§2.2.4 — an OSR is the unitary evolution iff with
ProvedBaconThesis.OSRUnitary.osr_unitary_iffLet be a finite, nonempty index set, a unitary matrix, and a finite family of complex matrices (the OSR operators). The following are equivalent:
- the operator-sum representation of the is the unitary evolution by and is normalized, that is,
- every OSR operator is a scalar multiple of : there are with
This is the claim of §2.2.4 of the thesis, eqs. (2.27)–(2.28): it answers when an OSR, which in general describes decoherence, is in fact a unitary evolution, and it is the prototype of the decoherence-free conditions of the later chapters.
Formalization Note The system is with a nonempty finite type, and the OSR index is any finite type, possibly empty. Unitarity of is Mathlib's Matrix.unitaryGroup. Density matrices are the published WildeQIT.IsDensityOperator; the OSR and its normalization are the definitions osrApply and IsTracePreservingOSR.
Differences from the source:
- encoding: a system Hilbert space of any dimension → a finite nonempty index type d; reason: for d empty the left side holds with an empty OSR while no c has Σ|c_i|² = 1; the hypothesis excludes only the zero-dimensional space, which the book does not consider
- reading: (2.27) "for all ρ" → for every density matrix ρ; reason: the book's ρ are states; density matrices span all matrices and both sides are linear in ρ, so the readings are equivalent
- reading: "(2.27), where U†U = Σ_i A_i†A_i = I (2.28)" → U†U = I is a standing hypothesis and the left side is (2.27) together with Σ_i A_i†A_i = I; reason: the book's proof reads both directions as "Eq. (2.27) and Eq. (2.28) both hold"; under Σ_i A_i†A_i = I the left side is (2.27) alone
- encoding: U unitary (U†U = I of (2.28)) → U ∈ Matrix.unitaryGroup d ℂ; reason: in finite dimension U†U = I implies UU† = I; same instances
- encoding: c_i(t) → c : ι → ℂ; reason: the A_i are taken at one fixed time t, so the c_i are scalars
- generalization: A_i the OSR operators √p_ν⟨μ|U(t)|ν⟩ of (2.11) → any finite family A : ι → Matrix d d ℂ; reason: the claim and its proof use only (2.27)–(2.28); every OSR of (2.11) is such a family
import Mathlib import Definitions.Def_WildeQIT_IsDensityOperator import Definitions.Def_BaconThesis_OSRUnitary_osrApply import Definitions.Def_BaconThesis_OSRUnitary_IsTracePreservingOSR open scoped Matrix
namespace BaconThesis.OSRUnitary
/-- Bacon, *Decoherence, Control, and Symmetry in Quantum Computers*, arXiv:quant-ph/0305025v1,
§2.2.4 "OSR and unitary evolution", (2.27)–(2.28), p.19 (PDF p.36), proof to (2.35), p.20
(PDF p.37): for a unitary `U`, the OSR `{A_i}` satisfies (2.27) `∑_i A_i ρ A_i† = U ρ U†` for
all `ρ` together with (2.28) `∑_i A_i† A_i = I` if and only if `A_i = c_i U` for scalars `c_i`
with `∑_i |c_i|² = 1`.
**Formalization Note.** The system is `d → ℂ` with `d` a finite, **nonempty** type (for an
empty `d` the left side holds with an empty OSR while no `c` has `∑ |c_i|² = 1`). The OSR index
`ι` is any finite type (possibly empty). `U ∈ Matrix.unitaryGroup d ℂ` is the standing
hypothesis `U†U = I` of (2.28). "For all `ρ`" ranges over density operators
(`WildeQIT.IsDensityOperator`: positive semidefinite, trace one). The book's `c_i(t)` are
scalars at the fixed time of the OSR, so `c : ι → ℂ`. -/
theorem osr_unitary_iff {d ι : Type} [Fintype d] [DecidableEq d] [Nonempty d] [Fintype ι]
(A : ι → Matrix d d ℂ) (U : Matrix d d ℂ) (hU : U ∈ Matrix.unitaryGroup d ℂ) :
((∀ ρ : Matrix d d ℂ, WildeQIT.IsDensityOperator ρ → osrApply A ρ = U * ρ * Uᴴ) ∧
IsTracePreservingOSR A) ↔
∃ c : ι → ℂ, (∀ i, A i = c i • U) ∧ ∑ i, ‖c i‖ ^ 2 = 1 := by sorry
end BaconThesis.OSRUnitary
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.