LeanMachineLearning

Bandits.ClippedUCB.ucb_eq_ucb'๐Ÿ”—

Lemma

No docstring.

Types
  • ฮฉ : Type u_1
Given
  • K : โ„•
  • l : โ„
  • u : โ„
  • ฯƒ2 : โ„
  • ฮด : โ„
  • A : โ„• โ†’ ฮฉ โ†’ Fin K
  • R : โ„• โ†’ ฮฉ โ†’ โ„
  • O : โ„• โ†’ ฮฉ โ†’ Unit
  • a : Fin K
  • n : โ„•
  • ฯ‰ : ฮฉ
Then
ucb A R l u ฯƒ2 ฮด a n ฯ‰ = ucb' n (Learning.history O A R n ฯ‰) l u ฯƒ2 ฮด a
Code
lemma ucb_eq_ucb' {O : โ„• โ†’ ฮฉ โ†’ Unit} {a : Fin K} {n : โ„•} {ฯ‰ : ฮฉ} :
    ucb A R l u ฯƒ2 ฮด a n ฯ‰ = ucb' n (history O A R n ฯ‰) l u ฯƒ2 ฮด a
Proof
by
  have hp : pullCount A a n ฯ‰ = pullCount' n (history O A R n ฯ‰) a := pullCount_eq_pullCount'
  have he : empMean A R a n ฯ‰ = empMean' n (history O A R n ฯ‰) a := empMean_eq_empMean'
  rw [ucb, ucb', hp, he]

Meaning last changed in v4.34.0-rc2-76-g565f652 (2026-09-10), the 2th recorded change.

Self-contained, with its dependencies inlined and proofs replaced by sorry: download the raw file ยท open it in the Lean web editor.

Dependency graph

Audit surface: 13 project declarations, 52 external constants

โœ“ Proved: no sorry anywhere in its closure

This is the tool's own reading of one build's recorded axioms, and it is not robust against an author who wants it to pass. Checking meant to be relied on should go through Comparator, which replays the proof through the kernel from an export against an explicit list of permitted axioms.