OctoInformBeta

OpenAI preprint, September 2026 · §2. Words and strict drift · §2

Lemma 2.1 [Reduction above an older supply]: Normal words preserve evaluation, unfilled orders consume older burgers in order, and prepending letters preserves the freshest surviving burger

Reduction of inventory words above an older burger supply.

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

In the exploration of random planar maps decorated by Fortuin-Kasteleyn percolation, the boundary of the explored region carries interfaces between percolation clusters. The inventory method records boundary conditions using an alphabet whose letters represent available interface endpoints, termed burgers, and requests to match them, termed orders. When an exploration step reveals new letters, earlier unmatched burgers remain as an older supply available to subsequent orders.

An inventory order o∈𝒪={0,1}∪{⋆}o \in \mathcal{O} = \{0, 1\} \cup \{\star\} is an order request that is either a fixed order of type b∈{0,1}b \in \{0, 1\} or the flexible order ⋆\star capable of matching either burger type.✦ A letter ℓ\ell of the inventory alphabet Σ\Sigma is an element of Σ\Sigma that is either a burger token of type b∈{0,1}b \in \{0, 1\} or an inventory order o∈𝒪o \in \mathcal{O}. An inventory evaluation state e=(oe,be)e = (o_e, b_e) is a pair consisting of a finite list oeo_e of pending orders in 𝒪\mathcal{O} and a finite stack beb_e of burgers in {0,1}\{0, 1\} stored freshest-first with freshest burger head⁡(be)\operatorname{head}(b_e), with order component orders⁡(e)=oe\operatorname{orders}(e) = o_e and burger component burgers⁡(e)=be\operatorname{burgers}(e) = b_e.✦

The evaluation run⁡(w,e)\operatorname{run}(w, e) of a finite word w∈Σ*w \in \Sigma^* on an evaluation state ee is the evaluation state obtained by sequentially processing the letters of ww: each incoming burger is placed at the top of the burger stack, and each incoming order consumes the freshest eligible burger from the stack or is appended to the list of pending orders if no eligible burger is available.✦ The reduction reduce⁡(w)\operatorname{reduce}(w) of a finite word w∈Σ*w \in \Sigma^* is the evaluation state obtained by evaluating ww against empty order and burger stacks, as in (1): reduce⁡(w)=run⁡(w,(∅,∅)).\operatorname{reduce}(w) = \operatorname{run}(w, (\emptyset, \emptyset)).(1)✦

The order evaluation evalOrders⁡(a,s)\operatorname{evalOrders}(a, s) of a list of orders aa against a burger stack ss is the evaluation state obtained by feeding the orders aa to ss with empty initial orders, as in (2): evalOrders⁡(a,s)=run⁡(a,(∅,s)).\operatorname{evalOrders}(a, s) = \operatorname{run}(a, (\emptyset, s)).(2)✦ The normal word normalWord⁡(e)\operatorname{normalWord}(e) of an evaluation state e=(oe,be)e = (o_e, b_e) is the word in Σ*\Sigma^* formed by the pending orders oeo_e followed by the burgers beb_e reversed to run from oldest to freshest, as in (3): normalWord⁡(e)=oe⋅berev.\operatorname{normalWord}(e) = o_e \cdot b_e^{\mathrm{rev}}.(3)

The Claim

Theorem 1. Let ww and w′w' be finite words over the inventory alphabet Σ\Sigma, let ss be a burger stack, and let ee be an evaluation state. With ow=orders⁡(reduce⁡(w))o_w = \operatorname{orders}(\operatorname{reduce}(w)) and bw=burgers⁡(reduce⁡(w))b_w = \operatorname{burgers}(\operatorname{reduce}(w)), when ww acts on an older burger supply ss with empty initial orders, the resulting state satisfies (4): run⁡(w,(∅,s))=(orders⁡(evalOrders⁡(ow,s)),bw⋅burgers⁡(evalOrders⁡(ow,s))),\operatorname{run}(w, (\emptyset, s)) = \bigl(\operatorname{orders}(\operatorname{evalOrders}(o_w, s)),\, b_w \cdot \operatorname{burgers}(\operatorname{evalOrders}(o_w, s))\bigr),(4)✦ Evaluating the normal word of the reduction of ww on ee matches the action of ww via (5): run⁡(normalWord⁡(reduce⁡(w)),e)=run⁡(w,e).\operatorname{run}(\operatorname{normalWord}(\operatorname{reduce}(w)), e) = \operatorname{run}(w, e).(5)✦ If the burger stack of reduce⁡(w)\operatorname{reduce}(w) has freshest burger b∈{0,1}b \in \{0, 1\}, then the freshest surviving burger head⁡(burgers⁡(reduce⁡(w′w)))\operatorname{head}(\operatorname{burgers}(\operatorname{reduce}(w' w))) of the prepended word satisfies (6): head⁡(burgers⁡(reduce⁡(w′w)))=b.\operatorname{head}(\operatorname{burgers}(\operatorname{reduce}(w' w))) = b.(6)✦

In the paper, the result reads: feed a finite word an arbitrary older list of burgers; every match between two letters of the word is the same as when the word is read from an empty supply, the orders unfilled in the latter evaluation act on the older list in their original chronology, the surviving burgers of the word remain above the surviving older burgers, in particular finite words may be replaced by their reductions when concatenating them, and once a suffix has a surviving burger, its freshest survivor is unchanged by prepending further letters.

What Is Checked

The three identities are stated over finite words of inventory letters, burger stacks, and inventory evaluation states.