Effects of a POVM are weakly -additive: quadratic forms and matrix entries are given by finite measures
ProvedBosonicLearn.fockPOVM_form_measureThe effects of a POVM are weakly -additive: their quadratic forms on finitely supported vectors are finite measures, and their matrix entries are complex combinations of four finite measures.
Let be an index set and let be a POVM on with outcomes in a measurable space , with effects for measurable .
- For every finite set and every there is a finite measure on such that for all measurable
- For all there are four finite measures on such that for all measurable
Role. The definition of a POVM used in this mission lists bounds on the effects, normalization and the Born rule, but no additivity of . This statement supplies it. Consequences used elsewhere are: entrywise; monotonicity of the quadratic forms in ; measurability of for a monotone family of sets ; and the possibility of integrating a bounded function against the entries of . It is a foundation for BosonicLearn.fockPOVM_postprocess and BosonicLearn.adaptive_scheme_product_povm.
Formalization Note is a FockPOVM. Measures take values in and enters the identities through ENNReal.toReal. The four measures of clause 2 are indexed by Fin 4.
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. Proved lemmas of this mission that can be imported: BosonicLearn.traceDist_le_one (nuclear expansions, truncation to finite index sets) and BosonicLearn.effect_matrix_operator (effect matrices as bounded positive operators).
import Mathlib import Definitions.Def_BosonicLearn_FockPOVM open MeasureTheory open scoped ENNReal
namespace BosonicLearn
/-- **The effects of a POVM are weakly σ-additive: quadratic forms and matrix entries are given by finite
measures.**
Chen, Anna Mele, Anna Mele, Preskill, *When are bosonic Gaussian states classical to learn?*, arXiv:2609.26705, §1.1, p. 5 ("general POVM measurements on n-mode bosonic systems"); continuous-outcome POVMs are used in
§5.4, pp. 34–35 (`∫ dz_t |ϕ⟩⟨ϕ| = P_k`). This makes precise the remark in the definition `BosonicLearn_FockPOVM`
that "the effects are weakly σ-additive: this is the usual definition of a POVM".
Let `P` be a POVM on `ℓ²(I)` with outcomes in a measurable space `Ω` and effects `E(B)`.
1. For every finite `s ⊆ I` and every `z`, there is a finite measure `m` on `Ω` with
`∑_{i,j ∈ s} conj(z_i) E(B)_{ij} z_j = m(B)` for all measurable `B`.
2. For all `i, j` there are four finite measures `m₀, m₁, m₂, m₃` on `Ω` with
`E(B)_{ij} = ¼ [(m₀(B) − m₁(B)) − i (m₂(B) − m₃(B))]` for all measurable `B`.
**Formalization Note.** `P` is a `FockPOVM`, whose fields do not include additivity of the effects; additivity is
derived here from the Born rule on rank-one densities with finite support. Measures take values in `ℝ≥0∞`, and
`m(B)` enters the identities through `ENNReal.toReal`. Clause 2 is clause 1 for the four vectors
`e_i + c e_j`, `c ∈ {1, −1, i, −i}`, by polarization. -/
theorem fockPOVM_form_measure {I Ω : Type} [DecidableEq I] [MeasurableSpace Ω] (P : FockPOVM I Ω) :
(∀ (s : Finset I) (z : I → ℂ), ∃ m : Measure Ω, IsFiniteMeasure m ∧ ∀ B, MeasurableSet B →
∑ i ∈ s, ∑ j ∈ s, star (z i) * P.effect B i j * z j = ((m B).toReal : ℂ)) ∧
(∀ i j : I, ∃ m : Fin 4 → Measure Ω, (∀ k, IsFiniteMeasure (m k)) ∧ ∀ B, MeasurableSet B →
P.effect B i j = (1 / 4 : ℂ) * ((((m 0 B).toReal - (m 1 B).toReal : ℝ) : ℂ) -
Complex.I * (((m 2 B).toReal - (m 3 B).toReal : ℝ) : ℂ))) := by sorry
end BosonicLearn