OctoInformBeta

OpenAI preprint, September 2026 · §2. Recursive colors and finite averages

Proposition 2.3: Positive displacement gives an explicit lower bound for relative boundaries in Thompson’s group FF

A fixed finite set forces boundary expansion.

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

An interval homeomorphism g:[0,1]→[0,1]g:[0,1]\to[0,1] is a homeomorphism of the closed unit interval onto itself.✦ A real number xx is dyadic if it has the form k/2nk/2^n, with k∈ℤk\in\mathbb Z and n≥0n\geq0 an integer.✦ A dyadic affine decomposition of an interval homeomorphism gg consists of m≥1m\geq1 closed cells [ti,ti+1][t_i,t_{i+1}] covering [0,1][0,1], with dyadic knots 0=t0<⋯<tm=10=t_0<\cdots<t_m=1 and integer exponents eie_i, on which (1) holds: g(x)=g(ti)+2ei(x−ti)(ti≤x≤ti+1,0≤i<m).g(x)=g(t_i)+2^{e_i}(x-t_i) \qquad(t_i\leq x\leq t_{i+1},\quad 0\leq i<m).(1)✦ An interval homeomorphism gg is dyadic piecewise linear if it admits a dyadic affine decomposition. Thompson’s group FF consists of the strictly increasing interval homeomorphisms with dyadic piecewise linearity, with multiplication hg=h∘ghg=h\circ g.✦

For a finite subset AA of FF, its left translate by h∈Fh\in F is hA={hg:g∈A}hA=\{hg:g\in A\}.✦ The relative boundary rA(h)r_A(h) of a nonempty finite subset AA of FF under h∈Fh\in F is the ratio in (2); for a fixed nonempty finite transport set S⊂FS\subset F, write η\eta for the maximum in the same equation: rA(h)=|hA△A||A|,η=maxh∈SrA(h).r_A(h)=\frac{|hA\triangle A|}{|A|}, \qquad \eta=\max_{h\in S}r_A(h).(2)✦ Here △\triangle denotes symmetric difference. The boundary estimate concerns how small these relative boundaries can be for one finite set AA and all transports in SS.

The closed unit ball BEB_E of a real Hilbert space EE is the set in (3): BE={x∈E:∥x∥≤1}.B_E=\{x\in E:\|x\|\leq1\}.(3) A map f:BE→BEf:B_E\to B_E has Lipschitz bound LL and displacement at least δ\delta when it satisfies (4) for all x,y∈BEx,y\in B_E: ∥f(x)−f(y)∥≤L∥x−y∥,δ≤∥f(x)−x∥.\|f(x)-f(y)\|\leq L\|x-y\|, \qquad \delta\leq\|f(x)-x\|.(4)✦

The Analytic Constant

The concrete Hilbert space is H=L2([0,1];ℝ2)H=L^2([0,1];\mathbb R^2), where ℝ2\mathbb R^2 is the Euclidean plane with coordinates indexed by 00 and 11. The measure is Lebesgue measure restricted to [0,1][0,1]; square-integrable functions equal almost everywhere are identified, and HH has its usual L2L^2 norm. The circle parametrization p:ℝ→ℝ2p:\mathbb R\to\mathbb R^2, its tangent p1:ℝ→ℝ2p_1:\mathbb R\to\mathbb R^2, the oscillator W:ℝ→HW:\mathbb R\to H and its velocity W1:ℝ→HW_1:\mathbb R\to H are the functions given by the representatives in (5), for z∈ℝz\in\mathbb R and w∈[0,1]w\in[0,1]: p(z)=(cos⁡z,sin⁡z),p1(z)=(−sin⁡z,cos⁡z),W(z)(w)=p(zw),W1(z)(w)=wp1(zw).\begin{aligned} p(z)&=(\cos z,\sin z), &p_1(z)&=(-\sin z,\cos z),\\ W(z)(w)&=p(zw), &W_1(z)(w)&=w\,p_1(zw). \end{aligned}(5)

The radial profile a:ℝ→ℝa:\mathbb R\to\mathbb R and angular profile θ:ℝ→ℝ\theta:\mathbb R\to\mathbb R, with their scalar velocity functions a′a' and θ′\theta', are given for real tt by (6) and (7): a(t)={t/8,t≤2,5/16−(3−t)2/16,2<t<3,5/16,t≥3,a′(t)={1/8,t≤2,(3−t)/8,2<t<3,0,t≥3.\begin{aligned} a(t)&=\begin{cases} t/8,&t\leq2,\\ 5/16-(3-t)^2/16,&2<t<3,\\ 5/16,&t\geq3, \end{cases} & a'(t)&=\begin{cases} 1/8,&t\leq2,\\ (3-t)/8,&2<t<3,\\ 0,&t\geq3. \end{cases} \end{aligned}(6) θ(t)={0,t≤1,(t−1)2/2,1<t<2,t−3/2,t≥2,θ′(t)={0,t≤1,t−1,1<t<2,1,t≥2.\begin{aligned} \theta(t)&=\begin{cases} 0,&t\leq1,\\ (t-1)^2/2,&1<t<2,\\ t-3/2,&t\geq2, \end{cases} & \theta'(t)&=\begin{cases} 0,&t\leq1,\\ t-1,&1<t<2,\\ 1,&t\geq2. \end{cases} \end{aligned}(7) The radial direction vv, oscillating curve γ\gamma, curve velocity γ′\gamma' and normalized tangent uu are the functions from ℝ\mathbb R to HH in (8): v(t)=W(θ(t)),γ(t)=a(t)v(t),γ′(t)=a′(t)v(t)+a(t)θ′(t)W1(θ(t)),u(t)=∥γ′(t)∥−1γ′(t).\begin{aligned} v(t)&=W(\theta(t)), &\gamma(t)&=a(t)v(t),\\ \gamma'(t)&=a'(t)v(t)+a(t)\theta'(t)W_1(\theta(t)), &u(t)&=\|\gamma'(t)\|^{-1}\gamma'(t). \end{aligned}(8)

A regular separated curve in a real Hilbert space EE is a differentiable map c:ℝ→Ec:\mathbb R\to E, with velocity V=c′V=c' and a tangent U:ℝ→EU:\mathbb R\to E, satisfying (9) for all real s,ts,t and some separation constant σC>0\sigma_C>0: ∥U(t)∥=1,V(t)=∥V(t)∥U(t),∥V(t)∥≥18,∥V(t)−V(s)∥≤|t−s|,∥U(t)−U(s)∥≤16|t−s|,∥c(t)−c(s)∥≥σCwhen |t−s|≥132.\begin{gathered} \|U(t)\|=1,\qquad V(t)=\|V(t)\|U(t),\qquad \|V(t)\|\geq\tfrac18,\\ \|V(t)-V(s)\|\leq|t-s|,\qquad \|U(t)-U(s)\|\leq16|t-s|,\\ \|c(t)-c(s)\|\geq\sigma_C \quad\text{when }|t-s|\geq\tfrac1{32}. \end{gathered}(9)✦ The admissible tube radius ρC\rho_C of a regular separated curve cc with separation constant σC\sigma_C is the real number in (10). For the oscillating curve, write σ\sigma for the chosen separation constant at scale 1/321/32 and ρ\rho for its tube radius: ρC=min⁡{1/1024,σC/8},ρ=min⁡{1/1024,σ/8}.\rho_C=\min\{1/1024,\sigma_C/8\}, \qquad \rho=\min\{1/1024,\sigma/8\}.(10)

For vectors x,yx,y in a real Hilbert space EE, the skew operator Kx,yK_{x,y} and rotation operator ℛx,y\mathcal R_{x,y} are defined by (11) for e∈Ee\in E, where II is the identity operator: Kx,y(e)=⟨x,e⟩y−⟨y,e⟩x,ℛx,y=I+Kx,y+(1+⟨x,y⟩)−1Kx,y,2.\begin{aligned} K_{x,y}(e)&=\langle x,e\rangle y-\langle y,e\rangle x,\\ \mathcal R_{x,y}&=I+K_{x,y} +(1+\langle x,y\rangle)^{-1}K_{x,y}^{,2}. \end{aligned}(11) The tangent rotation RR for γ\gamma is a family of operators on HH, and its radial correction q:ℝ→ℝq:\mathbb R\to\mathbb R is a scalar function, given for real tt by (12): R(t)=ℛu(t),v(t),q(t)=−max⁡{a(t)+1/8,0}.R(t)=\mathcal R_{u(t),v(t)}, \qquad q(t)=-\max\{a(t)+1/8,0\}.(12)

Radial correction data for a regular separated curve cc with tangent UU consist of scalar functions a,q:ℝ→ℝa,q:\mathbb R\to\mathbb R, a function v:ℝ→Ev:\mathbb R\to E and bounded linear operators R(t):E→ER(t):E\to E satisfying (13) for all real s,ts,t and e∈Ee\in E: c(t)=a(t)v(t),a(0)=0,a(t)≤516,a(t)≥18(t>1),q(t)=min⁡{a(t),−18}−a(t),|q(t)|≤716,|q(t)−q(s)|≤18|t−s|,∥v(t)∥=1,∥v(t)−v(s)∥≤|t−s|,∥c(t)−c(s)∥≤716|t−s|,∥R(t)e∥=∥e∥,R(t)e⟂v(t)(e⟂U(t)),∥(R(t)−R(s))e∥≤238|t−s|∥e∥,R(t)=I(t≤1),U(t)=v(t)(t≤1).\begin{gathered} c(t)=a(t)v(t),\qquad a(0)=0,\qquad a(t)\leq\tfrac5{16},\\ a(t)\geq\tfrac18\quad(t>1),\qquad q(t)=\min\{a(t),-\tfrac18\}-a(t),\\ |q(t)|\leq\tfrac7{16},\qquad |q(t)-q(s)|\leq\tfrac18|t-s|,\\ \|v(t)\|=1,\qquad \|v(t)-v(s)\|\leq|t-s|,\\ \|c(t)-c(s)\|\leq\tfrac7{16}|t-s|,\qquad \|R(t)e\|=\|e\|,\\ R(t)e\perp v(t)\quad(e\perp U(t)),\\ \|(R(t)-R(s))e\|\leq238|t-s|\|e\|,\\ R(t)=I\quad(t\leq1),\qquad U(t)=v(t)\quad(t\leq1). \end{gathered}(13)✦ For radial correction data with tube radius ρC\rho_C, the correction Lipschitz constant is κ(ρC)\kappa(\rho_C), where κ\kappa is the function of a real variable rr in (14); the displacement Lipschitz constant L*L_* uses the oscillating curve’s radius ρ\rho in the same equation: κ(r)=max⁡{1,30+74r+3808r,5+78r},L*=2κ(ρ)ρ/4.\begin{aligned} \kappa(r)&=\max\left\{1, 30+\frac7{4r}+3808r,\ 5+\frac7{8r}\right\},\\ L_*&=\frac{2\kappa(\rho)}{\rho/4}. \end{aligned}(14)

The Claim

For real L,δL,\delta and a natural number DD, the boundary constant β(L,δ,D)\beta(L,\delta,D) is defined by (15), assigning zero to a quotient whose denominator is zero: β(L,δ,D)=δ2/L2−4/D4(1−1/D).\beta(L,\delta,D)= \frac{\delta^2/L^2-4/D}{4(1-1/D)}.(15)✦

Theorem 1 (Proposition 2.3). Let EE be a real Hilbert space and let f:BE→BEf:B_E\to B_E satisfy (4) with L>0L>0 and δ>0\delta>0. Let D≥2D\geq2 be an integer with 4L2/δ2<D4L^2/\delta^2<D. There is a nonempty finite set S⊂FS\subset F such that, for every nonempty finite A⊂FA\subset F, some h∈Sh\in S satisfies (16): 0<β(L,δ,D)≤rA(h).0<\beta(L,\delta,D)\leq r_A(h).(16)✦

For the constant L*L_* in (14), there are an integer DD and a nonempty finite set S⊂FS\subset F such that, for every nonempty finite A⊂FA\subset F, some h∈Sh\in S satisfies (17): D≥2,4L*2(1/2)2<D,0<β(L*,1/2,D)≤rA(h).D\geq2,\qquad \frac{4L_*^2}{(1/2)^2}<D,\qquad 0<\beta(L_*,1/2,D)\leq r_A(h).(17)✦ For every natural number DD, this specialized constant has the formula (18): β(L*,1/2,D)=(1/2)2/L*2−4/D4(1−1/D).\beta(L_*,1/2,D)= \frac{(1/2)^2/L_*^2-4/D}{4(1-1/D)}.(18)✦

There are a nonempty finite set S⊂FS\subset F and a real number bb such that, for every nonempty finite A⊂FA\subset F, some h∈Sh\in S satisfies (19): 0<b,b|A|≤|hA△A|.0<b,\qquad b|A|\leq|hA\triangle A|.(19)✦

In the paper, the result reads: with E,BE,f,L,δ,DE,B_E,f,L,\delta,D and SS fixed as in its construction, together with the ordered list of internal parents chosen in that construction, every nonempty finite A⊂FA\subset F satisfies (20): η≥δ2/L2−4/D4(1−1/D)>0.\eta\geq\frac{\delta^2/L^2-4/D}{4(1-1/D)}>0.(20)

What Is Checked

The boundary statements concern finite subsets of the dyadic piecewise linear group FF; the general statement assumes a map of a real Hilbert space’s closed unit ball satisfying (4), and the concrete statements use the constant in (14).

Both oscillator representatives in (5) are square-integrable for every real zz. The curve γ\gamma, with velocity γ′\gamma' and tangent uu, satisfies the regular separated curve conditions in (9) with σ>0\sigma>0.✦ The functions a,v,qa,v,q and the tangent rotations RR satisfy the radial correction conditions in (13) for this curve.✦