Proposition 1 (Helstrom) —
ProvedFuchsVdGraaf.proposition_1information-theoryp2o-gran-per-chapterp2o-plan-paperp2o-v1quantum-information
Let be two arbitrary density matrices and let be the eigenvalues (with multiplicity) of the Hermitian matrix . Then the quantum probability of error, minimised over all POVMs, is
Both equalities are asserted. This is Helstrom's formula: the optimal error of discriminating two equiprobable states is a function of their trace distance.
Differences from the source:
- reading: → infimum over the set of values at all POVMs; reason: the paper's own "more carefully, one should use infimum" (p.10).
- encoding: POVMs and →
E : Fin m → Matrix n n ℂfor everym : ℕand the real part of the trace; reason: Lean convention, covers every finite POVM. - encoding: and the eigenvalues →
WildeQIT.traceNorm (ρ0 - ρ1)andIsHermitian.eigenvalues; reason: for Hermitian , ; same values.
Preamble
import Mathlib import Definitions.Def_WildeQIT_IsDensityOperator import Definitions.Def_WildeQIT_traceNorm import Definitions.Def_FuchsVdGraaf_qErrorProb open scoped ComplexOrder
Formal statement
namespace FuchsVdGraaf
/-- Fuchs and van de Graaf, arXiv:quant-ph/9712042v2, Proposition 1, eq. (17), p.11: for two
arbitrary density matrices `ρ₀`, `ρ₁`, the (quantum) probability of error equals
`PE(ρ₀, ρ₁) = ½ − ¼ ∑_{j=1}^{N} |λ_j| = ½ − ¼ Tr|ρ₀ − ρ₁|`, where the `λ_j` are the eigenvalues
of `ρ₀ − ρ₁`.
**Formalization Note.** `PE(ρ₀, ρ₁)` is `qErrorProb`, the infimum over all POVMs with any finite
number of outcomes. The eigenvalues are those of the Hermitian matrix `ρ₀ − ρ₁`
(`Matrix.IsHermitian.eigenvalues`, listed with multiplicity), the Hermitian proof being derived
from the two density-matrix hypotheses. `Tr|ρ₀ − ρ₁|` is the published `WildeQIT.traceNorm`. -/
theorem proposition_1 {n : Type} [Fintype n] [DecidableEq n] (ρ0 ρ1 : Matrix n n ℂ)
(h0 : WildeQIT.IsDensityOperator ρ0) (h1 : WildeQIT.IsDensityOperator ρ1) :
qErrorProb ρ0 ρ1 =
1 / 2 - 1 / 4 * ∑ j, |(h0.1.isHermitian.sub h1.1.isHermitian).eigenvalues j| ∧
qErrorProb ρ0 ρ1 = 1 / 2 - 1 / 4 * WildeQIT.traceNorm (ρ0 - ρ1) := by sorry
end FuchsVdGraaf
Source
Fuchs and van de Graaf, Cryptographic Distinguishability Measures for Quantum Mechanical States, arXiv:quant-ph/9712042v2, Proposition 1, eq. (17), p.11
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.