BanditRLlib
Lean gate passed before this site build; local proof declarations are shown as compiled.

Lean module · OFUL

BanditRLProof.OFULEllipticalPotentialFoundation

# Standard OFUL elliptical-potential endpoint This module packages the determinant-growth and clipped inverse-quadratic conclusions of the deterministic OFUL linear-algebra route in one theorem-facing statement. It is the handoff expected by a later self-normalized concentration and confidence-ellipsoid proof.

Module map

Declarations
1
Placeholders
0

Imports

BanditRLProof.OFULEllipticalPotential

Imported by

BanditRLProof, BanditRLProof.OFULSelectedWidthSummation, BanditRLProof.OFULSelfNormalizedConfidence

Declarations

Open an item to read its exact compact statement and source link. Detailed teaching notes are linked when registered.

theorem BanditRLProof.OFUL.standardLogDeterminantAndEllipticalPotential Compiled

Standard logarithmic determinant and elliptical-potential bounds. For a positive scalar regularization and uniformly bounded squared feature norms, the terminal regularized Gram determinant and the cumulative clipped inverse-quadratic updates are controlled by `d * log (1 + T * L2 / (d * lambda))`.

theorem standardLogDeterminantAndEllipticalPotential {Feature : Type u} [Fintype Feature] [DecidableEq Feature] [Nonempty Feature] (lambda : Real) (hlambda : 0 < lambda) (history : Nat -> Feature -> Real) (T : Nat) (L2 : Real) (hL2 : 0 <= L2) (hbound : forall t : Nat, t < T -> dotProduct (history t) (history t) <= L2) : (regularizedPrefixFeatureGram lambda history T).det <= lambda ^ Fintype.card Feature * Real.exp ((Fintype.card Feature : Real) * Real.log (1 + (T * L2) / ((Fintype.card Feature : Real) * lambda))) /\ (Finset.range T).sum (fun t => min 1 (dotProduct (history t) (Matrix.mulVec ((regularizedPrefixFeatureGram lambda history t)⁻¹) (history t)))) <= 2 * ((Fintype.card Feature : Real) * Real.log (1 + (T * L2) / ((Fintype.card Feature : Real) * lambda)))