Lemma 3: For t>0 and alpha>0,alpha!=1, phi_alpha(t)=psi(t)-psi(t^(1-alpha))/(1-alpha)
ProvedQAlgorithms.Papers.Independent.Intakenext100.P28.phiAlpha_identityprove2quantum
For and , , define off 1 and . With defined by (536), continuously extended by ,
Only the second term is divided by , including at .
Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P28
Formal statement
noncomputable section
open MeasureTheory Set Filter
open scoped ENNReal NNReal Topology BigOperators ProbabilityTheory
namespace QAlgorithms.Papers.Independent.Intakenext100.P28
universe u
attribute [local instance] Classical.propDecidable
/-- Lemma 3: For t>0 and alpha>0,alpha!=1, phi_alpha(t)=scalarLogSecant(t)-scalarLogSecant(t^(1-alpha))/(1-alpha). Sason--Verdu, arXiv:1508.00335v7, PDF p.76. -/
theorem phiAlpha_identity
(a t : ℝ) (ha : 0 < a ∧ a ≠ 1) (ht : 0 < t) :
phiAlpha a t = scalarLogSecant t - scalarLogSecant (Real.rpow t (1 - a)) / (1 - a) := by sorry
end QAlgorithms.Papers.Independent.Intakenext100.P28
end
Source
-divergence Inequalities, arXiv:1508.00335v7, PDF p. 76, Lemma 3
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
Let be real numbers satisfying , , and . For positive real , define
where the powers are real powers. Then
This includes , using the separately assigned values of both functions, and includes both and .
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.