{"id":"b50ddf8b-15f2-4325-b71c-2d321394bec8","arxiv_id":"2412.16753","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For an exponentially separated real analytic IFS without a common fixed point, every self-conformal measure has dimension equal to min{1, entropy divided by Lyapunov exponent}.","lead":"This paper proves a formula for the fractal dimension of measures generated by nonlinear analytic iterated function systems with exponentially separated maps. It extends a celebrated result for self-similar systems and introduces a technique for reducing infinite-dimensional analytic maps to finite-dimensional polynomials.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified","rationale":"I stress-tested the reader's flagged assumption and also the main internal steps. The reader's weakest_assumption is not, in my reading, load-bearing: [6, Theorem 2.8] is the standard Feng-Hu exact-dimensionality/fiber-entropy formula for conformal IFS and does not require exponential separation; the no-common-fixed-point and derivative assumptions match the setup. The exponential-separation scale issue I initially suspected is resolved by the downward propagation argument described above, since Definition 1.1's infinitely many good scales can be pulled back to the scales n+n' with only a bounded loss in the separation constant. The entropy-increase mechanism is a faithful adaptation of Hochman's inverse theorem: the linearization Lemma 3.9 and the uniform entropy dimension Proposition 3.3 supply exactly the inputs required by Theorem 3.2, and the parameter hierarchy in (3.14) and (4.4) is flexible enough to choose the constants in the right order. The only genuine dependency is on prior results ([6], [7], [9], [1]), all standard, and no circularity, fitting, or omitted contradiction is apparent. Hence I keep the ACCEPT verdict unchanged.","tokens_in":22116,"tokens_out":49517,"duration_ms":455863,"concrete_test":"Verify the exact statements of Feng-Hu's [6, Theorem 2.8] and [6, Proposition 4.10]: specifically that the theorem gives dim mu=(H(p)-H(beta,C1|Pi^{-1}BR))/chi for a C^{1+gamma} conformal IFS without any additional separation assumption, and that the proposition identifies the limit (1/l)H(beta_omega,C_l) with this same one-step conditional entropy. If either assertion is false, the contradiction setup in Theorem 1.2 would lose its starting point.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I found no load-bearing flaw in the proof of Theorem 1.2. The three places that look most fragile check out. First, the imported identity [6, Theorem 2.8] is a standard unconditional result for C^{1+gamma} conformal IFS: no separation hypothesis enters, so the reduction of a dimension drop to Delta'>0 is legitimate. Second, the step (4.5) is justified despite Definition 1.1 requiring separation only at infinitely many scales: if the separation constant is c at a scale m, then every level l<=m satisfies S_l>=c^m; choosing n approximately m/(1+alpha) for alpha=Delta/(2log|Lambda|) makes n+n'=m-O(1), so a fixed power c'=c^{O(1)} serves at level n+n'. Third, the polynomial reduction and the use of Theorem 3.1 are internally consistent; the constants in (4.4) can be chosen successively so that rho is smaller than the entropy-increase constant supplied by Theorem 3.1 and delta is small compared with epsilon_0 rho. I therefore do not have a substantive objection to raise.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper proves (Theorem 1.2) that for a real-analytic iterated function system Phi on [0,1] with 0<|phi_i'(x)|<1 for all i and x, with no common fixed point, and exponentially separated in the sense of Definition 1.1, every positive self-conformal measure mu satisfies dim mu = min{1,H(p)/chi}. The proof proceeds by contradiction: a positive dimension drop is extracted from the Feng–Hu identity dim mu = (H(p)-Delta')/chi, and the exponential separation is used to force, through a Taylor approximation of analytic semigroup elements, a large entropy increase for convolutions of mu with measures on spaces of polynomials. The entropy increase result (Theorem 3.1) is the technical core of the paper. The paper also derives the corresponding formula for self-conformal sets (Corollary 1.3) and a zero-dimensional exceptional set statement for one-parameter analytic families (Corollary 1.4).","tokens_in":22259,"tokens_out":23345,"duration_ms":193667,"significance":"If correct, Theorem 1.2 is a significant extension of Hochman's inverse-theorem approach beyond finite-dimensional parameter spaces: it appears to be the first exponential-separation dimension result for self-conformal measures whose IFS is not contained in a finite-dimensional Lie group. The reduction of convolutions with measures on the infinite-dimensional space of real analytic maps to convolutions with measures on finite-dimensional polynomial spaces is a genuine new idea and is likely to be reused in other stationary fractal settings. The proof is carefully organized, with a transparent hierarchy of parameters (M << k << rho^{-1} << delta^{-1} << n) and with external tools (Hochman's inverse theorem, the Feng–Hu dimension formula) imported as black boxes; I found no internal inconsistency or circular dependence on the main conclusion. The corollaries, especially the parametric family statement, increase the paper's utility.","major_comments":[],"minor_comments":[{"comment":"The sentence 'By exponential separation and the choice of c, we may clearly assume that (4.5) holds' is too terse. Exponential separation gives a lower bound c^m only at infinitely many scales m, whereas the proof needs a uniform bound at level n+n' for the particular n chosen in the parameter hierarchy. The step is valid: one chooses m comparable to (1+alpha)n with alpha = Delta/(2 log |Lambda|) and replaces c by a fixed power c^{1+alpha}; please spell this out explicitly.","section":"Section 4, Eq. (4.5)"},{"comment":"The application of Lemma 3.9 to the component measures nu_{p,i} and theta_{x,i} suppresses the usual truncation of small i. For the lemma to apply, the component diameter 2^{-i} must be smaller than the delta supplied by Lemma 3.9; this holds for i >= i0, and the finitely many exceptional indices contribute an error O(i0/n) that is absorbed by taking n large in the hierarchy (3.14). A sentence making this truncation explicit would improve rigor.","section":"Section 3.4, around Eq. (3.15)"},{"comment":"Lemmas 3.8 and 3.9 are stated without proof and are said to follow from [10, Lemma 6.9] and [1, Lemma 4.2], respectively. Since Theorem 3.1 rests on these two lemmas, the paper should either include their proofs or state precisely how the referenced arguments adapt to the present notation and to the class of measures under consideration.","section":"Section 3.3, Lemmas 3.8 and 3.9"},{"comment":"The phrase 'basic basic properties of disintegrations' contains a duplicated word; also, the verification that the relevant properties hold for beta-almost every omega would read more cleanly if the two displayed limit statements were labelled.","section":"Lemma 4.1 proof"},{"comment":"The verification that Phi_n inherits the hypotheses of Theorem 1.2 (no common fixed point and exponential separation) is only sketched as 'easy to see'. A short justification would be helpful, since Corollary 1.3 depends on applying the main theorem to Phi_n.","section":"Corollary 1.3 proof"}],"recommendation":"minor_revision","confidential_remarks":"In my reading the central proof is sound and the requested changes are expository. The paper is a strong fit for the journal and should be accepted after the minor clarifications above are made."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The headline is that this paper delivers the first exponential-separation dimension formula for self-conformal measures outside finite-dimensional Lie group settings. Hochman covered self-similar maps; Hochman–Solomyak covered Möbius systems in SL(2,R). Rapaport gets all real analytic contractions with no common fixed point. That is a genuine extension, and the main novelty—reducing convolutions against the infinite-dimensional space of analytic maps to convolutions against bounded-degree polynomials—is clever and seems to work. The entropy increase argument for polynomial convolutions is carefully assembled, and the constant hierarchy M << k << ρ^{-1} << δ^{-1} << n is transparent enough to follow without getting lost.\n\nI looked at the spots that usually kill this kind of paper. The imported identity from Feng–Hu (dim μ = (H(p) − Δ')/χ) is unconditional for C^{1+γ} conformal IFS, so using it to detect a dimension drop is legitimate, not circular. The exponential separation step at (4.5) is justified by the scaling argument in the stress-test note: separation at infinitely many scales gives a uniform constant at the intermediate scales actually needed. The Taylor remainder estimate in (4.14) checks out, and the polynomial map p is bounded with derivative bounded away from zero, matching the hypotheses of Theorem 3.1.\n\nSoft spots are minor and mostly about delegation. Lemmas 3.8 and 3.9 are omitted with references to earlier work; they are plausibly routine, but a referee should verify them since they do real work in the entropy-increase proof. The paper is also honest about its limits: analyticity is used for derivative bounds and cannot be relaxed to C^∞ here, and the authors do not claim algebraic parameter systems are exponentially separated. None of this undermines the main theorem.\n\nThis is a paper for fractal geometry and dimension theory readers. It deserves a serious referee: the result is important within the area, the proof is inventive, and the exposition is unusually clear about the hierarchy of parameters. I would send it to review without hesitation.","headline":"Extends Hochman's inverse-theorem approach to real analytic IFSs, with a genuinely new finite-dimensional reduction; the proof looks sound and the main theorem is a real advance, though it leans on imported results.","tokens_in":22814,"tokens_out":994,"would_cite":true,"duration_ms":10564,"reading_group":"yes","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 proves that exponentially separated real-analytic iterated function systems without a common fixed point yield self-conformal measures whose dimension equals the entropy-to-Lyapunov ratio, capped at one.","keywords":["self-conformal measures","exponential separation","real analytic IFS","dimension of measures","entropy increase","Lyapunov exponent","iterated function systems","fractal geometry"],"falsifier":"The claim would be refuted by exhibiting an exponentially separated real-analytic IFS on [0,1] with no common fixed point, a positive probability vector p, and a self-conformal measure μ with dim μ < min{1, H(p)/χ}. A concrete check is whether the identity from [6, Theorem 2.8] holds for any such system; if not, that is the point to test. A numerical computation of dim μ for a polynomial IFS at algebraic parameters would be a practical test.","tokens_in":21862,"feed_emoji":"📐","tokens_out":9023,"duration_ms":67564,"temperature":0.7,"pith_summary":"The paper proves that the natural entropy-to-Lyapunov dimension formula holds for every self-conformal measure generated by a real-analytic iterated function system on the interval, provided the maps are uniformly contracting, have no common fixed point, and are exponentially separated. Earlier theorems of this kind were confined to self-similar systems or to maps acting through a finite-dimensional Lie group. The paper replaces the infinite-dimensional space of all real-analytic maps by a finite-dimensional space of polynomials of bounded degree, where an entropy-increase theorem can be applied. If correct, the result gives the first confirmation of the expected dimension formula for a broad class of genuinely non-linear, infinite-dimensional analytic IFSs, and yields as corollaries the corresponding dimension formula for self-conformal sets and for almost every member of a one-parameter analytic family.","feed_headline":"Analytic IFS measures hit the expected dimension","feed_subtitle":"Extends the self-similar dimension formula to all real-analytic contractions without a common fixed point.","key_machinery":"The key machinery is the reduction of convolutions $\\nu.\\mu$, where $\\nu$ is a measure on the infinite-dimensional space $A(I)$ of real-analytic maps, to convolutions $\\nu'.\\mu$ with $\\nu'$ supported on the finite-dimensional space $P_k$ of polynomials of degree at most $k$. Taylor's theorem with Lagrange remainder, together with the bounded-distortion estimate (Lemma 2.1) bounding all higher derivatives of iterates $\\varphi_u$ in terms of $|\\varphi'_u(0)|$, makes the error uniform and exponentially small. On $P_k$, an entropy-increase theorem (Theorem 3.1) is proved along the lines of earlier work on convolutions of self-similar and self-affine measures with measures on affine maps: it uses the inverse theorem for entropy from [7] and a uniform-entropy-dimension statement for $C^1$-images of $\\mu$. The contradiction argument also uses the conditional-entropy identity from [6], namely $\\dim\\mu = (H(p) - \\Delta')/\\chi$, which converts a dimension drop into the positivity of the conditional entropy $\\Delta'$.","core_discovery":"The central claim is Theorem 1.2: for an analytic IFS $\\Phi$ on $I=[0,1]$ with $0<|\\varphi_i'(x)|<1$, no common fixed point, and exponential separation, every positive probability vector $p$ produces a self-conformal measure $\\mu$ with $\\dim\\mu = \\min\\{1, H(p)/\\chi(\\Phi,p)\\}$. The proof works by contradiction: assume a dimension deficit $\\Delta = \\min\\{1,H/\\chi\\} - \\dim\\mu > 0$, then decompose the semigroup according to a cut set and Taylor-expand the maps to order $k$. The analyticity and bounded distortion give uniform control on higher derivatives, so convolutions with measures on real-analytic maps are well approximated by convolutions with measures on polynomials of degree at most $k$. For those polynomial convolutions, an entropy-increase theorem (Theorem 3.1) shows that the entropy of the convolved measure exceeds that of $\\mu$ by a fixed amount, which contradicts the assumed dimension drop. The no-common-fixed-point condition guarantees $\\mu$ is non-atomic and the attractor is not a point, which is necessary for the formula.","pith_inferences":["The Taylor-polynomial reduction suggests a general template: any entropy-increase statement for finite-dimensional parameter spaces can lift to infinite-dimensional analytic families whenever all higher derivatives are controlled by the first derivative; this could extend to random walks on diffeomorphism groups.","The analyticity assumption may be close to necessary: the proof fails for $C^\\infty$ maps, and a $C^\\infty$ exponentially separated system with unbounded derivative growth would be a natural test of how far the theorem can stretch.","The paper leaves open whether exponential separation is equivalent to freeness for polynomial IFSs with algebraic coefficients, because degrees of iterates grow exponentially; a proof or counterexample for quadratic maps would sharpen the exact-overlaps picture in the analytic case.","Because the bottleneck is the imported conditional-entropy identity, a promising route to non-analytic conformal IFSs is to prove $\\dim\\mu=(H(p)-\\Delta')/\\chi$ afresh under exponential separation alone."],"forward_implications":["Theorem 1.2 gives the dimension formula for every exponentially separated real-analytic IFS without a common fixed point, going beyond any finite-dimensional acting group.","Corollary 1.3 computes the Hausdorff dimension of the self-conformal set as $\\min\\{1, s(\\Phi)\\}$, where $s(\\Phi)$ is the conformal similarity dimension.","Corollary 1.4 shows that in a non-degenerate one-parameter analytic family of IFSs, all but a Hausdorff-dimension-zero set of parameters satisfy both the measure and set dimension formulas.","Whenever the dimension formula fails for an exponentially separated analytic IFS, the proof forces strong restrictions on the system and ties the size of the failure to the conditional entropy $\\Delta'$.","The author states that the proof strategy is expected to transfer other recent results on stationary fractal measures to the real-analytic setting."],"supporting_citations":[{"why":"Supplies the inverse theorem for entropy and the exponential-separation framework used in the entropy-increase step.","marker":"[7]"},{"why":"Supplies the identity dim μ=(H(p)−Δ′)/χ and exact dimensionality, imported to turn a dimension drop into positivity of Δ′.","marker":"[6]"},{"why":"Introduces the convolution method for entropy increase with measures on affine maps, adapted here to polynomial maps.","marker":"[1]"},{"why":"Establishes existence and uniqueness of the self-conformal set and measure used throughout.","marker":"[12]"},{"why":"Provides the relationship between asymptotic entropy rate and exact dimension used to identify dim μ.","marker":"[19]"},{"why":"Used in the parametric-family corollary to show the non-exponentially-separated parameters form a zero-dimensional exceptional set.","marker":"[16]"}],"fun_headline_variants":["Analytic IFS dimension equals entropy over Lyapunov","Self-conformal measures: exact dimension for analytic IFS","No common fixed point? Analytic IFS still exact","Dimension formula for analytic IFS proven"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire contradiction argument collapses if the imported identity dim μ=(H(p)−Δ′)/χ fails for some exponentially separated analytic IFS without a common fixed point, and the paper does not prove this identity from scratch.","fun_headline_variants_meta":{"raw":{"variants":["Analytic IFS dimension equals entropy over Lyapunov","Self-conformal measures: exact dimension for analytic IFS","No common fixed point? Analytic IFS still exact","Dimension formula for analytic IFS proven"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000471,"raw_usage":{"total_tokens":2405,"prompt_tokens":1069,"completion_tokens":1336,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":685,"completion_tokens_details":{"reasoning_tokens":1274}},"tokens_in":685,"tokens_out":1336,"duration_ms":8467,"temperature":1.0,"reasoning_tokens":1274,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T10:16:18.275952+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"The claim would be refuted by exhibiting an exponentially separated real-analytic IFS on [0,1] with no common fixed point, a positive probability vector p, and a self-conformal measure μ with dim μ < min{1, H(p)/χ}. A concrete check is whether the identity from [6, Theorem 2.8] holds for any such system; if not, that is the point to test. A numerical computation of dim μ for a polynomial IFS at algebraic parameters would be a practical test.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the inverse theorem for entropy and the exponential-separation framework used in the entropy-increase step."},{"cited_title":"Feng and H","cited_arxiv_id":null,"evidence_quote":"Supplies the identity dim μ=(H(p)−Δ′)/χ and exact dimensionality, imported to turn a dimension drop into positivity of Δ′."},{"cited_title":"Bárány, M","cited_arxiv_id":null,"evidence_quote":"Introduces the convolution method for entropy increase with measures on affine maps, adapted here to polynomial maps."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the relationship between asymptotic entropy rate and exact dimension used to identify dim μ."},{"cited_title":"Solomyak and Y","cited_arxiv_id":null,"evidence_quote":"Used in the parametric-family corollary to show the non-exponentially-separated parameters form a zero-dimensional exceptional set."}],"review_version":1}