OctoInformBeta

OpenAI preprint, September 2026 · §4. The critical block tree · §4

§4.1 (setting): corners, blocks and generating functions: Generating series for typed words and Fortuin-Kasteleyn inventory streams, the one-link core word, and block offspring weights

Corner substitution decomposes nonempty stack-emptying typed words.

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

Take a Fortuin-Kasteleyn (FK) planar map with parameter p∈(1/2,1)p \in (1/2, 1), encoded as a burger-order inventory word. Removing the submaps inserted at the corners of its rooted block leaves a connected core word, so a nonempty stack-emptying word decomposes into a connected core word together with a stack-emptying subword at each corner, while word counts and block branching are tracked by generating series and offspring probability weights.

Inventory Words and the Stream Measure

Fix two boolean types, labeled 11 (the first type) and 22 (the second type). The set 𝒪\mathcal{O} of inventory orders consists of the fixed orders 𝖮τ\mathsf{O}_\tau for τ∈{1,2}\tau \in \{1, 2\} and the flexible order 𝖮*\mathsf{O}_*.✦ The burger-order alphabet 𝒜\mathcal{A} is the set {𝖡1,𝖡2,𝖮1,𝖮2,𝖮*}\{\mathsf{B}_1, \mathsf{B}_2, \mathsf{O}_1, \mathsf{O}_2, \mathsf{O}_*\}, where each letter in 𝒜\mathcal{A} is either a burger token 𝖡τ\mathsf{B}_\tau (τ∈{1,2}\tau \in \{1, 2\}) or an order token carrying an inventory order in 𝒪\mathcal{O}. An inventory stack evaluation state e=(O,S)e = (O, S) is a pair consisting of a finite list OO of pending orders in 𝒪\mathcal{O} and a finite list SS of boolean burger types stored freshest-first.✦

Given a word ww of letters in 𝒜\mathcal{A} and an evaluation state ee, the word evaluation eval(w,e)\mathrm{eval}(w, e) is the evaluation state returning ee when ww is empty, and on a letter aa followed by a suffix w′w' updating ee by the single-step transition on aa (in which an order consumes the freshest eligible burger) and continuing on w′w'.✦ For a word ww of letters in 𝒜\mathcal{A}, the word reduction red(w)\mathrm{red}(w) is the evaluation state eval(w,([],[]))\mathrm{eval}(w, ([], [])) obtained from empty pending orders and an empty burger stack.✦ A word ww of letters in 𝒜\mathcal{A} is empty-reducing if red(w)=([],[])\mathrm{red}(w) = ([], []).✦

For a parameter p∈(1/2,1)p \in (1/2, 1), the letter mass function mp:𝒜→ℝm_p \colon \mathcal{A} \to \mathbb{R} is the function assigning mass 1/41/4 to each burger 𝖡1,𝖡2\mathsf{B}_1, \mathsf{B}_2, mass p/2p/2 to the flexible order 𝖮*\mathsf{O}_*, and mass (1−p)/4(1 - p)/4 to each fixed order 𝖮1,𝖮2\mathsf{O}_1, \mathsf{O}_2. The letter distribution νp\nu_p is the probability mass function on 𝒜\mathcal{A} induced by mpm_p.✦ For any index set ι\iota, the independent and identically distributed (i.i.d.) inventory stream measure νp⊗ι\nu_p^{\otimes \iota} on 𝒜ι\mathcal{A}^{\iota} is the product measure of νp\nu_p across coordinates; on 𝒜ℕ\mathcal{A}^{\mathbb{N}}, write ℙp=νp⊗ℕ\mathbb{P}_p = \nu_p^{\otimes \mathbb{N}}.

For a stream x=(x0,x1,…)∈𝒜ℕx = (x_0, x_1, \dots) \in \mathcal{A}^{\mathbb{N}} and n≥0n \ge 0, the letter stream initial word wn(x)w_n(x) is the word (x0,…,xn−1)(x_0, \dots, x_{n-1}) of length nn.✦ For length N≥0N \ge 0 and parameter p∈(1/2,1)p \in (1/2, 1), the empty-reducing word probability εp(N)\varepsilon_p(N) is the probability ℙp({x∈𝒜ℕ:wN(x) is empty-reducing})\mathbb{P}_p\bigl(\{x \in \mathcal{A}^{\mathbb{N}} : w_N(x) \text{ is empty-reducing}\}\bigr).

Typed Words and Monomial Weights

The set 𝒯={𝖡1,𝖡2,𝖮1,𝖮2}\mathcal{T} = \{\mathsf{B}_1, \mathsf{B}_2, \mathsf{O}_1, \mathsf{O}_2\} of typed burger-order letters consists of the burger tokens 𝖡τ\mathsf{B}_\tau and the order tokens 𝖮τ\mathsf{O}_\tau carrying an explicit boolean type τ∈{1,2}\tau \in \{1, 2\}.✦ For a word ww of typed letters and an initial stack ss of boolean types, typed word stack execution is the process that evaluates ww letter by letter from left to right against ss, returning either the resulting stack or failure upon an illegal closure.✦ The set 𝒲0\mathcal{W}_0 of stack-emptying typed words consists of all typed words ww such that typed word stack execution of ww from the empty stack terminates successfully with the empty stack✦, and the set 𝒲+\mathcal{W}_+ of nonempty stack-emptying typed words consists of all words in 𝒲0\mathcal{W}_0 of positive length.✦

For a typed word ww and an initial stack ss of boolean types, the flexible closing count f(w,s)f(w, s) is the number of flexible order closures encountered when processing ww step by step against ss (stopping at any illegal closure).✦ Given elements t,r∈[0,∞)t, r \in [0, \infty) and a typed word ww, the inventory word monomial weight wtt,r(w)\mathrm{wt}_{t, r}(w) is tf(w,[])r|w|t^{f(w, [])} r^{|w|}, where |w||w| is the length of ww and f(w,[])f(w, []) is the flexible closing count starting from the empty stack.

Matchings, Core Words and Corner Insertion

Let α\alpha be a set. A non-crossing arch matching on a finite list ll of items in α\alpha with an arch list EE of ordered pairs in α×α\alpha \times \alpha is generated inductively by three rules: the empty list matches the empty arch list; an unmatched leading item may be skipped; and two items u,vu, v enclosing an interior list AA and followed by an exterior list BB may be paired as (u,v)(u, v) together with non-crossing arch matchings EAE_A on AA and EBE_B on BB, giving the arch list that places (u,v)(u, v) before EAE_A followed by EBE_B.✦ For an arch list E=[(a1,b1),…,(am,bm)]E = [(a_1, b_1), \dots, (a_m, b_m)], the list of endpoints of paired arches is the flattened list [a1,b1,…,am,bm][a_1, b_1, \dots, a_m, b_m].✦ For an arch list EE on α×α\alpha \times \alpha, the arch transposition product turns(E)\mathrm{turns}(E) is the permutation of α\alpha that is the identity on the empty list and equals turns(E′)(uv)\mathrm{turns}(E')(u\; v) when an arch (u,v)(u, v) precedes E′E', where (uv)(u\; v) is the transposition swapping uu and vv.✦

Given a map v:α→𝒯v \colon \alpha \to \mathcal{T} and a finite list ll of positions in α\alpha, a two-stack matching certificate M=(E1,E2)M = (E_1, E_2) for vv along ll consists of two non-crossing arch matchings E1E_1 and E2E_2 on ll whose combined endpoints (from E1E_1 followed by E2E_2) form a permutation of ll, such that every arch (i,j)∈E1(i, j) \in E_1 has v(i)=𝖡1v(i) = \mathsf{B}_1 and v(j)=𝖮1v(j) = \mathsf{O}_1, while every arch (i,j)∈E2(i, j) \in E_2 has v(i)=𝖡2v(i) = \mathsf{B}_2 and v(j)=𝖮2v(j) = \mathsf{O}_2.✦ The stack matching mate permutation πM\pi_M of a two-stack matching certificate M=(E1,E2)M = (E_1, E_2) is the permutation on α\alpha obtained by applying turns\mathrm{turns} to the concatenation of E1E_1 and E2E_2.✦

For an integer N≥0N \ge 0 and a permutation π\pi on {0,…,N−1}\{0, \dots, N - 1\}, a chord opening endpoint of π\pi is an index i∈{0,…,N−1}i \in \{0, \dots, N - 1\} satisfying i<π(i)i < \pi(i).✦ For an integer N≥0N \ge 0 and a permutation π\pi on {0,…,N−1}\{0, \dots, N - 1\}, two chord opening endpoints xx and yy of π\pi satisfy the chord crossing relation if either x<y<π(x)<π(y)x < y < \pi(x) < \pi(y) or y<x<π(y)<π(x)y < x < \pi(y) < \pi(x). The chord intersection graph G(π)G(\pi) is the simple graph on the chord opening endpoints of π\pi whose edges are the pairs satisfying the chord crossing relation.✦ The set 𝒦\mathcal{K} of connected core words consists of all typed words ww for which there exist an integer K≥0K \ge 0, a map v:{0,…,K−1}→𝒯v \colon \{0, \dots, K - 1\} \to \mathcal{T} with (v(0),…,v(K−1))=w(v(0), \dots, v(K - 1)) = w, and a two-stack matching certificate MM for vv along the index list [0,…,K−1][0, \dots, K - 1] such that 00 is a chord opening endpoint of πM\pi_M and every chord opening endpoint of πM\pi_M is reachable from 00 in G(πM)G(\pi_M).

The set 𝒞\mathcal{C} of valid corner-word lists consists of all finite lists l=[(c1,u1),…,(cK,uK)]l = [(c_1, u_1), \dots, (c_K, u_K)] of pairs (ci,ui)(c_i, u_i) with ci∈𝒯c_i \in \mathcal{T} and typed words uiu_i satisfying (c1,…,cK)∈𝒦(c_1, \dots, c_K) \in \mathcal{K} and each ui∈𝒲0u_i \in \mathcal{W}_0.✦ The corner insertion map ι:𝒞→𝒲+\iota \colon \mathcal{C} \to \mathcal{W}_+ is the map sending each valid corner-word list to its literal corner insertion.

The Claim

Definition 1. For t,r∈[0,∞]t, r \in [0, \infty], the word generating series W(t,r)∈[0,∞]W(t, r) \in [0, \infty] is the sum (1) of inventory word monomial weights over all stack-emptying typed words: W(t,r)=∑w∈𝒲0wtt,r(w).W(t, r) = \sum_{w \in \mathcal{W}_0} \mathrm{wt}_{t, r}(w) .(1)✦ For an FK model parameter p∈(1/2,1)p \in (1/2, 1) and x∈ℝx \in \mathbb{R}, the inventory generating series Ip(x)∈ℝI_p(x) \in \mathbb{R} is the series (2) in the even-length empty-reducing word probabilities: Ip(x)=∑m=0∞εp(2m)xm.I_p(x) = \sum_{m=0}^{\infty} \varepsilon_p(2m)\, x^m .(2)✦ The sequence b=(bk)k≥0b = (b_k)_{k \ge 0} of block offspring probability weights is a sequence of real numbers satisfying (3): bk≥0(k≥0),∑k=0∞bk=1.b_k \ge 0 \quad (k \ge 0), \qquad \sum_{k=0}^{\infty} b_k = 1 .(3)✦

Theorem 1. The corner insertion map ι\iota is a bijection (4) (the corner substitution bijection): ι:𝒞→∼𝒲+.\iota \colon \mathcal{C} \xrightarrow{\sim} \mathcal{W}_+ .(4)✦ The one-link typed word [𝖡1,𝖮1][\mathsf{B}_1, \mathsf{O}_1] consisting of a burger and an order of the first type is a connected core word (5): [𝖡1,𝖮1]∈𝒦.[\mathsf{B}_1, \mathsf{O}_1] \in \mathcal{K} .(5)✦

What Is Checked

The definitions and theorem are stated over the burger-order alphabet with FK parameter p∈(1/2,1)p \in (1/2, 1) and its i.i.d. stream measure, typed words under stack execution, non-crossing arch matchings and chord intersection graphs, valid corner-word lists, and real sequences. Beyond the claims, the library proves that for p∈(1/2,1)p \in (1/2, 1) the letter masses are nonnegative and sum to one, and that the chord crossing relation is symmetric and irreflexive.