§6, unnumbered: PQCs and source randomization are the same model
ProvedQAlgorithms.Papers.Independent.Advance50.P18.pqc_randomization_equivalenceFix a source of unit -qubit vectors, a finite classical key law , full-space keyed unitaries , a fixed mixed ancilla , and a fixed full-output density . These data form a private quantum channel if and only if applying the selected unitary and forgetting only its classical label randomizes the source to . Both conditions mean exactly
All density and normalization requirements are included in each predicate. The ancillary environment remains in the output. The entropy cost is the Shannon entropy of the same law . This identifies the two models without introducing a free discarded environment.
import Definitions.Def_QAlgorithms_Papers_Independent_Advance50_P18
namespace QAlgorithms.Papers.Independent.Advance50.P18
open scoped ComplexConjugate
noncomputable section
variable {K I A : Type*} [Fintype K] [DecidableEq K]
[Fintype I] [DecidableEq I] [Fintype A] [DecidableEq A]
/-- PDF p.8, §6: PQC and randomization are the same exact full-output condition,
with the same source, law, unitaries, mixed ancilla and fixed output. -/
theorem pqc_randomization_equivalence (n a N : ℕ) (S : Set ((EuclideanSpace ℂ (Qubits n))))
(E : KeyedUnitaryEnsemble (Fin N) (Qubits n × Qubits a))
(anc : (Matrix (Qubits a) (Qubits a) ℂ)) (rho0 : (Matrix (Qubits n × Qubits a) (Qubits n × Qubits a) ℂ)) :
PQC S E anc rho0 ↔ ExactSourceRandomization S E anc rho0 := by sorry
end
end QAlgorithms.Papers.Independent.Advance50.P18
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
For every choice of natural numbers , let , let , let an ensemble consist of arbitrary real weights and complex matrices on , indexed by , and let and be arbitrary complex matrices indexed respectively by and . The private-channel condition is equivalent to the exact-source-randomization condition, with each condition meaning precisely that , , every is unitary, both and are positive semidefinite with trace , and every has norm and satisfies
where has entries . There are no hypotheses requiring either condition to hold. If is empty, the requirements concerning its vectors are vacuous, while the requirements on the weights, matrices, and density matrices remain. If , both conditions are false because the weights cannot sum to . The cases and are included, with a singleton.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.