Proposition 4: convexity
OpenQAlgorithms.Papers.Independent.Intakethird100.P22.nonlocality_convexAll input and output alphabets are finite and nonempty. Let be a normalized nonnegative bipartite non-signaling device, and let be its complete extension, with possibly input-dependent pure-member labels. All Shannon information quantities are in bits, with . Write for the maximum over honest settings of the infimum, over Eve's general measurements, of intrinsic information; intrinsic information itself minimizes classical conditional mutual information over arbitrary stochastic channels on Eve.
For two devices in the same fixed polytope and every , with the mixture computed pointwise,
The claim holds for arbitrary complete extensions of each indicated device and includes both endpoints.
Eve retains the dice result and extension output . Squashed CMI conditions on ; intrinsic-information channels may depend on both. This is the operational convention of XI.B.3 used by the proof, rather than discarding the dice result in display (7).
import Definitions.Def_QAlgorithms_Papers_Independent_Intakethird100_P22
noncomputable section
open scoped BigOperators
namespace QAlgorithms.Papers.Independent.Intakethird100.P22
attribute [local instance] Classical.propDecidable Classical.decEq
open QAlgorithms.Papers.Independent.Advance50.P47
(Box IsCorrelation IsNoSignaling HasLocalModel correlationMixture prBox)
open QAlgorithms.Papers.Independent.Information100.P56 (IsLaw)
open QAlgorithms.Papers.Independent.Information100.P27 (shannon)
open QAlgorithms.Papers.Independent.Gap48.P25 (classicalMI)
section Devices
variable {X Y A B : Type} [Fintype X] [Fintype Y] [Fintype A] [Fintype B]
[Nonempty X] [Nonempty Y] [Nonempty A] [Nonempty B]
/-- PDF p.46, Proposition 4, (222): convexity, including both endpoints.
The CE arguments are arbitrary complete extensions of the specified devices. Source convention: Eve keeps the dice
result (p.22, XI.B.3); CE members may depend on (z,e) -/
theorem nonlocality_convex (p q : Box X Y A B) (t : ℝ) (ht : 0 ≤ t) (htOne : t ≤ 1)
(cp : CompleteExtension p) (cq : CompleteExtension q)
(cm : CompleteExtension (correlationMixture p q t)) :
squashedNonlocality cm ≤ t * squashedNonlocality cp + (1-t) * squashedNonlocality cq := by sorry
end Devices
end QAlgorithms.Papers.Independent.Intakethird100.P22
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
For finite nonempty alphabets , any real arrays , any real with , and any complete extensions of , respectively, where , the assertion is
Here a complete extension supplies positive finite indices , probability weights , and common nonnegative normalized no-signaling boxes reconstructing its box for every . Every positive-weight support consists of vertices and has no proper subset supporting another finite valid decomposition, and every such minimal vertex decomposition has a unique with the same total weight at every box. For each such extension,
with , marginals formed by summation, and . The endpoints and are included, and the three complete extensions are quantified independently.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.