OctoInformBeta

OpenAI preprint, September 2026 · §5. Conditioned trees and uniform path sums · §5

Proposition 5.1 [Joint coding limit]: Rescaled odd-size joint coding paths and Fortuin-Kasteleyn tree heights converge to Brownian excursion limits, matching the Vervaat bridge law

Joint Brownian excursion limits for critical even tree codes.

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 critical Galton-Watson tree with finite offspring variance σ2\sigma^2 and even offspring degrees, conditioned for n≥0n \ge 0 to have N=2n+1N = 2n+1 vertices. Its Lukasiewicz exploration walk WjW_j and vertex height process HjH_j trace the tree in preorder, posing the question of what joint scaling limit their rescaled piecewise linear interpolations approach as nn grows.

In the continuum, shifting a Brownian bridge at its first global minimum via the Vervaat transform invites comparison with a standard normalized Brownian excursion ee on [0,1][0,1]. Specializing the offspring law to the block weights of the Fortuin-Kasteleyn (FK) burger-order inventory model with parameter p∈(1/2,1)p \in (1/2, 1) then asks whether the rescaled height path of a marked FK tree sample approaches the same Brownian excursion limit.

Tree Codes, Offspring Laws, and Path Interpolations

For a sequence d:{0,…,N−1}→{0,…,N−1}d : \{0,\dots,N-1\} \to \{0,\dots,N-1\} of length N≥0N \ge 0, the degree sequence partial walk (Wj(d))j≥0(W_j(d))_{j \ge 0} of dd is the integer sequence Wj(d)=∑i=0N−1(di−1)𝟏i<j.W_j(d) = \sum_{i=0}^{N-1} (d_i - 1)\,\mathbf{1}_{i < j}.(1)✦ For any sequence d:{0,…,N−1}→ℕd : \{0,\dots,N-1\} \to \mathbb{N} of length N≥0N \ge 0, the total step Δ(d)\Delta(d) of dd is the integer ∑i=0N−1(di−1)\sum_{i=0}^{N-1} (d_i - 1). A discrete bridge sequence of length NN is a sequence d:{0,…,N−1}→{0,…,N−1}d : \{0,\dots,N-1\} \to \{0,\dots,N-1\} satisfying WN(d)=−1W_N(d) = -1 in (1).✦ A sequence d:{0,…,N−1}→{0,…,N−1}d : \{0,\dots,N-1\} \to \{0,\dots,N-1\} satisfies the valid tree code condition if WN(d)=−1W_N(d) = -1 and Wj(d)≥0W_j(d) \ge 0 for all 0≤j<N0 \le j < N✦; the Lukasiewicz tree code set 𝒞N\mathcal{C}_N consists of all sequences of length NN satisfying this condition.

Given a positive integer NN and an offset m∈{0,…,N−1}m \in \{0,\dots,N-1\}, the cyclic rotation of degree array ρm\rho_m is the permutation of {0,…,N−1}N\{0,\dots,N-1\}^N sending a sequence dd to (di+mmod⁡N)i=0N−1(d_{i+m \bmod N})_{i=0}^{N-1}.✦ Its restriction to discrete bridge sequences of length NN defines the cyclic rotation of bridges ρm\rho_m.✦ For a discrete bridge sequence dd of positive length NN, the good rotation index m*(d)∈{0,…,N−1}m^*(d) \in \{0,\dots,N-1\} is the unique shift index whose cyclic rotation turns dd into a valid tree code✦, and the canonical tree code of dd is the Lukasiewicz tree code obtained by applying that shift.✦ For n≥0n \ge 0, the tree code from bridge map is the map sending each integer sequence d:{0,…,2n}→ℕd : \{0,\dots,2n\} \to \mathbb{N} to the Lukasiewicz tree code in 𝒞2n+1\mathcal{C}_{2n+1} obtained by applying its good rotation shift.

A critical even offspring distribution ν\nu is a probability mass function on ℕ\mathbb{N} (its offspring probability mass function) such that ν(k)=0\nu(k) = 0 for every odd kk, ν(0)>0\nu(0) > 0, ν(2)>0\nu(2) > 0, the mean is ∑k=0∞kν(k)=1\sum_{k=0}^\infty k\,\nu(k) = 1, and the second moment ∑k=0∞k2ν(k)\sum_{k=0}^\infty k^2 \nu(k) is finite.✦ For a critical even offspring distribution ν\nu, the offspring distribution variance σ2\sigma^2 is the nonnegative real number (2), and the offspring standard deviation σ\sigma is the nonnegative real number σ2\sqrt{\sigma^2}. σ2=∑k=0∞(k−1)2ν(k).\sigma^2 = \sum_{k=0}^\infty (k-1)^2 \nu(k).(2) A sequence of block offspring probability weights b=(bm)m≥0b = (b_m)_{m \ge 0} is a nonnegative real sequence satisfying ∑m=0∞bm=1\sum_{m=0}^\infty b_m = 1. The doubled block mass sequence b̃\tilde{b} of a sequence b=(bm)m≥0b = (b_m)_{m \ge 0} is the sequence (b̃k)k≥0(\tilde{b}_k)_{k \ge 0} defined by b̃k=bk/2\tilde{b}_k = b_{k/2} for even kk and b̃k=0\tilde{b}_k = 0 for odd kk. For block offspring probability weights bb satisfying b0>0b_0 > 0, b1>0b_1 > 0, ∑m=0∞mbm=1/2\sum_{m=0}^\infty m b_m = 1/2, and ∑m=0∞m2bm<∞\sum_{m=0}^\infty m^2 b_m < \infty, the constructed even offspring distribution is the critical even offspring distribution with mass sequence b̃\tilde{b}.✦

For a critical even offspring distribution ν\nu and length N≥0N \ge 0, the independent and identically distributed (i.i.d.) offspring product measure on {0,…,N−1}→ℕ\{0,\dots,N-1\} \to \mathbb{N} is the NN-fold product of ν\nu.✦ For n≥0n \ge 0, conditioning the i.i.d. offspring product measure of length 2n+12n+1 on total step Δ(d)=−1\Delta(d) = -1 gives the conditioned offspring bridge law✦, and pushing this law forward under the tree code from bridge map yields the conditioned tree code law on 𝒞2n+1\mathcal{C}_{2n+1}.✦ Given a family of measures (κk)k≥0(\kappa_k)_{k \ge 0} on a measurable space AA, the conditional mark measure on code for d∈𝒞Nd \in \mathcal{C}_N is the product measure ⨂i=0N−1κdi\bigotimes_{i=0}^{N-1} \kappa_{d_i} on ANA^N, forming the tree code marking kernel from 𝒞N\mathcal{C}_N to ANA^N, and the joint distribution of the conditioned tree code law at size nn with this kernel on 𝒞2n+1\mathcal{C}_{2n+1} gives the marked tree code law on 𝒞2n+1×A2n+1\mathcal{C}_{2n+1} \times A^{2n+1}. For a critical even offspring distribution ν\nu and length N≥0N \ge 0, the tree code partition function ZN(ν)Z_N(\nu) is the nonnegative real number (3), and for each Lukasiewicz tree code d∈𝒞Nd \in \mathcal{C}_N the tree code weight wν(d)w_\nu(d) is the nonnegative real number (3). wν(d)=∏i=0N−1ν(di),ZN(ν)=∑d∈𝒞Nwν(d).w_\nu(d) = \prod_{i=0}^{N-1} \nu(d_i), \qquad Z_N(\nu) = \sum_{d \in \mathcal{C}_N} w_\nu(d).(3)

For a sequence d:{0,…,N−1}→{0,…,N−1}d : \{0,\dots,N-1\} \to \{0,\dots,N-1\} of length N≥0N \ge 0 and an index j≥0j \ge 0, the tree vertex height Hj(d)H_j(d) is the number of indices 0≤k<min⁡(j,N)0 \le k < \min(j, N) such that Wk(d)≤Wr(d)W_k(d) \le W_r(d) for every integer r∈{0,…,N}r \in \{0,\dots,N\} with k≤r≤jk \le r \le j, where Wr(d)W_r(d) is given by (1).✦ For N≥0N \ge 0 and a real sequence a=(ai)i=0Na = (a_i)_{i=0}^N, the piecewise linear interpolation IN(a)∈C([0,1],ℝ)I_N(a) \in C([0,1], \mathbb{R}) is defined using the interpolation hat functions ΛN,i(t)=max⁡(0,1−|Nt−i|)\Lambda_{N,i}(t) = \max(0, 1 - |Nt - i|) for t∈[0,1]t \in [0,1] by IN(a)(t)=∑i=0NaiΛN,i(t).I_N(a)(t) = \sum_{i=0}^N a_i \Lambda_{N,i}(t).(4) For a critical even offspring distribution ν\nu with standard deviation σ\sigma from (2) and a Lukasiewicz tree code d∈𝒞Nd \in \mathcal{C}_N of length N≥1N \ge 1, the normalized preorder height path hN(ν,d)∈C([0,1],ℝ)h_N(\nu, d) \in C([0,1], \mathbb{R}) and the joint tree code path JN(σ,d)∈C([0,1],ℝ2)J_N(\sigma, d) \in C([0,1], \mathbb{R}^2) are formed from (4) as hN(ν,d)=IN((σHi(d)2N)i=0N),JN(σ,d)=(IN((Wi(d)σN)i=0N),IN((σHi(d)2N)i=0N)).h_N(\nu, d) = I_N\!\left(\left(\frac{\sigma H_i(d)}{2\sqrt{N}}\right)_{i=0}^N\right), \qquad J_N(\sigma, d) = \left(I_N\!\left(\left(\frac{W_i(d)}{\sigma\sqrt{N}}\right)_{i=0}^N\right),\; I_N\!\left(\left(\frac{\sigma H_i(d)}{2\sqrt{N}}\right)_{i=0}^N\right)\right).(5) For a critical even offspring distribution ν\nu with standard deviation σ\sigma, any function F:C([0,1],ℝ2)→ℝF : C([0,1], \mathbb{R}^2) \to \mathbb{R}, and N≥0N \ge 0, the tree code test expectation ENν[F]E_N^\nu[F] of FF under ν\nu is the real number given using (3) and (5) by ENν[F]=∑d∈𝒞Nwν(d)ZN(ν)F(JN(σ,d)).E_N^\nu[F] = \sum_{d \in \mathcal{C}_N} \frac{w_\nu(d)}{Z_N(\nu)}\, F\bigl(J_N(\sigma, d)\bigr).(6)

Brownian Bridges, the Vervaat Transform, and Excursions

Given a real-valued stochastic process W=(Wt)t≥0W = (W_t)_{t \ge 0} on a probability space (Ω,ℙ)(\Omega, \mathbb{P}), its Brownian motion path W̃∈C([0,1],ℝ)\widetilde{W} \in C([0,1], \mathbb{R}) is the restriction t↦Wtt \mapsto W_t on [0,1][0,1], and its Brownian bridge path Wbr∈C([0,1],ℝ)W^{\mathrm{br}} \in C([0,1], \mathbb{R}) is given for t∈[0,1]t \in [0,1] by Wtbr=W̃t−tW1W^{\mathrm{br}}_t = \widetilde{W}_t - t W_1.✦

For any continuous path f∈C([0,1],ℝ)f \in C([0,1], \mathbb{R}), its first minimum time c(f)∈[0,1]c(f) \in [0,1] is the least time c∈[0,1]c \in [0,1] satisfying f(c)≤f(t)f(c) \le f(t) for all t∈[0,1]t \in [0,1].✦ Given c∈[0,1]c \in [0,1] and f∈C([0,1],ℝ)f \in C([0,1], \mathbb{R}), the cyclic path shift θcf∈C([0,1],ℝ)\theta_c f \in C([0,1], \mathbb{R}) is defined for t∈[0,1]t \in [0,1] by (θcf)(t)={f(c+t)−f(c),c+t≤1,f(1)+f(c+t−1)−f(0)−f(c),c+t>1.(\theta_c f)(t) = \begin{cases} f(c+t) - f(c), & c+t \le 1, \\ f(1) + f(c+t-1) - f(0) - f(c), & c+t > 1. \end{cases}(7)✦ The Vervaat transform V(f)V(f) of f∈C([0,1],ℝ)f \in C([0,1], \mathbb{R}) is the continuous path θc(f)f\theta_{c(f)} f defined by (7).

Given three real-valued processes B=(B(0),B(1),B(2))B = (B^{(0)}, B^{(1)}, B^{(2)}) on (Ω,ℙ)(\Omega, \mathbb{P}), the three-bridge excursion path sBs_B of BB is the continuous path on [0,1][0,1] defined by sB(t)=∑i=02(Bt(i)−tB1(i))2,t∈[0,1],s_B(t) = \sqrt{\sum_{i=0}^2 \bigl(B^{(i)}_t - t B^{(i)}_1\bigr)^2}, \qquad t \in [0,1],(8) and the standard Brownian excursion path e(B)∈C([0,1],ℝ)e(B) \in C([0,1], \mathbb{R}) is the continuous path equal to sBs_B from (8) when continuous and zero otherwise.✦ Associated with BB is the diagonal Brownian excursion path ediag(B)=(e(B),e(B))∈C([0,1],ℝ2)e_{\mathrm{diag}}(B) = (e(B), e(B)) \in C([0,1], \mathbb{R}^2), and for any function F:C([0,1],ℝ2)→ℝF : C([0,1], \mathbb{R}^2) \to \mathbb{R}, the joint Brownian test expectation 𝔼ℙ[F(ediag(B))]\mathbb{E}_{\mathbb{P}}\bigl[F(e_{\mathrm{diag}}(B))\bigr] is the real number 𝔼ℙ[F(ediag(B))]=∫ΩF(ediag(B)(ω))dℙ(ω).\mathbb{E}_{\mathbb{P}}\bigl[F(e_{\mathrm{diag}}(B))\bigr] = \int_{\Omega} F\bigl(e_{\mathrm{diag}}(B)(\omega)\bigr)\, d\mathbb{P}(\omega).(9)

The Burger-order Inventory Model and Fortuin-kasteleyn Heights

An FK model parameter is a real probability p∈(1/2,1)p \in (1/2, 1).✦ An inventory order is either a flexible order or a fixed order of one of two burger types in {0,1}\{0,1\}✦, and a burger-order alphabet letter is either a burger token of type b∈{0,1}b \in \{0,1\} or an inventory order, forming a five-letter alphabet 𝒜\mathcal{A}.✦ For p∈(1/2,1)p \in (1/2, 1), the letter mass function assigns probability 1/41/4 to each burger token, p/2p/2 to the flexible order, and (1−p)/4(1-p)/4 to each fixed order, inducing the letter distribution on 𝒜\mathcal{A}. The i.i.d. inventory stream measure ℙpst\mathbb{P}_p^{\mathrm{st}} is the infinite product measure of this letter distribution on 𝒜ℕ\mathcal{A}^{\mathbb{N}}✦, and for a stream x=(x0,x1,…)∈𝒜ℕx = (x_0, x_1, \dots) \in \mathcal{A}^{\mathbb{N}} and length n≥0n \ge 0, the letter stream initial word wn(x)w_n(x) of xx is the word (x0,…,xn−1)(x_0, \dots, x_{n-1}).

An inventory stack evaluation state is a pair consisting of a list of pending orders and a freshest-first stack of burger types in {0,1}\{0,1\}.✦ The evaluation of a word maps a finite word of letters in 𝒜\mathcal{A} and an evaluation state to the state obtained by updating sequentially via single-step transitions (where an order consumes the freshest eligible burger).✦ The word reduction of a word in 𝒜\mathcal{A} is its evaluation starting from empty pending orders and an empty burger stack✦, and a word satisfies the empty-reducing word condition if its word reduction leaves both lists empty.✦ Under p∈(1/2,1)p \in (1/2, 1), the empty-reducing word probability qN(p)q_N(p) is the probability under ℙpst\mathbb{P}_p^{\mathrm{st}} that wN(x)w_N(x) satisfies the empty-reducing word condition✦, and the total empty word probability τ\tau is the real number ∑n=0∞q2n(p)\sum_{n=0}^\infty q_{2n}(p). For p∈(1/2,1)p \in (1/2, 1), the flexible order parameter tpt_p and inventory radius parameter uu are the real numbers tp=1+2p1−p,u=1−p16.t_p = 1 + \frac{2p}{1-p}, \qquad u = \frac{1-p}{16}.(10)

A typed burger-order letter is either a burger token or an order token carrying an explicit type in {0,1}\{0,1\}; the resulting four-letter alphabet 𝒜typ\mathcal{A}_{\mathrm{typ}} consists of the tokens burgerb\mathrm{burger}_b and orderb\mathrm{order}_b for b∈{0,1}b \in \{0,1\}.✦ Given a stack of types in {0,1}\{0,1\}, the typed word stack execution processes a finite word ww of typed letters step by step and returns either the resulting stack or failure on an illegal closure✦; the empty-reducing typed word set ℰN\mathcal{E}_N consists of all typed words of length N≥0N \ge 0 whose execution starting from an empty stack succeeds and terminates with an empty stack.✦ For a typed word ww and an initial stack ss of types in {0,1}\{0,1\}, the flexible closing count f(w,s)f(w, s) counts the flexible order closures encountered before any illegal closure when processing ww from ss; writing f(w)=f(w,∅)f(w) = f(w, \emptyset), the inventory word monomial weight of ww at ϑ,r∈ℝ\vartheta, r \in \mathbb{R} is ϑf(w)r|w|\vartheta^{f(w)} r^{|w|}, where |w||w| is the length of ww.✦

For a finite sequence ll and a list EE of ordered pairs, EE forms a non-crossing arch matching on ll under three inductive rules: the empty pair list matches the empty sequence; EE matches al′a l' whenever EE matches l′l'; and (a,b)E′F′(a,b) E' F' matches aLbRa L b R whenever E′E' matches LL and F′F' matches RR.✦ The arch transposition product π(E)\pi(E) of a pair list EE is the permutation defined recursively by π(∅)=id\pi(\emptyset) = \mathrm{id} and π((a,b)E′)=π(E′)(ab)\pi((a,b)E') = \pi(E')(a\;b), where (ab)(a\;b) is the transposition of aa and bb.✦ For a set α\alpha, a two-stack matching certificate MM for a map v:α→𝒜typv : \alpha \to \mathcal{A}_{\mathrm{typ}} along a finite sequence ll in α\alpha consists of two non-crossing arch matchings E1E_1 and E2E_2 on ll whose combined endpoints form a permutation of ll, such that every pair (i,j)∈E1(i,j) \in E_1 satisfies v(i)=burger1v(i) = \mathrm{burger}_1 and v(j)=order1v(j) = \mathrm{order}_1 while every pair (i,j)∈E2(i,j) \in E_2 satisfies v(i)=burger0v(i) = \mathrm{burger}_0 and v(j)=order0v(j) = \mathrm{order}_0, with stack matching mate permutation π(M)\pi(M) defined as the permutation π(E1E2)\pi(E_1 E_2).

Given a positive integer KK and a permutation aa on {0,…,K−1}\{0,\dots,K-1\}, its chord opening endpoints are the indices i∈{0,…,K−1}i \in \{0,\dots,K-1\} with i<a(i)i < a(i), two openers xx and yy satisfy the chord crossing relation if either x<y<a(x)<a(y)x < y < a(x) < a(y) or y<x<a(y)<a(x)y < x < a(y) < a(x)✦, and the chord intersection graph of aa is the simple graph on its openers whose edges are pairs satisfying this crossing relation.✦ A typed word ww of length KK satisfies the connected core word condition if, writing v:{0,…,K−1}→𝒜typv : \{0,\dots,K-1\} \to \mathcal{A}_{\mathrm{typ}} for its sequence of letters (so w=(v(0),…,v(K−1))w = (v(0),\dots,v(K-1))), there exists a two-stack matching certificate MM for vv along (0,…,K−1)(0,\dots,K-1) such that 00 is a chord opening endpoint of π(M)\pi(M) from which every chord opening endpoint of π(M)\pi(M) is reachable in the chord intersection graph of π(M)\pi(M).✦ A connected typed core is a typed word satisfying the connected core word condition✦, and the optional core label set ℒ\mathcal{L} consists of an empty label ∘\circ with arity ar(∘)=0\mathrm{ar}(\circ) = 0 and all connected typed cores cc with arity ar(c)=|c|\mathrm{ar}(c) = |c|; an optional core label is an element of ℒ\mathcal{L}.

For ϑ∈ℝ\vartheta \in \mathbb{R} and m≥0m \ge 0, the core generating polynomial coefficient cm(ϑ)c_m(\vartheta) and block generating polynomial coefficient Bm(ϑ)B_m(\vartheta) are the real numbers cm(ϑ)=∑w∈ℰ2mw connected coreϑf(w),Bm(ϑ)=𝟏m=0+cm(ϑ).c_m(\vartheta) = \sum_{\substack{w \in \mathcal{E}_{2m} \\ w \text{ connected core}}} \vartheta^{f(w)}, \qquad B_m(\vartheta) = \mathbf{1}_{m=0} + c_m(\vartheta).(11) Under FK model parameters p∈(1/2,1)p \in (1/2, 1), the actual FK block mass bm(p)b_m(p) at size m≥0m \ge 0 is the real number defined from (10) and (11) by bm(p)=Bm(tp)(uτ2)mτ,b_m(p) = \frac{B_m(t_p)\,(u\tau^2)^m}{\tau},(12) For p∈(1/2,1)p \in (1/2, 1), the critical FK corner offspring law νp\nu_p is the critical even offspring distribution constructed from the block weights (bm(p))m≥0(b_m(p))_{m \ge 0} in (12).

Given scalars ϑ,r,τ′∈ℝ\vartheta, r, \tau' \in \mathbb{R}, the parametric core label mass assigns weight 1/τ′1/\tau' to the empty label ∘\circ and weight ϑf(c)(rτ′)|c|/τ′\vartheta^{f(c)}(r\tau')^{|c|}/\tau' (via the inventory word monomial weight at parameters ϑ\vartheta and rτ′r\tau') to each connected typed core cc.✦ Evaluating this mass at (ϑ,r,τ′)=(tp,u,τ)(\vartheta, r, \tau') = (t_p, \sqrt{u}, \tau) from (10) gives the root core mark mass and defines the root core mark distribution on ℒ\mathcal{L}. For each k≥0k \ge 0, the conditional core mark distribution κp,k\kappa_{p,k} on ℒ\mathcal{L} is the root core mark distribution conditioned on labels of arity kk if any such label exists, and the point mass at ∘\circ otherwise.✦ For n≥0n \ge 0, a marked tree code sample of size nn is a pair (d,ℓ)∈𝒞2(n+1)+1×ℒ2(n+1)+1(d, \ell) \in \mathcal{C}_{2(n+1)+1} \times \mathcal{L}^{2(n+1)+1}.✦ The law of marked code draw ℚp,n\mathbb{Q}_{p,n} is the marked tree code law of νp\nu_p at size n+1n+1 with mark kernel (κp,k)k≥0(\kappa_{p,k})_{k \ge 0}✦, and the actual tree height path H(p,n)(d,ℓ)∈C([0,1],ℝ)H^{(p,n)}(d, \ell) \in C([0,1], \mathbb{R}) scales (5) by the height scaling factor αn\alpha_n: αn=1+12(n+1),H(p,n)(d,ℓ)=αnh2(n+1)+1(νp,d).\alpha_n = \sqrt{1 + \frac{1}{2(n+1)}}, \qquad H^{(p,n)}(d, \ell) = \alpha_n\, h_{2(n+1)+1}(\nu_p, d).(13)

The Claim

Theorem 1. Let ν\nu be a critical even offspring distribution with standard deviation σ\sigma. For every probability space (Ω,ℙ)(\Omega, \mathbb{P}), any three independent standard Brownian motions B=(B(0),B(1),B(2))B = (B^{(0)}, B^{(1)}, B^{(2)}) on (Ω,ℙ)(\Omega, \mathbb{P}), and every bounded continuous function F:C([0,1],ℝ2)→ℝF : C([0,1], \mathbb{R}^2) \to \mathbb{R}, the tree code test expectations (6) and joint Brownian test expectation (9) satisfy limn→∞E2n+1ν[F]=𝔼ℙ[F(ediag(B))].\lim_{n \to \infty} E_{2n+1}^\nu[F] = \mathbb{E}_{\mathbb{P}}\bigl[F(e_{\mathrm{diag}}(B))\bigr].(14)✦ Furthermore, for any FK model parameter p∈(1/2,1)p \in (1/2, 1), any probability space (Ω,ℙ)(\Omega, \mathbb{P}), any three independent standard Brownian motions B=(B(0),B(1),B(2))B = (B^{(0)}, B^{(1)}, B^{(2)}) on (Ω,ℙ)(\Omega, \mathbb{P}), and pairs (dn,ℓn)(d_n, \ell_n) distributed according to ℚp,n\mathbb{Q}_{p,n} for each n≥0n \ge 0, the actual tree height path (13) converges in distribution as n→∞n \to \infty to the standard Brownian excursion path e(B)e(B): H(p,n)(dn,ℓn)→n→∞de(B).H^{(p,n)}(d_n, \ell_n) \xrightarrow[n \to \infty]{d} e(B).(15)✦ Finally, for any probability space (Ω,ℙ)(\Omega, \mathbb{P}), any standard Brownian motion WW on (Ω,ℙ)(\Omega, \mathbb{P}), and any three independent standard Brownian motions B=(B(0),B(1),B(2))B = (B^{(0)}, B^{(1)}, B^{(2)}) on (Ω,ℙ)(\Omega, \mathbb{P}), the Vervaat transform of the Brownian bridge path WbrW^{\mathrm{br}} has the same distribution in C([0,1],ℝ)C([0,1], \mathbb{R}) as e(B)e(B): V(Wbr)=de(B).V(W^{\mathrm{br}}) \stackrel{d}{=} e(B).(16)✦

In the paper, the result reads: under the critical even offspring assumptions on ν\nu with standard deviation σ\sigma, the rescaled pair (WNt/(σN),σHNt/(2N))0≤t≤1(W_{Nt}/(\sigma\sqrt{N}),\, \sigma H_{Nt}/(2\sqrt{N}))_{0 \le t \le 1} (interpolating the walk WjW_j and height HjH_j at tree size NN) converges in distribution in C([0,1],ℝ2)C([0,1], \mathbb{R}^2) to (e(t),e(t))0≤t≤1(e(t), e(t))_{0 \le t \le 1}, where ee is a standard normalized Brownian excursion. The library states this convergence along N=2n+1N = 2n+1 through expectations of bounded continuous test functions on the piecewise linear paths J2n+1(σ,d)J_{2n+1}(\sigma, d) and the diagonal Brownian excursion path ediag(B)e_{\mathrm{diag}}(B) in (14), alongside the FK height limit (15) and the Vervaat identity (16).

What Is Checked

The joint tree coding claim (14) is stated over any critical even offspring distribution ν\nu with standard deviation σ\sigma, its tree code weights wνw_\nu and partition functions Z2n+1(ν)Z_{2n+1}(\nu), the piecewise linear joint tree code paths J2n+1(σ,d)J_{2n+1}(\sigma, d) built from the partial walk WjW_j and vertex height HjH_j, three independent standard Brownian motions B=(B(0),B(1),B(2))B = (B^{(0)}, B^{(1)}, B^{(2)}) on a probability space (Ω,ℙ)(\Omega, \mathbb{P}), the diagonal Brownian excursion path ediag(B)e_{\mathrm{diag}}(B), and bounded continuous test functions F:C([0,1],ℝ2)→ℝF : C([0,1], \mathbb{R}^2) \to \mathbb{R}. The FK height limit (15) is stated over FK model parameters p∈(1/2,1)p \in (1/2, 1), the marked tree code law ℚp,n\mathbb{Q}_{p,n} constructed from the critical FK corner offspring law νp\nu_p (via the actual FK block masses (12) and the constructed even offspring distribution) and conditional core mark distributions κp,k\kappa_{p,k} (via connected typed cores, two-stack matching certificates, and chord intersection graphs), the actual tree height path H(p,n)H^{(p,n)}, and the standard Brownian excursion path e(B)e(B). The Vervaat identity (16) is stated over a standard Brownian motion WW and three independent standard Brownian motions BB on (Ω,ℙ)(\Omega, \mathbb{P}), the Brownian bridge path WbrW^{\mathrm{br}}, the Vervaat transform VV via the first minimum time and cyclic path shift, and e(B)e(B).

Beyond these claims, the library establishes that the actual FK block masses (12) satisfy bm(p)≥0b_m(p) \ge 0 for all m≥0m \ge 0✦, with b0(p)>0b_0(p) > 0✦ and b1(p)>0b_1(p) > 0✦. It also establishes that rotating any discrete bridge sequence dd of positive length NN by its good rotation index m*(d)m^*(d) satisfies the valid tree code condition✦, and that every continuous function on [0,1][0,1] attains its global minimum at a least time in [0,1][0,1].✦