OctoInformBeta

OpenAI preprint, September 2026 · §6. Overlap conventions and dimension of the attractor

Corollary 6.4: Lower Hausdorff dimension from symbol entropy and a positive common contraction ratio

Dimension for homogeneous systems without exact overlaps.

Hover over a blue or green term in the text, or a node in the graph, to find it in the other. Select it to read its declaration both in Lean and in words.

The Setting

A homogeneous affine iterated function system on the line uses one contraction ratio for every symbol, while its translations and weights may vary. A weighted affine iterated function system SS on a finite nonempty alphabet II consists of similarities φi\varphi_i with real signed slopes rir_i, real translations tit_i, and weights pip_i satisfying (1): φi(x)=rix+ti,0<|ri|<1,pi>0,∑i∈Ipi=1.\varphi_i(x)=r_i x+t_i,\qquad 0<|r_i|<1,\qquad p_i>0,\qquad \sum_{i\in I}p_i=1.(1)✦

A measure μ\mu on ℝ\mathbb R is self-similar for SS when it satisfies (2), where (φi)*μ(\varphi_i)_*\mu denotes pushforward: μ=∑i∈Ipi(φi)*μ.\mu=\sum_{i\in I}p_i(\varphi_i)_*\mu.(2)✦

The lower Hausdorff dimension dim⁡Hμ\dim_H\mu of a measure μ\mu on ℝ\mathbb R is the extended nonnegative real number defined by positive-mass Borel sets in (3): dim⁡Hμ=inf⁡{dim⁡HE:E⊆ℝ is Borel and μ(E)>0}∈[0,∞].\dim_H\mu =\inf\{\dim_H E:E\subseteq\mathbb R\text{ is Borel and }\mu(E)>0\} \in[0,\infty].(3)✦

For a finite address ww over II, the word affine composition (rw,tw)(r_w,t_w) is the pair of signed slope and translation of the map φw(x)=rwx+tw\varphi_w(x)=r_wx+t_w, specified by (4); the empty address is denoted by ⌀\varnothing, and iwiw denotes prefixing ii to ww: (r⌀,t⌀)=(1,0),(riw,tiw)=(rirw,ritw+ti),φiw=φi∘φw.(r_{\varnothing},t_{\varnothing})=(1,0),\qquad (r_{iw},t_{iw})=(r_i r_w,r_i t_w+t_i),\qquad \varphi_{iw}=\varphi_i\circ\varphi_w.(4)✦ The system SS has no exact overlaps when the map w↦(rw,tw)w\mapsto(r_w,t_w) is injective on all finite addresses, including the empty address.✦

The symbol entropy H(p)H(p) of the weight vector p=(pi)i∈Ip=(p_i)_{i\in I} of SS is the natural-logarithmic Shannon entropy in (5): H(p)=−∑i∈Ipiln⁡pi.H(p)=-\sum_{i\in I}p_i\ln p_i.(5)✦

The Claim

Theorem 1. Let SS be a weighted affine iterated function system on a finite nonempty alphabet II. Suppose that all slopes equal a common number λ>0\lambda>0, that SS has no exact overlaps, and that μ\mu is a self-similar probability measure for SS. The homogeneous dimension formula is (6): dim⁡Hμ=min⁡{1,H(p)ln⁡(1/λ)}.\dim_H\mu=\min\left\{1,\frac{H(p)}{\ln(1/\lambda)}\right\}.(6)✦ Under the same hypotheses, the direct homogeneous dimension formula is (7): dim⁡Hμ=min⁡{1,H(p)ln⁡(1/λ)}.\dim_H\mu=\min\left\{1,\frac{H(p)}{\ln(1/\lambda)}\right\}.(7)✦ The real values on the right are regarded as elements of [0,∞][0,\infty].

In the paper, the result reads: if m≥2m\geq2, 0<λ<10<\lambda<1, and t1,…,tm∈ℝt_1,\ldots,t_m\in\mathbb R are such that the maps φi(x)=λx+ti\varphi_i(x)=\lambda x+t_i have no exact overlaps, then for every strictly positive probability vector pp their self-similar measure μ\mu satisfies dim⁡Hμ=min⁡{1,H2(p)/log⁡2(1/λ)}\dim_H\mu=\min\{1,H_2(p)/\log_2(1/\lambda)\}, where H2(p)=−∑i=1mpilog⁡2piH_2(p)=-\sum_{i=1}^m p_i\log_2 p_i.

What Is Checked

Both formulas are stated over finite nonempty weighted affine systems with a common positive contraction ratio, a self-similar probability measure, and injectivity of the complete word affine map on all finite addresses.