Convexity under the fractional-linear perspective
ProvedQAlgorithms.Papers.Independent.Bridge50.P43.fractional_linear_perspective_convexprove2quantum
Let be convex. Define by
Then is convex on . The input to is always greater than one, and is excluded. This is the perspective transformation used for the order variable in (4.29)–(4.31).
Formalization Note A total real function is constrained only on , so its other values are immaterial.
Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Bridge50_P43
Formal statement
noncomputable section
open scoped Matrix ComplexConjugate ComplexOrder Topology
open Filter
namespace QAlgorithms.Papers.Independent.Bridge50.P43
attribute [local instance] Classical.propDecidable
attribute [local instance] Classical.decEq
open QAlgorithms.Papers.Independent.Dependency50.P09 (Divergence relativeD petzD sandwichedD)
open QAlgorithms.Papers.Independent.Information100.P71
(SupportIncluded supportPower AdaptiveTest adaptiveInput adaptiveAccept adaptiveBeta channelPower)
open QAlgorithms.Papers.Independent.Information100.P58 (testingD)
-- Exactly the trace-preparation map of PDF pp.5--6, on every operator.
open QAlgorithms.Papers.Independent.Information100.P69 (replacer)
/-- Lemma 13, arXiv:1408.3373v3, PDF pp.19--20.
SOURCE CORRECTION: printed f:(0,1)->R is undefined at 1/(1-u)>1.
Use convexity on (1,infinity), the order domain of F in (4.29)--(4.31).
ConvexOn of a total real function only uses its restriction to that domain. -/
theorem fractional_linear_perspective_convex
(f : ℝ → ℝ) (hf : ConvexOn ℝ (Set.Ioi 1) f) :
ConvexOn ℝ (Set.Ioo 0 1) (fun u => (1-u) * f (1/(1-u))) := by sorry
end QAlgorithms.Papers.Independent.Bridge50.P43
end
Source
Strong converse exponents for a quantum channel discrimination problem and quantum-feedback-assisted communication, arXiv:1408.3373v3, PDF p. 19, Lemma 13, corrected domain
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
For every function that is convex on , the function
is convex on . Thus for all and all nonnegative real weights with , . The function is defined on all real numbers, but its assumed convexity concerns only arguments strictly greater than ; the asserted convexity of excludes both endpoints and .
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.