Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

The trace distance of two density operators is at most one: 12∥ρ−σ∥1≤1\tfrac12\|\rho-\sigma\|_1\le121​∥ρ−σ∥1​≤1

Proved
BosonicLearn.traceDist_le_one

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

bosonic-gaussian-statesquantum-informationquantum-learning

The trace distance between two density operators never exceeds one.

Let III be an arbitrary index set and let ρ\rhoρ and σ\sigmaσ be density operators on ℓ2(I)\ell^2(I)ℓ2(I), given by their matrices in the standard basis: each is positive, trace class and of trace one. Write ∥⋅∥1\|\cdot\|_1∥⋅∥1​ for the trace norm and Dtr(ρ,σ)=12∥ρ−σ∥1D_{\mathrm{tr}}(\rho,\sigma)=\tfrac12\|\rho-\sigma\|_1Dtr​(ρ,σ)=21​∥ρ−σ∥1​ for the trace distance. Then

Dtr(ρ,σ)≤1.D_{\mathrm{tr}}(\rho,\sigma)\le 1 .Dtr​(ρ,σ)≤1.

Role. An estimate of an unknown state to trace-distance accuracy ε≥1\varepsilon\ge1ε≥1 therefore needs no measurement: every fixed state is within distance 111 of the unknown one. The reduction of Corollary 19 uses this for ϵ≥1\epsilon\ge1ϵ≥1, where the claimed number of copies O(n2ϵ−2min⁡{n,1+1/ν})O\big(n^2\epsilon^{-2}\min\{n,1+1/\nu\}\big)O(n2ϵ−2min{n,1+1/ν}) can be smaller than one. The bound is also what makes every success event {Dtr(ρ^,ρ)≤ε}\{D_{\mathrm{tr}}(\hat\rho,\rho)\le\varepsilon\}{Dtr​(ρ^​,ρ)≤ε} certain once ε≥1\varepsilon\ge1ε≥1.

Formalization Note Density operators are the elements of State I, and the trace norm is traceSize: the infimum of ∑k∥ℓk∥ ∥rk∥\sum_k\|\ell_k\|\,\|r_k\|∑k​∥ℓk​∥∥rk​∥ over all nuclear expansions Xij=∑kℓk(i) rk(j)‾X_{ij}=\sum_k\ell_k(i)\,\overline{r_k(j)}Xij​=∑k​ℓk​(i)rk​(j)​ with ℓk,rk∈ℓ2(I)\ell_k,r_k\in\ell^2(I)ℓk​,rk​∈ℓ2(I), a value in [0,∞][0,\infty][0,∞] that is ∞\infty∞ when no expansion exists. The statement therefore contains that ρ−σ\rho-\sigmaρ−σ is trace class and that its nuclear expansions have total weight arbitrarily close to at most 222. The source defines the trace distance but does not state this bound separately.

Preamble
import Mathlib
import Definitions.Def_BosonicLearn_traceDist
open QAlgorithms.Papers.Independent.Dependency50.P33 (State traceSize)

open scoped ENNReal
Formal statement
namespace BosonicLearn

/-- **The trace distance of two density operators is at most one.**
Chen, Anna Mele, Anna Mele, Preskill, *When are bosonic Gaussian states classical to learn?*, arXiv:2609.26705, footnote 1, p. 3: "We write D_tr(ρ, σ) := ½‖ρ − σ‖₁ = ½‖ρ − σ‖_tr, with both ‖·‖₁ and ‖·‖_tr denoting the
trace norm. … if two objects have ε distinguishability distance, the maximal probability that an observer can
distinguish them correctly is (1 + ε)/2."

For any two density operators `ρ, σ` on `ℓ²(I)`, `½‖ρ − σ‖₁ ≤ 1`.

**Formalization Note.** The source does not state the bound; it is the standard consequence of
`‖ρ − σ‖₁ ≤ ‖ρ‖₁ + ‖σ‖₁ = 2`, and it is implicit in the quoted success probability `(1 + ε)/2 ≤ 1`. Here `‖·‖₁` is
`traceSize`, the infimum of `∑ₖ ‖ℓₖ‖‖rₖ‖` over nuclear expansions, so the content is that a density operator has
nuclear expansions of total weight arbitrarily close to its trace. The index type `I` is arbitrary. -/
theorem traceDist_le_one {I : Type} (ρ σ : State I) :
    traceDist ρ.coefficients σ.coefficients ≤ 1 := by sorry

end BosonicLearn
Source
Chen, Anna Mele, Anna Mele, Preskill, When are bosonic Gaussian states classical to learn?, arXiv:2609.26705, footnote 1, p. 3 (definition D_tr(ρ, σ) := ½‖ρ − σ‖₁, with optimal distinguishing probability (1 + ε)/2); used in the reduction of Corollary 19, Eq. (221), p. 64

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