incomplete, and consequently Corollary 2.7, which depends on those expressions, is not established by the given argument. A corrected proof would need to sum over the actual maximal elements of the cut set rather than over all children of a maximal word.","section":"Section 3, proof of Proposition 2.6"}],"minor_comments":[{"comment":"The line 'E(|Π_ω(u_1)-Π_ω(u_2)|^{-s}...|Π_ω(u_n)-Π_ω(v)|^{-s}) = E(E(X_1...X_n|F_n)|F_n)' is a typo; the right-hand side should be E(E(X_1...X_n|F_n)), not the nested conditional expectation with the same sigma-field twice.","section":"Section 5, proof of Proposition 5.2"},{"comment":"The notation u^{-}j is never defined in the paper; it appears to mean the word obtained by appending the letter j to u^{-}, but the notation is ambiguous and should be defined explicitly.","section":"Section 3"},{"comment":"There are numerous typographical errors, including 'mutltifractal' in Section 1.1, 'nonautonommous' in Sections 2.1 and 2.3, 'generlized' at the start of Section 5, and inconsistent accenting of names such as Rényi and Barański; a careful proofreading pass is needed.","section":"Throughout"},{"comment":"The change of variables in equation (6.22) is not fully explained: after defining y=a_1-a_2+H(a), the treatment of a_2,...,a_τ as independent variables and the domain of integration for the Jacobian should be stated explicitly so that the application of Lemma 6.1 is transparent.","section":"Section 6, proof of Theorem 2.12"},{"comment":"The lemma should specify that the inequality holds almost surely with respect to the underlying product measure, and the dependence of the constant C on the hypotheses (e.g., on any boundedness or moment condition on the densities) should be stated; otherwise the inequality is not a well-defined pointwise assertion.","section":"Section 5, Lemma 5.1"}],"recommendation":"major_revision","confidential_remarks":"The paper's most substantial new claim, Theorem 2.9, rests entirely on Lemma 5.1, which is imported from the authors' own unpublished preprint and is in fact false under the stated 'absolutely continuous' hypothesis. This is a correctability problem only if the authors add a bounded-density or equivalent hypothesis and provide a full proof of the lemma within the paper. The error in the proof of Proposition 2.6 is also substantive and affects Corollary 2.7. Given the otherwise sound upper-bound framework and the clear value of the intended results, I recommend major revision rather than rejection, but the authors should be told plainly that the random-translation theorem and the simplified Bernoulli critical-value formulas are currently unproved."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this paper has a real core and a serious hole. The clean upper bounds (Theorems 2.1 and 2.8) are sound, and Theorem 2.4 for nonautonomous similar sets under SSC is a genuine and probably correct extension of Cawley-Mauldin. The random affine lower bound, however, does not hold as stated.\n\nThe genuinely new items are the nonautonomous similar-set formula and the q=1 cases. The proof of Theorem 2.4 is elementary and convincing; the upper-bound arguments are routine but correct. The finitely-many-translations theorem (2.12) uses Falconer's Lemma 6.1 and looks self-contained, though Corollaries 2.13 and 2.14 are dismissed with \"proof is similar\"—a real omission for advertised improvements.\n\nNow the soft spots. First, Proposition 2.6 contains a false step: it claims that for a maximal-length u in Sigma*(s,r), every sibling (u-)j is also in Sigma*(s,r). That is not automatic. When c_{u-} is just above r and c_{K1,j} is close to 1, the product c_{u-}c_{K1,j} can exceed r. The proof of the simplified critical-value formulas needs repair.\n\nSecond, and far more important, Lemma 5.1 is imported from your own unpublished preprint and is false under the hypotheses stated in Section 2.3. You only assume the translation distributions are absolutely continuous. With an unbounded density, say f(t) ~ |t|^{-1/2} in d=1, the conditional expectation in Lemma 5.1 diverges for s=3/4, while the claimed bound is finite. You need a bounded-density or uniform absolute-continuity hypothesis, and you need to prove Lemma 5.1 in the paper. Without that, Theorem 2.9 and the q=1 extension in Corollary 2.11 do not follow. This is not a typo; it is a load-bearing gap.\n\nWho gets value from this? People working on dimension theory of nonautonomous IFS will want the similar-set result, and the upper bounds are a useful reference. But no one should rely on the random affine lower bound as currently written.\n\nRecommendation: send this to peer review, but expect major revision. The similar-set part may be publishable on its own; the random affine part needs a fixed lemma and a self-contained proof, or an explicit bounded-density assumption.","headline":"Solid upper bounds and a likely correct similar-set formula, but the advertised random affine lower bound rests on a false imported lemma and needs substantive repair.","tokens_in":38724,"tokens_out":4839,"would_cite":true,"duration_ms":43153,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["28A80","37C45"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper establishes that the generalized (Lq) dimensions of measures supported on nonautonomous fractals are governed by critical values of weighted singular-value sums, and proves exact formulas for nonautonomous similar sets and for…","keywords":["generalized q-dimensions","nonautonomous attractors","singular value function","almost self-affine sets","random translations","Lq-spectrum","Moran sets"],"falsifier":"Simulate or compute both sides of Lemma 5.1 for a two-level nonautonomous affine system in $R^{2}$ whose translation distribution is absolutely continuous but has an unbounded density concentrated near a point, and check whether the ratio $E(|\\Pi(u)-\\Pi(v)|^{-s}|\\mathcal{F})/\\psi^s(T_{u\\wedge v})$ is bounded uniformly over pairs (u,v); an unbounded ratio would disprove the existence of the constant C and invalidate the almost-sure formula.","tokens_in":37540,"feed_emoji":"📐","tokens_out":7955,"duration_ms":66590,"temperature":0.7,"pith_summary":"Generalized q-dimensions measure the fluctuation of a probability measure at small scales. This paper asks what they are for the natural measures sitting on nonautonomous fractals—attractors of iterated function systems whose contraction ratios, and even the number of contractions, change from level to level, so ergodic-theoretic tools are unavailable. The paper's claim is that, for large q at least, these dimensions are pinned by a single critical value: the threshold s where the weighted sums $\\sum \\psi^s(T_u)^{1-q} \\mu(C_u)^q$ over all cylinder words change from summable to divergent, with $\\psi^s$ the singular value function of the accumulated linear parts. The authors establish this as a universal upper bound for all nonautonomous affine sets, as an exact formula for nonautonomous similar sets with strong separation, and as an almost-sure formula for affine sets whose translations are randomly perturbed (q>1) or chosen from a finite set (1<q<=2). If the formulas hold, generalized q-dimensions of such measures become computable from the contraction data and cylinder weights alone.","feed_headline":"A critical sum decides the q-dimension on nonautonomous fractals","feed_subtitle":"Generalized dimensions of projected measures are pinned by the threshold where cylinder-mass sums diverge.","key_machinery":"The load-bearing object is the singular value function $\\psi^s(T)$, defined as the product of the first $m-1$ singular values of $T$ times the $m$-th singular value raised to $s-m+1$ (with $s>d$ handled by the determinant power). The critical exponents $d^-_q$ and $d^+_q$ are defined by requiring the weighted sums $\\sum \\psi^s(T_u)^{1-q}\\mu(C_u)^q$ over cylinder sets to be summable or lim-sup bounded; these sums act as the effective entropy of the nonautonomous system. For the lower bounds, the proof uses the multienergy kernel $\\psi^s(u_1,\\ldots,u_n,v)$, formed by multiplying $\\psi^s$ at the vertices of the join set of the words, together with a hierarchy of join-class estimates (Propositions 5.2–5.4) that convert the singular-value weights of a tree into volume estimates for $r$-balls under random translations.","core_discovery":"On its own terms, the paper's central discovery is Theorem 2.9: for a nonautonomous affine set with i.i.d. random translations (each absolutely continuous with respect to Lebesgue measure), and for any q>1, almost surely $\\underline{D}_q(\\mu_\\omega)=\\min\\{d^-_q,d\\}$, where $d^-_q=\\sup\\{s:\\sum_{k=1}^\\infty\\sum_{u\\in\\Sigma^k}\\psi^s(T_u)^{1-q}\\mu(C_u)^q<\\infty\\}$. Alongside it stands Theorem 2.4, giving two-sided formulas for nonautonomous similar sets under strong separation: $\\underline{D}_q(\\mu_\\omega)=\\min\\{d^-_q,d\\}$ and $\\overline{D}_q(\\mu_\\omega)=\\min\\{d^+_q,d\\}$. The paper also proves that these critical values are universal upper bounds for all nonautonomous affine sets (Theorem 2.8), and, for the finite-translation family, obtains the lower bound for $1<q\\le 2$ (Theorem 2.12), with the identical-Ξ case extended to $q\\ge 1$ (Corollary 2.11).","pith_inferences":["If Theorem 2.9 is correct, the lower generalized q-dimension of a random nonautonomous affine measure is a geometric quantity controlled by the linear part and the cylinder masses alone, independent of the fine details of the translation law once it is absolutely continuous; this suggests the same critical-sum formula may extend to other families of random non-conformal fractals.","The proof's dependence on the conditional expectation inequality in Lemma 5.1, imported from the authors' unpublished preprint, means the almost-sure upper bound should be read as conditional on that lemma; checking whether the inequality survives for translation laws with density zero sets or non-i.i.d. choices would delimit the theorem's true scope.","One could test the formula numerically on a simple nonautonomous system with slowly varying contraction ratios, where $d^-_q$ and $d^+_q$ differ, and compare $\\underline{D}_q(\\mu_\\omega)$ to $\\min\\{d^-_q,d\\}$ almost surely; agreement would support the conjecture that the lower generalized dimension is always the summability threshold."],"forward_implications":["For every nonautonomous similar attractor satisfying strong separation, the generalized q-dimensions of any projected measure are explicitly given by the critical values $\\min\\{d^-_q,d\\}$ and $\\min\\{d^+_q,d\\}$; when the two critical values agree, $D_q(\\mu_\\omega)$ exists.","For nonautonomous affine sets with random translations, the lower generalized q-dimension is almost surely $\\min\\{d^-_q,d\\}$ for q>1, and the paper upgrades the earlier almost-self-affine result from q>1 to q>=1 when the linear parts are identical.","For self-affine sets built with finitely many allowed translations and contraction norms below 1/2, the formula $D_q(\\mu_a)=\\min\\{d_q,d\\}$ holds for $1\\le q\\le 2$ for Lebesgue-almost every translation vector.","The upper bound $\\overline{D}_q(\\mu_\\omega)\\le\\min\\{d^+_q,d\\}$ holds with no separation or randomness assumption, so the critical value is a universal ceiling for nonautonomous affine measures.","For Bernoulli measures, Proposition 2.6 reduces the critical values to products over levels of $\\sum_{j=1}^{n_i} c_{i,j}^{s(1-q)} p_{i,j}^q$, making the dimensions computable directly from contraction ratios and weights."],"supporting_citations":[{"why":"Unpublished preprint that supplies Lemma 5.1, the conditional expectation inequality that drives the almost-sure lower bound in Theorem 2.9.","marker":"[22]"},{"why":"Falconer's study of almost self-affine sets; supplies the q>1 dimension formula and the equality $d^-_q=d^+_q=d_q$ used in Corollary 2.11.","marker":"[13]"},{"why":"Introduces the singular value function and the integral estimate (Lemma 6.1) used to control differences of affine projections in the finite-translation case.","marker":"[10]"},{"why":"Falconer's bounds for generalized dimensions of self-affine measures on typical sets, the baseline that Corollary 2.14 extends to q=1 and to the nonautonomous setting.","marker":"[12]"},{"why":"Jordan–Pollicott–Simon's random-perturbation model of almost self-affine sets, which motivates the random-translation version studied here.","marker":"[27]"},{"why":"Peres–Solomyak's integral representation and existence results for Lq dimensions, used in Proposition 4.1 to pass between cube sums and integrals over balls.","marker":"[45]"}],"fun_headline_variants":["Critical sum decides q-dimension on nonautonomous attractors","Series threshold yields exact dimension for random affine fractals","Generalized q-dimensions from convergence of a cylinder sum","Nonautonomous fractals: a critical series pins the q-dimension"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof of the almost-sure lower bound depends on Lemma 5.1, a conditional expectation inequality imported from the authors' own unpublished preprint; if that inequality fails outside its stated hypotheses—for instance for translation distributions with unbounded density or for non-i.i.d. choices—the formula in Theorem 2.9 does not follow.","fun_headline_variants_meta":{"raw":{"variants":["Critical sum decides q-dimension on nonautonomous attractors","Series threshold yields exact dimension for random affine fractals","Generalized q-dimensions from convergence of a cylinder sum","Nonautonomous fractals: a critical series pins the q-dimension"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000766,"raw_usage":{"total_tokens":3378,"prompt_tokens":911,"completion_tokens":2467,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":527,"completion_tokens_details":{"reasoning_tokens":2409}},"tokens_in":527,"tokens_out":2467,"duration_ms":18827,"temperature":1.0,"reasoning_tokens":2409,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T12:16:16.531140+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate or compute both sides of Lemma 5.1 for a two-level nonautonomous affine system in $R^{2}$ whose translation distribution is absolutely continuous but has an unbounded density concentrated near a point, and check whether the ratio $E(|\\Pi(u)-\\Pi(v)|^{-s}|\\mathcal{F})/\\psi^s(T_{u\\wedge v})$ is bounded uniformly over pairs (u,v); an unbounded ratio would disprove the existence of the constant C and invalidate the almost-sure formula.","supporting_citations":[{"cited_title":"Gu and J","cited_arxiv_id":null,"evidence_quote":"Unpublished preprint that supplies Lemma 5.1, the conditional expectation inequality that drives the almost-sure lower bound in Theorem 2.9."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Falconer's study of almost self-affine sets; supplies the q>1 dimension formula and the equality $d^-_q=d^+_q=d_q$ used in Corollary 2.11."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the singular value function and the integral estimate (Lemma 6.1) used to control differences of affine projections in the finite-translation case."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Falconer's bounds for generalized dimensions of self-affine measures on typical sets, the baseline that Corollary 2.14 extends to q=1 and to the nonautonomous setting."},{"cited_title":"Jordan, M","cited_arxiv_id":null,"evidence_quote":"Jordan–Pollicott–Simon's random-perturbation model of almost self-affine sets, which motivates the random-translation version studied here."},{"cited_title":"Peres and B","cited_arxiv_id":null,"evidence_quote":"Peres–Solomyak's integral representation and existence results for Lq dimensions, used in Proposition 4.1 to pass between cube sums and integrals over balls."}],"review_version":1}