Private-dit entanglement cost is at least the key size
OpenQAlgorithms.Papers.Independent.Gap48.P42.private_cost_lower_boundLet be positive integers, let be a positive semidefinite trace-one shield state on , and let each be a unitary on that joint shield. Write
with the bipartition and local shield dimensions . All logarithms and resource rates are in bits.
Let be operational entanglement cost in singlets per input copy, allowing trace-preserving adaptive LOCC, arbitrary local ancillary systems and vanishing preparation error. Then
The statement includes . It uses the accepted operational cost, without assuming the formation lower bound for tensor powers or attainment of an optimal cost.
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.4, Property 1a: operational singlet cost is at least log_2 d.
Uses the accepted Hayden--Horodecki--Terhal cost, not a prescribed scalar or
an assumed formation bound for tensor powers. The case d=1 is included. -/
theorem private_cost_lower_bound {d a b : ℕ}
(hd : 0 < d) (ha : 0 < a) (hb : 0 < b)
(rho : Matrix (Fin a × Fin b) (Fin a × Fin b) ℂ) (hrho : IsDensityMatrix rho)
(U : Fin d → Matrix.unitaryGroup (Fin a × Fin b) ℂ) :
Real.logb 2 d ≤ Dependency50.P42.entanglementCost Dependency50.P42.paperBures
(bipartitePrivate rho U) := by sorry
end
end QAlgorithms.Papers.Independent.Gap48.P42
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
For all positive natural numbers , every positive semidefinite trace-one matrix on , and every family of unitaries indexed by , form on local dimensions by splitting local indices with the fixed bijections and : its block between equal row keys on both parties and equal column keys on both parties is , and all other key blocks vanish. Then . Here is the real infimum of all real numbers such that, for every , there exist natural numbers with and a finite local-operations map from local dimensions to satisfying and . The state has , all tensor copies are grouped by party through the fixed cardinality bijections, and , where , with matrix square roots evaluated by continuous functional calculus and the real square root taken as a total function. The allowed maps are exactly those built by finitely many applications of identity maps and the recursion , where and every is itself allowed, or the corresponding recursion with for the other party. Rectangular Kraus operators and branch-dependent continuations are allowed. The block sizes may depend on ; the definition requires existence of a block and does not require a construction at every sufficiently large block size. Zero input singlets are allowed, giving a scalar input state, and gives the asserted lower bound .
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.