Theorem 8 — heterodyne tomography under a vacuum gap with copies
ProvedBosonicLearn.theorem_8bosonic-gaussian-statesp2o-gran-per-chapterp2o-plan-paperp2o-v1quantum-informationquantum-learning
There is a universal constant such that the following holds. For every , every and every there are
and a non-adaptive protocol that measures copies by independent heterodyne detections and always outputs a valid Gaussian state , such that for every -mode Gaussian state with ,
For warm states () this is the classical rate .
Differences from the source:
- encoding: O(·) and "there is a … protocol" → ∃ C > 0 before every other variable; N and the estimator chosen before the state; reason: the constant is universal and the protocol must work for every state of the class
- encoding: a protocol "using only independent heterodyne measurements" → a measurable deterministic estimator of N i.i.d. heterodyne outcomes; reason: the paper's protocol; a deterministic estimator is the stronger existence claim
Preamble
import Mathlib import Definitions.Def_BosonicLearn_heterodyneSuccess import Definitions.Def_BosonicLearn_promiseClass open QAlgorithms.Papers.Independent.Dependency50.P33 (State traceSize) open QAlgorithms.Papers.Independent.Claimed50.P47 (Occupation) open MeasureTheory ProbabilityTheory open scoped ENNReal Matrix
Formal statement
namespace BosonicLearn
/-- **Theorem 8 (Heterodyne tomography under a vacuum gap).**
Chen, Anna Mele, Anna Mele, Preskill, *When are bosonic Gaussian states classical to learn?*, arXiv:2609.26705, Theorem 8, Eqs. (205)–(206), p. 61: "Let ρ(µ, Σ) be an n-mode Gaussian state satisfying Σ ⪰ (1/2 + ν)I
for some ν > 0. For every ϵ, δ ∈ (0, 1), there is a non-adaptive protocol using only independent heterodyne
measurements that always outputs a valid Gaussian state ρ̂ and, with probability at least 1 − δ, satisfies
½‖ρ̂ − ρ(µ, Σ)‖₁ ≤ ϵ using N = O((1 + 1/ν) n(n + log(1/δ))/ϵ²) copies."
There is a universal `C > 0` such that for every `n`, every `ν > 0` and every `ϵ, δ ∈ (0, 1)` there are
`N ≤ C(1 + 1/ν) n(n + log(1/δ))/ϵ²` and a measurable estimator `est` of `N` heterodyne outcomes, every value of
which is a Gaussian state, such that every Gaussian state with `Σ ⪰ (½ + ν)I` is learned to trace distance `ϵ`
with probability at least `1 − δ`.
**Formalization Note.** The protocol (`N` and `est`) is chosen before the state; `O(·)` is one constant uniform in
`n, ν, ϵ, δ`. The estimator is deterministic; that is the paper's protocol. -/
theorem theorem_8 :
∃ C : ℝ, 0 < C ∧ ∀ (n : ℕ) (ν : ℝ), 0 < ν → ∀ ϵ δ : ℝ, 0 < ϵ → ϵ < 1 → 0 < δ → δ < 1 →
∃ N : ℕ, (N : ℝ) ≤ C * ((1 + 1 / ν) * ((n : ℝ) * (n + Real.log (1 / δ))) / ϵ ^ 2) ∧
∃ est : (Fin N → EuclideanSpace ℝ (Fin n ⊕ Fin n)) → State (Occupation (Fin n)),
Measurable est ∧
(∀ y, ∃ (μ : Fin n ⊕ Fin n → ℝ) (S : Matrix (Fin n ⊕ Fin n) (Fin n ⊕ Fin n) ℝ),
IsGaussianState (est y) μ S) ∧
∀ ρ ∈ promiseClass n ν, ENNReal.ofReal (1 - δ) ≤ heterodyneSuccess N est ρ ϵ := by sorry
end BosonicLearn
Source
Chen, Anna Mele, Anna Mele, Preskill, When are bosonic Gaussian states classical to learn?, arXiv:2609.26705, Theorem 8, Eqs. (205)–(206), p. 61
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.