An adaptive single-copy scheme is one POVM on with the transcript as outcome
ProvedBosonicLearn.adaptive_scheme_product_povmAn adaptive sequence of single-copy measurements on copies is one measurement on the -copy system.
Let be the -mode Fock space. An adaptive single-copy scheme on copies with outcomes in a measurable space assigns to every round and every record of earlier outcomes a POVM on with outcomes in , measured on a fresh copy, such that the outcome law of round depends measurably on . For density operators on let be the law on of the transcript when round measures : the first outcome has law , and given the -th outcome has law .
Then there is a POVM on with outcomes in such that for all density operators on and every measurable
For the right-hand side is the transcript law of the scheme on copies of .
Role. This is the precise sense in which single-copy (non-entangled) schemes are a subclass of collective measurements on . With BosonicLearn.fockPOVM_postprocess applied to the estimator it gives BosonicLearn.adaptive_scheme_collective.
Formalization Note The scheme is an AdaptiveScheme n N O; the right-hand side is sequentialLaw of the round kernels S.round τ_t t. The product state has entries on the index type . is a FockPOVM, so its Born law is a probability measure on every density operator of the -copy space, entangled ones included; the identity is asserted on product states. Non-identical factors are allowed because the statement composes round by round in that form.
Related platform material The space used here (Hilbert A) is the same Lean type as OAI.TensorLp.H A, and is the same type as OAI.EntropyPhotonNumber.Fock n. The published definitions OAI_Analysis_TensorLp (tensor products, slices and operator lifts on ) and OAI_MathematicalPhysics_Fock_Operators (OAI.CrossEnsemble.operator, the bounded operator of a family, with hasSum_operator and entry_hasSum) are sorry-free and can be imported. The proof of BosonicLearn.traceDist_le_one contains lemmas on nuclear expansions, their sums, and truncation to finite index sets.
import Mathlib import Definitions.Def_BosonicLearn_AdaptiveScheme import Definitions.Def_BosonicLearn_sequentialLaw open QAlgorithms.Papers.Independent.Dependency50.P33 (IsDensity) open QAlgorithms.Papers.Independent.Claimed50.P47 (Occupation) open MeasureTheory ProbabilityTheory open scoped ENNReal Matrix
namespace BosonicLearn
/-- **An adaptive single-copy scheme is one POVM on the `N` copies.**
Chen, Anna Mele, Anna Mele, Preskill, *When are bosonic Gaussian states classical to learn?*, arXiv:2609.26705, §5.1, p. 30: "A non-entangled adaptive learning protocol on N copies can be specified by a family of
sequences of POVMs {M_{x_1}}, …, {M^{(x_{<t})}_{x_t}}, … Here, the t-th POVM can depend on the first t − 1
measurement outcomes x_{<t}"; the transcript likelihood is the product over rounds, Eq. (63), p. 35.
For every adaptive single-copy scheme `S` on `N` copies of an `n`-mode system with outcomes in `O`, there is one POVM
`M` on the `N`-copy Fock space `ℓ²(ℕⁿ)^{⊗N}` with outcomes in `O^N` (the transcripts) such that, for all density
operators `τ₁, …, τ_N` on one copy, the Born law of `M` on the product state `τ₁ ⊗ ⋯ ⊗ τ_N` is the law of the
transcript obtained when round `t` measures `τ_t`: the first outcome has law `S.round τ₁`, and given the first
`t − 1` outcomes `x_{<t}` the `t`-th outcome has law `S.round τ_t` at `x_{<t}`.
**Formalization Note.** The product state has entries `∏ₜ τₜ(uₜ, vₜ)` on the index type `Fin N → ℕⁿ`; with all `τₜ`
equal to `ρ` it is `ρ^{⊗N}` and the right-hand side is `transcriptLaw S ρ`. `M` is a `FockPOVM`, so its Born law is
a probability measure on every density of the `N`-copy space, entangled ones included. Non-identical factors are
allowed because that is the form in which the statement composes round by round. -/
theorem adaptive_scheme_product_povm {n N : ℕ} {O : Type} [MeasurableSpace O] (S : AdaptiveScheme n N O) :
∃ M : FockPOVM (Fin N → Occupation (Fin n)) (Fin N → O),
∀ τ : Fin N → Matrix (Occupation (Fin n)) (Occupation (Fin n)) ℂ, (∀ t, IsDensity (τ t)) →
M.law (Matrix.of fun u v => ∏ t, τ t (u t) (v t)) =
sequentialLaw (fun t => S.round (τ t) t) N le_rfl := by sorry
end BosonicLearn