Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Finite positive entropy measures and their exact performance interfaces

Definition
PGDRevision20261113_MeasureExtension

by Somehow · Nov 13, 2026 · Mathlib 0df444a (Lean v4.33.1)

bregman-geometryconvex-optimization

For a finite positive measure supported in [0,aˉ][0,\bar a][0,aˉ], with aˉ<1\bar a<1aˉ<1 and positive first moment, define

ψ(t)=∫H(a t) dν(a),A(t)=∫a1−at2 dν(a).\psi(t)=\int H(\sqrt a\,t)\,d\nu(a),\qquad A(t)=\int\frac{a}{1-at^2}\,d\nu(a).ψ(t)=∫H(a​t)dν(a),A(t)=∫1−at2a​dν(a).

This module supplies the integrated loss, power geometries, actual implicit-profile performance, and the target proposition extending the finite-mixture theorem. Finiteness, support, positive moment, and positive penalty scale are explicit hypotheses.

Definition code
import Definitions.Def_PGDRevision20261113_Geometry
import Mathlib.MeasureTheory.Integral.Bochner.Basic

set_option autoImplicit false
noncomputable section
open scoped ENNReal
open MeasureTheory

namespace PGDRevision20261113

def normalizedEntropy (t : ℝ) : ℝ :=
  ((1 + t) * Real.log (1 + t) + (1 - t) * Real.log (1 - t)) / 2

def measureEntropyLoss (nu : Measure ℝ) (t : ℝ) : ℝ :=
  ∫ a : ℝ, normalizedEntropy (Real.sqrt a * t) ∂nu

def measureEntropyCurvature (nu : Measure ℝ) (t : ℝ) : ℝ :=
  ∫ a : ℝ, a / (1 - a * t ^ 2) ∂nu

def measureEntropyGeometry (nu : Measure ℝ) (lam : ℝ) : GeometryData :=
  ⟨.interval 1, measureEntropyCurvature nu, powerCurvature 1, lam⟩

def measureEntropyProfile (nu : Measure ℝ) (lam c : ℝ) : ScalarProfile :=
  ⟨Set.Ioo (-1 : ℝ) 1, measureEntropyLoss nu,
    (fun t => lam * normalizedEntropy t), doublePrimitive (powerCurvature c)⟩

def AdmissibleEntropyMeasure (nu : Measure ℝ) (abar lam : ℝ) : Prop :=
  IsFiniteMeasure nu ∧ 0 < abar ∧ abar < 1 ∧ 0 < lam ∧
  nu (Set.Icc (0 : ℝ) abar)ᶜ = 0 ∧ 0 < ∫ a : ℝ, a ∂nu

def entropyMeasureA0 (nu : Measure ℝ) : ℝ := ∫ a : ℝ, a ∂nu
def entropyMeasureA1 (nu : Measure ℝ) : ℝ := ∫ a : ℝ, a / (1 - a) ∂nu
def entropyThreshold (a : ℝ) : ℝ := (Real.sqrt (1 + 8 * a) - 1) / 2

def MeasureEntropyLocation (nu : Measure ℝ) (abar lam : ℝ) (z : ℝ × ℝ) : Prop :=
  0 < z.1 ∧ z.1 < entropyThreshold abar ∧ entropyThreshold abar < 1 ∧
  entropyMeasureA0 nu / 2 < z.2 ∧ z.2 < entropyMeasureA1 nu ∧
  0 < geometryRho (measureEntropyGeometry nu lam) z.1 z.2 ∧
  geometryRho (measureEntropyGeometry nu lam) z.1 z.2 <
    ENNReal.ofReal (entropyMeasureA0 nu / (entropyMeasureA0 nu + 2 * lam)) ∧
  entropyMeasureA0 nu / (entropyMeasureA0 nu + 2 * lam) < 1

def NonnegativeAlternatingContacts (G : GeometryData) (z : ℝ × ℝ) (r : ℝ) : Prop :=
  ∃ (t : Fin 3 → ℝ) (sigma : ℝ),
    0 ≤ t 0 ∧ t 0 < t 1 ∧ t 1 < t 2 ∧ t 2 < 1 ∧
    (sigma = 1 ∨ sigma = -1) ∧
    ∀ i, geometryError G z.1 z.2 (t i) = alternatingSign sigma i * r

def fixedExponentRates (G : GeometryData) (c : ℝ) : Set ℝ≥0∞ :=
  {r | ∃ p > (0 : ℝ), r = geometryRho G c p}

def PositiveMeasureEntropyExtensionClaim : Prop :=
  ∀ (nu : Measure ℝ) (abar lam : ℝ), AdmissibleEntropyMeasure nu abar lam →
    let G := measureEntropyGeometry nu lam
    AdmissibleGeometry G ∧
    (∀ t ∈ Set.Ioo (-1 : ℝ) 1,
      deriv (deriv (measureEntropyLoss nu)) t = measureEntropyCurvature nu t) ∧
    (∀ c p : ℝ, 0 ≤ c → 0 < p →
      AdmissibleProfile (measureEntropyProfile nu lam c) p ∧
      ∀ N : ℕ, 1 ≤ N → profileOmega (measureEntropyProfile nu lam c) p N =
        geometryRho G c p ^ (2 * N)) ∧
    (∃! z : ℝ × ℝ, GeometryOptimum G z) ∧
    (∀ z, GeometryOptimum G z →
      MeasureEntropyLocation nu abar lam z ∧
      NonnegativeAlternatingContacts G z (geometryRho G z.1 z.2).toReal ∧
      geometryRho G z.1 z.2 < sInf (fixedExponentRates G 0) ∧
      geometryRho G z.1 z.2 < sInf (fixedExponentRates G 1) ∧
      ∀ N : ℕ, 1 ≤ N → ∀ w : ℝ × ℝ, FeasiblePair w →
        profileOmega (measureEntropyProfile nu lam z.1) z.2 N ≤
          profileOmega (measureEntropyProfile nu lam w.1) w.2 N ∧
        (profileOmega (measureEntropyProfile nu lam w.1) w.2 N =
          profileOmega (measureEntropyProfile nu lam z.1) z.2 N → w = z)) ∧
    (∀ (z : ℝ × ℝ) (r : ℝ), FeasiblePair z → 0 < r →
      UniformGeometryEnvelope G z r → NonnegativeAlternatingContacts G z r →
      GeometryOptimum G z ∧ geometryRho G z.1 z.2 = ENNReal.ofReal r ∧
      ∀ w, GeometryOptimum G w → w = z) ∧
    (∀ a : ℝ, 0 < a → a < 1 →
      (∀ t ∈ Set.Ioo (-1 : ℝ) 1,
        measureEntropyLoss nu t = normalizedEntropy (Real.sqrt a * t)) →
      ∀ z, GeometryOptimum G z → a < z.2 ∧ z.2 < a / (1 - a) ∧
        ∃ t2 t3 : ℝ, 0 < t2 ∧ t2 < t3 ∧ t3 < 1 ∧
          geometryError G z.1 z.2 0 = (geometryRho G z.1 z.2).toReal ∧
          geometryError G z.1 z.2 t2 = -(geometryRho G z.1 z.2).toReal ∧
          geometryError G z.1 z.2 t3 = (geometryRho G z.1 z.2).toReal)

end PGDRevision20261113
Source
Author manuscript, Exact Residual Bounds and Optimal Geometry Selection for Proximal Gradient Methods, revised reading edition, 13 November 2026; Sections 4--8 and Appendix B, with exact source references on each associated theorem.

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