Bandits.UCB.forall_arm_prop
No docstring.
Bandits.UCB.forall_arm_prop.{u_1} {K : โ} {hK : 0 < K} {c : โ} {ฮฝ : ProbabilityTheory.Kernel (Fin K) โ} [ProbabilityTheory.IsMarkovKernel ฮฝ] {ฮฉ : Type u_1} {mฮฉ : MeasurableSpace ฮฉ} {P : MeasureTheory.Measure ฮฉ} [MeasureTheory.IsProbabilityMeasure P] {A : โ โ ฮฉ โ Fin K} {R : โ โ ฮฉ โ โ} (h : Learning.IsAlgEnvSeq A R (ucbAlgorithm hK c) (Learning.stationaryEnv ฮฝ) P) : โแต (h : ฮฉ) โP, (โ n < K, A n h = โจn % K, โฏโฉ) โง โ (n : โ), K โค n โ โ (a : Fin K), Learning.empMean A R a n h + ucbWidth A c a n h โค Learning.empMean A R (A n h) n h + ucbWidth A c (A n h) n hBandits.UCB.forall_arm_prop.{u_1} {K : โ} {hK : 0 < K} {c : โ} {ฮฝ : ProbabilityTheory.Kernel (Fin K) โ} [ProbabilityTheory.IsMarkovKernel ฮฝ] {ฮฉ : Type u_1} {mฮฉ : MeasurableSpace ฮฉ} {P : MeasureTheory.Measure ฮฉ} [MeasureTheory.IsProbabilityMeasure P] {A : โ โ ฮฉ โ Fin K} {R : โ โ ฮฉ โ โ} (h : Learning.IsAlgEnvSeq A R (ucbAlgorithm hK c) (Learning.stationaryEnv ฮฝ) P) : โแต (h : ฮฉ) โP, (โ n < K, A n h = โจn % K, โฏโฉ) โง โ (n : โ), K โค n โ โ (a : Fin K), Learning.empMean A R a n h + ucbWidth A c a n h โค Learning.empMean A R (A n h) n h + ucbWidth A c (A n h) n h
Code
lemma forall_arm_prop
(h : IsAlgEnvSeq A R (ucbAlgorithm hK c) (stationaryEnv ฮฝ) P) :
โแต h โP,
(โ n < K, A n h = โจn % K, Nat.mod_lt _ hKโฉ) โง
(โ n, K โค n โ โ a, empMean A R a n h + ucbWidth A c a n h โค
empMean A R (A n h) n h + ucbWidth A c (A n h) n h)Proof
by
simp only [eventually_and]
constructor
ยท exact forall_arm_eq_mod_of_lt h
ยท simp_rw [ae_all_iff]
intro n hn a
have h_ae := forall_ucbIndex_le_ucbIndex_arm h a
simp_rw [ae_all_iff] at h_ae
exact h_ae n hnActions: Source ยท Open Issue
Meaning last changed in v4.34.0-rc2-14-gf86702d (2026-08-25), the 5th 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: 21 project declarations, 117 external constants
โ 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.