Triangle inequality for the trace distance:
ProvedBosonicLearn.traceDist_triangleThe trace distance satisfies the triangle inequality.
Let be an index set and coefficient arrays on . With the trace norm defined as the infimum of over nuclear expansions , (and if is not trace class),
Role. In the proof of Theorem 8 the error of the estimate is split through the intermediate state into a covariance part (Theorem 7) and a first-moment part (Lemma 18).
Formalization Note traceDist A B = traceSize (A − B) / 2 with values in ℝ≥0∞. No positivity or normalization of the arrays is assumed.
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_traceDist open QAlgorithms.Papers.Independent.Dependency50.P33 (State traceSize) open scoped ENNReal
namespace BosonicLearn
/-- **Triangle inequality for the trace distance.**
Chen, Anna Mele, Anna Mele, Preskill, *When are bosonic Gaussian states classical to learn?*, arXiv:2609.26705, §7.2, proof of Theorem 8, Eq. (218), p. 62: "Inserting the intermediate state ρ(µ̂, Σ) and using the
triangle inequality, we find ½‖ρ̂ − ρ(µ, Σ)‖₁ ≤ ½‖ρ(µ̂, Σ̃) − ρ(µ̂, Σ)‖₁ + ½‖ρ(µ̂, Σ) − ρ(µ, Σ)‖₁."
For coefficient arrays `A, B, C` on any index set, `½‖A − C‖₁ ≤ ½‖A − B‖₁ + ½‖B − C‖₁`.
**Formalization Note.** `traceDist A B = traceSize (A − B) / 2` with values in `[0, ∞]`; the trace norm is the
infimum over nuclear expansions, and is `∞` for arrays that are not trace class, so the inequality holds for
all arrays. -/
theorem traceDist_triangle {I : Type} (A B C : Matrix I I ℂ) :
traceDist A C ≤ traceDist A B + traceDist B C := by sorry
end BosonicLearn