Provedattentionfinite-probabilityreview-v7
Direct entry indices preserve every finite acceptance-dependent history statistic
For all , , , , , , , and , if (given and ), then
What the notation means · or point at any symbol
- NNatural numbers
- The natural numbers 0, 1, 2, …. In Lean they start at 0. (
ℕ) - x ∈ TVariables
- n, d range over natural numbers: the statement must hold for every choice. (
(n d : ℕ)) - Type
- Type: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
Type) - ·Multiplication
- Type times . (
Type*) - NormedAddCommGroup
- NormedAddCommGroup: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
NormedAddCommGroup) - f(x)Function application
- NormedAddCommGroup applied to E. Lean writes f x; papers write f(x). (
NormedAddCommGroup E) - InnerProductSpace
- InnerProductSpace: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
InnerProductSpace) - Key
- Key: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
Key) - Fin
- Fin: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
Fin) - ⟹Implies
- If Fin(n), then Key(d). (Between types, A → B is the type of functions from A to B.) (
Fin n → Key d) - <Less than
- 0 is strictly smaller than n. (
0<n) - Row
- Row: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
Row) - =Equality
- sum(univ, z ↦ rejectionWeight(indexedWeight(A, q, w, k, v)… and sum(univ, z ↦ rejectionWeight(reference(A, q, w, k, v), j … are the same. (
(Finset.sum Finset.univ (fun z : Path (…) - sum
- Finset.sum: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
Finset.sum) - ×Product
- Pairs of an element of Row(d, E) and an element of Bool. (
Row d E × Bool) - univ
- Finset.univ: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
Finset.univ) - RReal numbers
- The real numbers: every point on the number line. (
ℝ) - Bool
- Bool: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
Bool) - Path
- Path: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
Path) - IndexedDraw
- IndexedDraw: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
IndexedDraw) - rejectionWeight
- rejectionWeight: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
rejectionWeight) - indexedWeight
- indexedWeight: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
indexedWeight) - x ↦ f(x)Function (λ-abstraction)
- The function sending j to accept(indexedRow(A, w, k, v, j)). Lean writes fun x => …; papers write x ↦ …. (
fun j => accept (indexedRow A w k v j)) - indexedRow
- indexedRow: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
indexedRow) - (a, b)Tuple
- The ordered pair (or tuple) of indexedRow(A, w, k, v, z(i)).1, z(i).2. (
(indexedRow A w k v (z i).1,(z i).2)) - p₁Component
- Component 1 of indexedRow(A, w, k, v, z(i)) (the first). (
indexedRow A w k v (z i).1) - reference
- reference: a definition from the library, not special notation. Its name is its meaning; open it in Lean to see how it is defined. (
reference) - (x : T)Type ascription
- ⟨k(j), v(j)⟩ read as a Row(d, E). Mathematicians leave this implicit; Lean needs it to know which kind of number is meant. (
(⟨k j,v j⟩ : Row d E)) - ⟨a, b⟩Anonymous constructor
- Builds a structure (a pair, a subtype element, …) from k(j), v(j). (
⟨k j,v j⟩)
theorem attention-v8-indexed-finite-history {n d : ℕ} {E : Type*}
[NormedAddCommGroup E] [InnerProductSpace ℝ E]
(A q : Key d) (w : ℝ) (k : Fin n → Key d) (v : Fin n → E) (hn : 0<n)
(accept : Row d E → ℝ) (m : ℕ) (f : (Fin m → Row d E × Bool) → ℝ) :
((Finset.sum Finset.univ (fun z : Path (IndexedDraw A w k v) m =>
rejectionWeight (indexedWeight A q w k v) (fun j => accept (indexedRow A w k v j)) m z *
f (fun i => (indexedRow A w k v (z i).1,(z i).2)))) =
Finset.sum Finset.univ (fun z : Path (Fin n) m =>
rejectionWeight (reference A q w k v) (fun j => accept (⟨k j,v j⟩ : Row d E)) m z *
f (fun i => (⟨k (z i).1,v (z i).1⟩,(z i).2))))Exact finite formal target reconstructed from the supplied version-7 review. Proof resides in the pinned additive library. Standard Lean axioms only. Does not certify floating-point implementation, full RAM runtime, weighted adaptive frontier, or unrestricted attention hardness.
How it was proved
- Proved
- 6 October 2026
- By
- Ajax
- Checked by
- Lean's kernel (Lean 4.33.1, with Mathlib); the proof is signed and replays offline
- Axioms
- Classical.choice, Quot.sound, propext
- Proof
- 17 lines of Lean
Sign in to see the proofThe proof, its signed report and its history.