Theorem 2.5 — is a quantum channel for a unit-vector Gram matrix
ProvedDistQuantumQuery.Supplement.theorem_2_5adversary-methodcomplexity-theoryp2o-gran-per-chapterp2o-plan-paperp2o-v1quantum-informationquantum-query-complexity
Let be a Gram matrix of unit vectors , i.e. ; equivalently, is positive semidefinite with every diagonal entry equal to . Then the map
is a quantum channel: it is linear, completely positive (for every finite-dimensional auxiliary space , maps positive semidefinite operators to positive semidefinite operators) and trace preserving ( for all ).
Hadamard products with Gram matrices are the channels through which the paper processes weighted Gram matrices of state families.
Differences from the source:
- reading: G a Gram matrix → G positive semidefinite with G i i = 1 for every i; reason: p.16 defines a Gram matrix (corresponding to unit vectors) as P = W†W with unit columns and shows this holds iff P ⪰ 0 has unit diagonal; without the unit diagonal X ↦ X ∘ G^⊤ is not trace preserving
- encoding: ℂ^n, n ∈ ℕ → ℂ^n for a finite index type n; reason: relabelling coordinates; Fin n is the source case
- encoding: quantum channel: linear, CPTP, CP for all finite-dimensional R (§2.3.1, p.15) → published WildeQIT.EntBreaking.IsQuantumChannel: a ℂ-linear map, id_{ℂ^r} ⊗ E positive for every r, Tr E(X) = Tr X for all X; reason: every finite-dimensional auxiliary space is isomorphic to some ℂ^r; same notion
Preamble
import Mathlib import Definitions.Def_WildeQIT_EntBreaking_IsQuantumChannel import Definitions.Def_DistQuantumQuery_Supplement_schurChannel open scoped Matrix ComplexOrder
Formal statement
namespace DistQuantumQuery.Supplement
/-- **Theorem 2.5** (Ben-David, Ebtehaj, *Distributional Quantum Query Complexity*,
arXiv:2610.06835, held version, sha256 6334ffaf…, Theorem 2.5, p.16). Let `G ∈ ℂ^{n×n}` be a Gram
matrix. Then the map `E_G : L(ℂ^n) → L(ℂ^n)` defined by `E_G(X) := X ∘ G^⊤` is a quantum channel.
**Formalization Note** "Gram matrix" is the page-16 notion, a Gram matrix corresponding to unit
vectors, which the page shows is the same as a positive semidefinite matrix with all diagonal
entries one; it is stated that way. A quantum channel is a linear, completely positive and trace
preserving map (§2.3.1, p.15), the published `WildeQIT.EntBreaking.IsQuantumChannel`. The index
set `[n]` is a finite type `n`. -/
theorem theorem_2_5 {n : Type} [Fintype n] [DecidableEq n] (G : Matrix n n ℂ)
(hG : G.PosSemidef) (hdiag : ∀ i, G i i = 1) :
WildeQIT.EntBreaking.IsQuantumChannel (schurChannel G) := by sorry
end DistQuantumQuery.Supplement
Source
Ben-David, Ebtehaj, Distributional Quantum Query Complexity, arXiv:2610.06835 (held version, sha256 6334ffaf…), §2.3.3, Theorem 2.5, p.16