Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

One-sided covariance perturbation for Σ⪰12I\Sigma\succeq\tfrac12IΣ⪰21​I: 12∥ρ(μ,Σ′)−ρ(μ,Σ)∥1≤2 Tr[Σ−1(Σ′−Σ)]\tfrac12\|\rho(\mu,\Sigma')-\rho(\mu,\Sigma)\|_1\le2\,\mathrm{Tr}[\Sigma^{-1}(\Sigma'-\Sigma)]21​∥ρ(μ,Σ′)−ρ(μ,Σ)∥1​≤2Tr[Σ−1(Σ′−Σ)]

Proved
BosonicLearn.cov_perturbation_classical

by Senrui Chen · Nov 13, 2026 · Mathlib 0df444a (Lean v4.33.1)

bosonic-gaussian-statesquantum-informationquantum-learning

An energy-independent trace-distance bound between two Gaussian states with the same mean and ordered covariances, proved without fidelity formulas.

Let ρ(μ,Σ)\rho(\mu,\Sigma)ρ(μ,Σ) and ρ(μ,Σ′)\rho(\mu,\Sigma')ρ(μ,Σ′) be nnn-mode Gaussian states with Σ⪰121\Sigma\succeq\tfrac12\mathbf 1Σ⪰21​1 and Σ′⪰Σ\Sigma'\succeq\SigmaΣ′⪰Σ. Then

12∥ρ(μ,Σ′)−ρ(μ,Σ)∥1 ≤ 2 Tr⁡[Σ−1(Σ′−Σ)].\tfrac12\big\|\rho(\mu,\Sigma')-\rho(\mu,\Sigma)\big\|_1\ \le\ 2\,\operatorname{Tr}\big[\Sigma^{-1}(\Sigma'-\Sigma)\big].21​​ρ(μ,Σ′)−ρ(μ,Σ)​1​ ≤ 2Tr[Σ−1(Σ′−Σ)].

Comparison with BosonicLearn.bmem25_theorem_8_cov (the covariance part of [BMEM25, Theorem 8], with constant 1+38\frac{1+\sqrt3}881+3​​ and Σ−1+Σ′−1⪯2Σ−1\Sigma^{-1}+\Sigma'^{-1}\preceq2\Sigma^{-1}Σ−1+Σ′−1⪯2Σ−1): same form, constant 222 instead of 1+34\frac{1+\sqrt3}441+3​​, under the extra hypothesis Σ⪰121\Sigma\succeq\tfrac12\mathbf 1Σ⪰21​1 — which is the promise of [BMEM25, Theorem 10] as used in Corollary 19. The bound is linear in the relative perturbation and does not depend on the energy of the states.

Proof idea. ρ(μ,Σ)\rho(\mu,\Sigma)ρ(μ,Σ) is a mixture of coherent states over N(μ,Σ−12)\mathcal N(\mu,\Sigma-\tfrac12)N(μ,Σ−21​). The extra noise x∼N(0,Σ′−Σ)x\sim\mathcal N(0,\Sigma'-\Sigma)x∼N(0,Σ′−Σ) is split as (Σ−12)Σ−1x(\Sigma-\tfrac12)\Sigma^{-1}x(Σ−21​)Σ−1x, absorbed in the mixing measure through a Cameron–Martin weight, plus 12Σ−1x\tfrac12\Sigma^{-1}x21​Σ−1x, a displacement of the coherent states; by the symmetry x↦−xx\mapsto-xx↦−x and the even/odd decomposition of coherent vectors the first-order terms cancel, leaving a second-order weight ≲xTΣ−1x\lesssim x^T\Sigma^{-1}x≲xTΣ−1x, whose expectation is Tr⁡[Σ−1(Σ′−Σ)]\operatorname{Tr}[\Sigma^{-1}(\Sigma'-\Sigma)]Tr[Σ−1(Σ′−Σ)]. Checked numerically for one-mode thermal states.

Formalization Note Hypotheses (S - (1/2) • 1).PosSemidef and (S' - S).PosSemidef; the bound is ENNReal.ofReal (2 * (S⁻¹ * (S' - S)).trace).

Preamble
import Mathlib
import Definitions.Def_BosonicLearn_traceDist
import Definitions.Def_BosonicLearn_IsGaussianState

open QAlgorithms.Papers.Independent.Dependency50.P33 (State traceSize)
open QAlgorithms.Papers.Independent.Claimed50.P47 (Occupation)
open MeasureTheory
open scoped ENNReal Matrix
Formal statement
namespace BosonicLearn

/-- **One-sided covariance perturbation for classical covariances: `½‖ρ(µ, Σ′) − ρ(µ, Σ)‖₁ ≤ 2 Tr[Σ⁻¹(Σ′ − Σ)]`.**
Chen, Anna Mele, Anna Mele, Preskill, *When are bosonic Gaussian states classical to learn?*, arXiv:2609.26705, §1.4, p. 10 (trace-distance bounds between Gaussian states [BMEM25]); proof of Corollary 19, p. 65 ([BMEM25, Theorem 10]);
[BMEM25] = Bittel, Mele, Eisert, Mele, arXiv:2508.14979, Theorem 8.

Let `ρ(µ, Σ)` and `ρ(µ, Σ′)` be Gaussian states with the same mean, `Σ ⪰ ½I` and `Σ′ ⪰ Σ`. Then
`½‖ρ(µ, Σ′) − ρ(µ, Σ)‖₁ ≤ 2 Tr[Σ⁻¹(Σ′ − Σ)]`.

**Formalization Note.** This is the form of `bmem25_theorem_8_cov` needed for [BMEM25, Theorem 10] under the promise
`Σ ⪰ ½I`: the bound is linear in the relative perturbation and independent of the energy. The constant `2` replaces
`(1 + √3)/4` (and `Σ⁻¹ + Σ′⁻¹ ⪯ 2Σ⁻¹`). It is proved without fidelity formulas: `ρ(µ, Σ)` is a mixture of
coherent states over `N(µ, Σ − ½I)`, the extra noise `x ∼ N(0, Σ′ − Σ)` is split as
`(Σ − ½I)Σ⁻¹x` (absorbed in the mixing measure through a Cameron–Martin weight) plus `½Σ⁻¹x` (a displacement of the
coherent states), and the first-order terms cancel by the symmetry `x ↦ −x`. -/
theorem cov_perturbation_classical {n : ℕ} (ρ ρ' : State (Occupation (Fin n)))
    (μ : Fin n ⊕ Fin n → ℝ) (S S' : Matrix (Fin n ⊕ Fin n) (Fin n ⊕ Fin n) ℝ)
    (hS : (S - (1 / 2 : ℝ) • (1 : Matrix (Fin n ⊕ Fin n) (Fin n ⊕ Fin n) ℝ)).PosSemidef)
    (hord : (S' - S).PosSemidef)
    (h : IsGaussianState ρ μ S) (h' : IsGaussianState ρ' μ S') :
    traceDist ρ'.coefficients ρ.coefficients ≤
      ENNReal.ofReal (2 * (S⁻¹ * (S' - S)).trace) := by sorry

end BosonicLearn
Source
Chen, Anna Mele, Anna Mele, Preskill, When are bosonic Gaussian states classical to learn?, arXiv:2609.26705, §1.4, p. 10 (trace-distance bounds between Gaussian states [BMEM25]); proof of Corollary 19, p. 65 ([BMEM25, Theorem 10]); [BMEM25] = arXiv:2508.14979, Theorem 8

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me