Corrected quantitative private-bit filtering and distillation bound
OpenQAlgorithms.Papers.Independent.Gap48.P42.private_bit_rateConsider the preceding private-state construction with exactly two parties, , arbitrary positive finite local shields, normalized shield state and joint-shield unitaries . For normalized local product vectors and , define
The maximum is attained. For every maximizing pair, put and . Then and . The balanced local success filters are contractions. Their joint unnormalized output equals , with trace , where are the projectors onto , and
Here , , and with zero terms defined by continuity. The balanced filters use the same maximizing vectors and remove the phase of their cross matrix element. The yield is per original input copy, so it includes the success probability. This makes the strict positivity quantitative.
Source correction. Eq. (2.4) prints ; its two filtering cases instead give .
import Definitions.Def_QAlgorithms_Papers_Independent_Gap48_P42
namespace QAlgorithms.Papers.Independent.Gap48.P42
open scoped BigOperators Matrix ComplexConjugate ComplexOrder ENNReal
noncomputable section
attribute [local instance] Classical.propDecidable Classical.decEq
/-- PDF p.2, Eqs.(2.2)--(2.4), note [11] p.5. Source correction:
printed a_max=max(a_1,a_2) is changed to min(a_1,a_2), since the two
filtering cases succeed with the smaller weight. The Bell parameter remains
1/2+eta/(2 sqrt(a_1*a_2)). Maximizers exist and no a_s is an input parameter. -/
theorem private_bit_rate (S : Fin 2 → ℕ) (hS : ∀ k, 0 < S k)
(rho : Matrix (PartyBasis S) (PartyBasis S) ℂ) (hrho : IsDensityMatrix rho)
(U : Fin 2 → Matrix.unitaryGroup (PartyBasis S) ℂ) :
(∃ f g : ProductVectors S, Maximizes (shieldBlock rho U 0 1) f g) ∧
∀ f g : ProductVectors S, Maximizes (shieldBlock rho U 0 1) f g →
PairConclusion rho U 0 1 f g := by sorry
end
end QAlgorithms.Papers.Independent.Gap48.P42
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
Let be positive integers, let , let be a positive semidefinite matrix of trace on , and let be arbitrary unitaries on that space. Write , and, for families and , write , . Define . There exist unit factor families attaining this supremum, and every such pair has all the following properties. Set , , , , , and , where . Then , , , and . Identify each local basis of size with a key bit and a shield basis element using the fixed cardinality bijection , and define by its key blocks when both parties have key in the row and key in the column, and by zero for all other key patterns. Define and , with these bras interpreted through the same bijections. Both are positive semidefinite, and, with and , one has and . Finally, , where viewed in , and is the supremum of over nonnegative real rates such that, for every , all sufficiently large positive integers admit an integer and a density matrix obtainable deterministically from , with and . Tensor copies here are grouped by party using the fixed cardinality bijections, and . The deterministic conversions are exactly those represented by finite trees starting with the unchanged state, applying at each node an instrument on one party with finitely many Kraus operators, nonnegative branch probabilities summing to , normalized branch states satisfying , and , followed by branch-dependent trees; one may also forget an output ensemble by replacing it with its weighted sum when its local dimensions agree. All states in these trees have positive local dimensions, and deterministic output means the final list is . Local shield dimensions equal to are included.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.