Entrywise convergence of density operators implies trace-norm convergence
ProvedBosonicLearn.density_tendsto_of_entrywiseFor density operators, entrywise convergence in a fixed basis implies convergence in trace norm.
Let be an index set and let and be density operators on (trace class, positive semidefinite, trace one), given by their coefficient arrays in the standard basis. If
then .
The hypothesis that the limit has trace one cannot be dropped: no mass escapes to infinity. Quantitatively, for a density operator and a finite , the part of outside the block has trace norm at most .
Role. For the hard family of the single-copy lower bound, the Fock-basis entries are polynomial in the fugacity matrix, hence continuous in ; this lemma upgrades that to the trace-norm continuity used in the proof of Theorem 5 ("which can be shown by appropriate photon number cutoff"). It applies equally to any family of states whose matrix elements depend continuously on parameters.
Formalization Note Density operators are IsDensity (trace class in the nuclear-expansion sense, non-negative quadratic forms on finitely supported vectors, diagonal sum ); is traceSize with values in ℝ≥0∞; the conclusion is convergence to in ℝ≥0∞ along atTop. The index set is arbitrary.
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_QAlgorithms_Papers_Independent_Dependency50_P33 open QAlgorithms.Papers.Independent.Dependency50.P33 (IsDensity traceSize) open Filter Topology open scoped ENNReal
namespace BosonicLearn
/-- **Entrywise convergence of density operators implies convergence in trace norm.**
Chen, Anna Mele, Anna Mele, Preskill, *When are bosonic Gaussian states classical to learn?*, arXiv:2609.26705, proof of Theorem 5, p. 40: "the map X ↦ ρ_{R_X} is continuous in trace norm (which can be shown by
appropriate photon number cutoff)".
Let `X₀, X₁, X₂, …` and `X` be density operators on `ℓ²(I)` (trace class, positive semidefinite, trace one), given
by their coefficient arrays. If `(X_k)_ij → X_ij` for all `i, j`, then `‖X_k − X‖₁ → 0`.
**Formalization Note.** `‖·‖₁` is `traceSize` with values in `ℝ≥0∞`. Because all operators have trace one, no
mass can escape: for a finite set `F` of basis indices the part of a density operator outside the block `F × F`
has trace norm at most `τ + 2√τ`, `τ = 1 − ∑_{i ∈ F} X_ii`, and `τ` converges along the sequence. -/
theorem density_tendsto_of_entrywise {I : Type} (X : ℕ → Matrix I I ℂ) (X₀ : Matrix I I ℂ)
(hX : ∀ k, IsDensity (X k)) (hX₀ : IsDensity X₀)
(h : ∀ i j, Tendsto (fun k => X k i j) atTop (𝓝 (X₀ i j))) :
Tendsto (fun k => traceSize (X k - X₀)) atTop (𝓝 0) := by sorry
end BosonicLearn