Data processing for a thresholded photon count:
ProvedBosonicLearn.fugacity_count_testData processing for the two-outcome measurement "rotate by a passive unitary, count the photons in a set of modes, compare with a threshold".
Let be fugacity matrices (positive semidefinite with positive definite), , a unitary matrix, a set of modes and . Let be the law of the photon count in the modes after the passive unitary of . Then
Role. Last step of the proof of Lemma 12 (Eqs. (147)–(148)): a gap between the two tail probabilities of the photon count lower-bounds the trace distance of the two states.
Formalization Note The right-hand side is the full trace norm traceSize (without the factor of the trace distance; the constant is immaterial for Lemma 12). is fugacityCoeff R; the rotated state is written fugacityCoeff (U * R * Uᴴ), which is .
import Mathlib import Definitions.Def_BosonicLearn_fugacityCoeff import Definitions.Def_QAlgorithms_Papers_Independent_Dependency50_P33 open QAlgorithms.Papers.Independent.Dependency50.P33 (traceSize) open QAlgorithms.Papers.Independent.Claimed50.P47 (Occupation) open scoped ENNReal Matrix ComplexOrder
namespace BosonicLearn
/-- **Data processing for a thresholded photon count.**
Chen, Anna Mele, Anna Mele, Preskill, *When are bosonic Gaussian states classical to learn?*, arXiv:2609.26705, §5.7, proof of Lemma 12, Eqs. (147)–(148): "Coarse-graining the photon count into the two outcomes {K̂ > τ} and {K̂ ≤ τ}
gives a classical total variation distance equal to the absolute value of the left-hand side of Eq. (147). By data
processing for trace distance …".
Let `R₁, R₂` be fugacity matrices (positive semidefinite, `1 − R_i` positive definite), `U` a unitary matrix, `A` a set
of modes and `τ` a threshold. For `i = 1, 2` let `p^{(i)}_k = ∑_{w : ∑_{j∈A} w_j = k} ⟨w|ρ_{U R_i U†}|w⟩` be the law of the
photon count in the modes `A` after the passive unitary of `U`. Then
`|∑_{k > τ} p^{(2)}_k − ∑_{k > τ} p^{(1)}_k| ≤ ‖ρ_{R₂} − ρ_{R₁}‖₁`.
**Formalization Note.** The right-hand side is the full trace norm `traceSize` (no factor `½`); the constant is
irrelevant for Lemma 12. `ρ_R` is `fugacityCoeff R`. -/
theorem fugacity_count_test {n : ℕ} (R₁ R₂ : Matrix (Fin n) (Fin n) ℂ) (h₁ : R₁.PosSemidef)
(h₁' : (1 - R₁).PosDef) (h₂ : R₂.PosSemidef) (h₂' : (1 - R₂).PosDef)
(U : Matrix (Fin n) (Fin n) ℂ) (hU : Uᴴ * U = 1) (A : Finset (Fin n)) (τ : ℕ) :
ENNReal.ofReal |(∑' k : ℕ, if τ < k then (∑' w : Occupation (Fin n),
if ∑ i ∈ A, w i = k then (fugacityCoeff (U * R₂ * Uᴴ) w w).re else 0) else 0) -
∑' k : ℕ, if τ < k then (∑' w : Occupation (Fin n),
if ∑ i ∈ A, w i = k then (fugacityCoeff (U * R₁ * Uᴴ) w w).re else 0) else 0| ≤
traceSize (fugacityCoeff R₂ - fugacityCoeff R₁) := by sorry
end BosonicLearn