Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

BanditAlgorithm.isotropic_gaussian_unregularized_quadratic_mixture

Proved

by Harry_Xu · Jul 29, 2026 · Mathlib 0df444a (Lean v4.33.1)

bandit-algorithmsconcentration-inequalitiesgaussianprobability

Let d∈Nd\in\mathbb Nd∈N and λ>0\lambda>0λ>0. Prove that there exists a centered isotropic Gaussian probability measure h=N(0,λ−1I)h=N(0,\lambda^{-1}I)h=N(0,λ−1I) on Rd\mathbb R^dRd such that, for every vector S∈RdS\in\mathbb R^dS∈Rd and every positive-semidefinite matrix V∈Rd×dV\in\mathbb R^{d\times d}V∈Rd×d, the function

x⟼exp⁡ ⁣(⟨x,S⟩−12x⊤Vx)x\longmapsto \exp\!\left( \langle x,S\rangle-\frac12 x^\top Vx \right)x⟼exp(⟨x,S⟩−21​x⊤Vx)

is integrable with respect to hhh and

∫Rdexp⁡ ⁣(⟨x,S⟩−12x⊤Vx) dh(x)=exp⁡ ⁣[12(S⊤(λI+V)−1S−log⁡det⁡(λI+V)λd)].\int_{\mathbb R^d} \exp\!\left( \langle x,S\rangle-\frac12 x^\top Vx \right)\,dh(x) = \exp\!\left[ \frac12\left( S^\top(\lambda I+V)^{-1}S -\log\frac{\det(\lambda I+V)}{\lambda^d} \right) \right].∫Rd​exp(⟨x,S⟩−21​x⊤Vx)dh(x)=exp[21​(S⊤(λI+V)−1S−logλddet(λI+V)​)].

This is the Gaussian completion-of-the-square identity used in the method of mixtures. The term λI\lambda IλI comes from the Gaussian density, while VVV is the unregularized quadratic variation.

Source: Tor Lattimore and Csaba Szepesvári, Bandit Algorithms (Cambridge University Press, 2020), Lemma 20.3 and equation (20.8), p. 259.

Preamble
import Mathlib.Probability.Distributions.Gaussian.Multivariate
import Mathlib.Analysis.SpecialFunctions.Log.Basic
import Mathlib.LinearAlgebra.Matrix.PosDef

open MeasureTheory ProbabilityTheory Matrix
Formal statement
theorem BanditAlgorithm.isotropic_gaussian_unregularized_quadratic_mixture
    {d : ℕ} {lam : ℝ} (hlam : 0 < lam) :
    ∃ h : Measure (Fin d → ℝ),
      IsProbabilityMeasure h ∧
      ∀ (S : Fin d → ℝ) (V : Matrix (Fin d) (Fin d) ℝ),
        V.PosSemidef →
        Integrable
            (fun x => Real.exp
              (x ⬝ᵥ S - 1 / 2 * (x ⬝ᵥ V *ᵥ x))) h ∧
          (∫ x, Real.exp
              (x ⬝ᵥ S - 1 / 2 * (x ⬝ᵥ V *ᵥ x)) ∂h) =
            Real.exp
              (1 / 2 *
                (S ⬝ᵥ (lam • (1 : Matrix (Fin d) (Fin d) ℝ) + V)⁻¹ *ᵥ S -
                  Real.log
                    ((lam • (1 : Matrix (Fin d) (Fin d) ℝ) + V).det / lam ^ d))) := by
  sorry
Source
Tor Lattimore and Csaba Szepesvári, Bandit Algorithms, Cambridge University Press, 2020, p. 259, Lemma 20.3 and Eq. (20.8).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me