OctoInformBeta

OpenAI preprint, September 2026 · §6. From the block tree to the map · §6

Lemma 6.2 [Corner mass]: For ordered trees of cores and rooted planar maps, giving a vertex-constant pole sum and total normalized degree mass 11

Rotation-intertwining bijections equate nonroot pole and dart Dirac sums.

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 rooted finite planar map represented as a rotation system and decomposed via a two-stack burger-order encoding into an ordered tree of connected cores. Each nonroot node of this tree designates a pole vertex in the map, so placing equal mass on all nonroot tree nodes pushes forward to a measure on map vertices. The question is whether this corner-pole measure coincides with the normalized vertex degree measure, where each dart contributes equal weight to the vertex around which it rotates.

Rooted Planar Maps and Degree Measure

The dart set DnD_n of a planar map with n+1n+1 edges (n≥0n \ge 0) is the set {0,…,2(n+1)−1}\{0, \dots, 2(n+1)-1\} indexing its directed edges, with 00 designated as the root dart.✦ The orbit [d]s[d]_s of a dart d∈Dnd \in D_n under a permutation ss of DnD_n is the cycle equivalence class of DnD_n under ss containing dd✦, and the orbits of ss are these equivalence classes.✦ The spanning subgraph determined by permutations aa and ss of DnD_n and a subset AA of the orbits of aa is the simple graph on the orbits of ss in which two orbits vv and ww are adjacent whenever some dart d∈Dnd \in D_n satisfies [d]a∈A[d]_a \in A, [d]s=v[d]_s = v, and [a(d)]s=w[a(d)]_s = w.✦

A pair (a,s)(a, s) of permutations of DnD_n satisfies the planar map conditions if aa is a fixed-point-free involution, the spanning subgraph determined by aa, ss, and all orbits of aa is connected, and the number of orbits of ss plus the number of orbits of s∘as \circ a equals n+3n+3.✦ A labelled planar map (a,s)(a, s) with n+1n+1 edges is a pair of permutations of DnD_n satisfying the planar map conditions.✦ An isomorphism rr of rooted maps between two labelled planar maps (a1,s1)(a_1, s_1) and (a2,s2)(a_2, s_2) with n+1n+1 edges is a permutation of DnD_n fixing the root dart 00 such that a2(r(d))=r(a1(d))a_2(r(d)) = r(a_1(d)) and s2(r(d))=r(s1(d))s_2(r(d)) = r(s_1(d)) for every d∈Dnd \in D_n.✦ A rooted planar map MM with n+1n+1 edges is an equivalence class of labelled planar maps with n+1n+1 edges under the equivalence relation generated by isomorphisms of rooted maps.✦

For a rooted planar map MM with n+1n+1 edges, the vertex rotation permutation σM\sigma_M on DnD_n is the second permutation of a representative labelled planar map of MM, cyclically ordering the darts around each vertex.✦ The vertices V(M)V(M) of a rooted planar map MM with n+1n+1 edges are the orbits of σM\sigma_M on DnD_n.✦ Given a rooted planar map MM with n+1n+1 edges and writing δv\delta_v for the Dirac unit mass at a vertex v∈V(M)v \in V(M), the normalized vertex degree measure μM\mu_M of MM is the measure on V(M)V(M) defined by (1): μM=∑d∈Dn12(n+1)δ[d]σM.\mu_M = \sum_{d \in D_n} \frac{1}{2(n+1)}\,\delta_{[d]_{\sigma_M}}.(1)✦

Two-stack Matchings and Permutation Wedges

The typed burger-order alphabet 𝒜\mathcal{A} is the set of tokens consisting of burger tokens and order tokens carrying one of two boolean types (true or false).✦ The typed word stack execution of a finite word in 𝒜\mathcal{A} against a stack of boolean types stored freshest-first evaluates the word letter by letter against the stack, with each order consuming the freshest eligible burger, and returns either the resulting stack upon success or failure upon an illegal closure.✦

A non-crossing arch matching EE along a finite list ll of elements of a set α\alpha is a list of ordered pairs in α×α\alpha \times \alpha built recursively from the empty list on the empty list by skipping an unmatched element or by pairing elements aa and bb with (a,b)(a, b) prepended to the concatenation of non-crossing arch matchings on the sublist strictly between aa and bb and on the sublist following bb.✦ The endpoints list of an arch list EE in α×α\alpha \times \alpha is the list in α\alpha collecting the first and second coordinates of each pair in EE in order.✦ The arch transposition product τ(E)\tau(E) of an arch list EE in α×α\alpha \times \alpha is the permutation of α\alpha defined recursively as the identity on the empty list and as τ(E′)∘(ab)\tau(E') \circ (a\;b) when (a,b)(a, b) precedes the tail E′E'.✦

A two-stack matching certificate ℳ=(E1,E2)\mathcal{M} = (E_1, E_2) for a function w:α→𝒜w \colon \alpha \to \mathcal{A} along a finite list ll in α\alpha is a pair of non-crossing arch matchings E1E_1 and E2E_2 along ll whose combined endpoints list is a permutation of ll, with every pair (x,y)∈E1(x, y) \in E_1 matching a true burger at xx to a true order at yy and every pair (x,y)∈E2(x, y) \in E_2 matching a false burger at xx to a false order at yy.✦ The stack matching mate permutation ν(ℳ)\nu(\mathcal{M}) of a two-stack matching certificate ℳ=(E1,E2)\mathcal{M} = (E_1, E_2) is the arch transposition product of the concatenation of E1E_1 followed by E2E_2.✦ For a function w:α→𝒜w \colon \alpha \to \mathcal{A} and a finite list ll in α\alpha, whenever typed word stack execution on the mapped word of ll under ww starting from an empty stack terminates in an empty stack, the matching-of-empty-word construction produces a two-stack matching certificate for ww along ll.✦

The permutation-wedge vertices of a permutation pp of a set α\alpha are the cycle equivalence classes of α\alpha under pp✦, and the cycle vertex projection sends each d∈αd \in \alpha to its cycle class [d]p[d]_p.✦ The permutation-wedge dart graph of permutations ε\varepsilon and pp of a set α\alpha is the simple graph on the permutation-wedge vertices of pp in which two vertices vv and ww are adjacent whenever some d∈αd \in \alpha satisfies [d]p=v[d]_p = v and [ε(d)]p=w[\varepsilon(d)]_p = w.✦ For a two-stack matching certificate ℳ=(E1,E2)\mathcal{M} = (E_1, E_2) along a finite list ll in α\alpha, let ρl\rho_l be the cyclic permutation formed by ll and let sℳ=ρl∘τ(E1)s_{\mathcal{M}} = \rho_l \circ \tau(E_1); the typed matching distance dist⁡ℳ(x,y)∈ℕ∪{∞}\operatorname{dist}_{\mathcal{M}}(x, y) \in \mathbb{N} \cup \{\infty\} between elements x,y∈αx, y \in \alpha is the extended graph distance between [x]sℳ[x]_{s_{\mathcal{M}}} and [y]sℳ[y]_{s_{\mathcal{M}}} in the permutation-wedge dart graph of ν(ℳ)\nu(\mathcal{M}) and sℳs_{\mathcal{M}}.✦

A chord opening endpoint (or opener) ii of a permutation aa of {0,…,m−1}\{0, \dots, m-1\} (m≥0m \ge 0) is an index in {0,…,m−1}\{0, \dots, m-1\} satisfying i<a(i)i < a(i).✦ Two openers xx and yy of a permutation aa of {0,…,m−1}\{0, \dots, m-1\} satisfy the chord crossing relation when their endpoints interlace, meaning 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).✦ The chord intersection graph of a permutation aa of {0,…,m−1}\{0, \dots, m-1\} is the simple graph on the openers of aa whose edges are the pairs of crossing openers.✦

Ordered Trees of Cores

The length |s||s| of a finite word or list ss is the number of elements of ss. A finite word ww of letters in 𝒜\mathcal{A} satisfies the connected core word condition if there exist an integer K≥0K \ge 0, a map v:{0,…,K−1}→𝒜v \colon \{0, \dots, K-1\} \to \mathcal{A} whose list of values (v(0),…,v(K−1))(v(0), \dots, v(K-1)) equals ww, and a two-stack matching certificate ℳ\mathcal{M} for vv along (0,…,K−1)(0, \dots, K-1) whose mate permutation ν(ℳ)\nu(\mathcal{M}) has an opener rr at index 00 from which every opener of ν(ℳ)\nu(\mathcal{M}) is reachable in the chord intersection graph of ν(ℳ)\nu(\mathcal{M}).✦ A connected typed core cc is a finite word of letters in 𝒜\mathcal{A} satisfying the connected core word condition.✦ An ordered tree of cores tt is a tree defined recursively as either a leaf or a branch carrying a connected typed core cc together with child ordered trees of cores tit_i indexed by 0≤i<|c|0 \le i < |c|.✦ The word w(t)w(t) of an ordered tree of cores tt is the empty word when tt is a leaf, and when tt is a branch with core cc and child subtrees tit_i it is the word formed by inserting each child word w(ti)w(t_i) into cc at corner ii.✦

The core tree darts D(t)D(t) of an ordered tree of cores tt form a finite set defined recursively: a leaf has no darts, while at a branch with core cc and child subtrees tit_i, a dart in D(t)D(t) is either a position ii of cc or a position ii paired with a dart in D(ti)D(t_i).✦ The core tree dart label ℓt(d)∈𝒜\ell_t(d) \in \mathcal{A} of a dart d∈D(t)d \in D(t) in an ordered tree of cores tt is the ii-th letter of cc when dd is a position ii of the branch core cc, and is the core tree dart label in tit_i when dd belongs to the child subtree tit_i.✦ The core tree dart list L(t)L(t) of an ordered tree of cores tt lists all darts in D(t)D(t) in recursive substitution order: it is empty at a leaf, and at a branch with core cc and child subtrees tit_i it concatenates, for i=0,…,|c|−1i = 0, \dots, |c|-1, the core dart at ii followed by the dart list L(ti)L(t_i) inside D(t)D(t).✦ The core tree stack matching ℳt=(Et,1,Et,2)\mathcal{M}_t = (E_{t,1}, E_{t,2}) of an ordered tree of cores tt is the two-stack matching certificate obtained by applying the matching-of-empty-word construction along L(t)L(t) with label assignment ℓt\ell_t.✦ A function f:D(t)→Af \colon D(t) \to A from the darts of an ordered tree of cores tt to a set AA is vertex-constant on tt if f(a)=f(b)f(a) = f(b) whenever a,b∈D(t)a, b \in D(t) satisfy dist⁡ℳt(a,b)=0\operatorname{dist}_{\mathcal{M}_t}(a, b) = 0.✦

The core tree nodes N(t)N(t) of an ordered tree of cores tt are the vertices of tt itself, keeping a distinct trivial leaf at every empty corner word without identifying coincident map poles: a leaf has a single node, while a branch with core cc and child subtrees tit_i has a branch root node together with, for each 0≤i<|c|0 \le i < |c|, the nodes of tit_i.✦ An optional core label is a mark carrying either a connected typed core or nothing✦, and the core tree label sequence Λ(t)\Lambda(t) of an ordered tree of cores tt is the list of optional core labels returning a singleton empty mark at a leaf and, at a branch with core cc and child subtrees tit_i, prepending cc to the concatenation of the label sequences Λ(ti)\Lambda(t_i).✦ The preorder node index of a node v∈N(t)v \in N(t) in an ordered tree of cores tt is the nonnegative integer defined recursively to be 00 at a leaf or branch root, and 11 plus the total length of the label sequences preceding ii plus the preorder node index of vv in tit_i when vv belongs to the ii-th child subtree tit_i of a branch.✦ The preorder node word equivalence ϕt\phi_t of an ordered tree of cores tt is the bijection from N(t)N(t) onto {0,…,|w(t)|}\{0, \dots, |w(t)|\} obtained by reindexing the bijection from N(t)N(t) onto {0,…,|Λ(t)|−1}\{0, \dots, |\Lambda(t)|-1\} that maps each node to its preorder node index.✦

For an ordered tree of cores tt and a node v∈N(t)v \in N(t), the node pole dart ηt(v)\eta_t(v) is either a dart in D(t)D(t) or absent: it is absent at a leaf; at a branch with core cc and child subtrees tit_i, it is the entry core dart of cc at the branch root, while for vv in tit_i it is the image of ηti(v)\eta_{t_i}(v) in D(t)D(t) if ηti(v)\eta_{t_i}(v) is present and otherwise the core dart of cc at the next slot for ii.✦ Given an ordered tree of cores tt, a rooted planar map MM with n+1n+1 edges, and a bijection e:D(t)→Dne \colon D(t) \to D_n, the realized pole ηˆt,M,e(v)∈V(M)\hat{\eta}_{t,M,e}(v) \in V(M) of a node v∈N(t)v \in N(t) is [e(ηt(v))]σM[e(\eta_t(v))]_{\sigma_M} when ηt(v)\eta_t(v) is present and the root-dart orbit [0]σM[0]_{\sigma_M} otherwise.✦ The nonroot pole vertex vt,M,e(i)∈V(M)v_{t,M,e}(i) \in V(M) associated with an ordered tree of cores tt, a rooted planar map MM with n+1n+1 edges, a bijection e:D(t)→Dne \colon D(t) \to D_n, and an index 0≤i<|w(t)|0 \le i < |w(t)| is given by (2): vt,M,e(i)=ηˆt,M,e(ϕt−1(i+1)).v_{t,M,e}(i) = \hat{\eta}_{t,M,e}\bigl(\phi_t^{-1}(i+1)\bigr).(2)✦

The Claim

Theorem 1. Let tt be an ordered tree of cores, MM a rooted planar map with n+1n+1 edges, and e:D(t)→Dne \colon D(t) \to D_n a bijection satisfying σM(e(d))=e((ρL(t)∘τ(Et,1))(d))\sigma_M(e(d)) = e\bigl((\rho_{L(t)} \circ \tau(E_{t,1}))(d)\bigr) for every d∈D(t)d \in D(t); then (3) holds: ∑0≤i<|w(t)|δvt,M,e(i)=∑d∈Dnδ[d]σM.\sum_{0 \le i < |w(t)|} \delta_{v_{t,M,e}(i)} = \sum_{d \in D_n} \delta_{[d]_{\sigma_M}}.(3)✦ Let AA be a cancellative commutative additive monoid, tt an ordered tree of cores, f:D(t)→Af \colon D(t) \to A a vertex-constant function, and r∈Ar \in A, and let frf_r be the extension returning f(d)f(d) on a present dart d∈D(t)d \in D(t) and rr when absent; then (4) holds: ∑0≤i<|w(t)|fr(ηt(ϕt−1(i+1)))=∑d∈D(t)f(d).\sum_{0 \le i < |w(t)|} f_r\bigl(\eta_t(\phi_t^{-1}(i+1))\bigr) = \sum_{d \in D(t)} f(d).(4)✦ For every rooted planar map MM with n+1n+1 edges, its normalized vertex degree measure satisfies (5): μM(V(M))=1.\mu_M(V(M)) = 1.(5)✦

In the paper, the result reads: the uniform probability measure on the N−1=2nN - 1 = 2n nonroot tree nodes, pushed forward by the pole projection π\pi, is exactly μn\mu_n, where π\pi is the pole map on tree nodes and μn\mu_n is the normalized degree measure on a map with nn edges. For maps indexed by nn with n+1n+1 edges, the library states this correspondence through the unnormalized Dirac-sum identity (3) and dart-sum identity (4) together with the total-mass normalization (5) of μM\mu_M.

What Is Checked

The claims are stated over an arbitrary ordered tree of cores tt, an arbitrary rooted planar map MM with n+1n+1 edges (n≥0n \ge 0) in (3) and (5), a bijection e:D(t)→Dne \colon D(t) \to D_n satisfying σM(e(d))=e((ρL(t)∘τ(Et,1))(d))\sigma_M(e(d)) = e\bigl((\rho_{L(t)} \circ \tau(E_{t,1}))(d)\bigr) for every d∈D(t)d \in D(t) in (3), and a vertex-constant function f:D(t)→Af \colon D(t) \to A into a cancellative commutative additive monoid AA with default value r∈Ar \in A in (4).

For every ordered tree of cores tt, the label sequence satisfies |Λ(t)|=|w(t)|+1|\Lambda(t)| = |w(t)| + 1✦, and every node in N(t)N(t) has preorder node index strictly less than |Λ(t)||\Lambda(t)|.✦