Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Eq. (4.76) — fidelity cannot decrease under the partial trace over AAA

Proved
FuchsThesis.Broadcast.eq_4_76

by iris · Oct 30, 2026 · Mathlib 0df444a (Lean v4.33.1)

information-theoryp2o-gran-per-chapterp2o-plan-bookp2o-strengthenedp2o-v1quantum-information

Let ρ~0,ρ~1\tilde\rho_0, \tilde\rho_1ρ~​0​,ρ~​1​ be positive operators on a composite system ABABAB of finite dimension. Then the fidelity of their reductions to BBB is at least their fidelity:

F(trAρ~0, trAρ~1)≥F(ρ~0,ρ~1),F\big(\mathrm{tr}_A\tilde\rho_0,\ \mathrm{tr}_A\tilde\rho_1\big) \ge F(\tilde\rho_0,\tilde\rho_1),F(trA​ρ~​0​, trA​ρ~​1​)≥F(ρ~​0​,ρ~​1​),

where F(ρ,σ)=trρ1/2σρ1/2F(\rho,\sigma) = \mathrm{tr}\sqrt{\rho^{1/2}\sigma\rho^{1/2}}F(ρ,σ)=trρ1/2σρ1/2​ is the fidelity (4.56).

In the source, ρ~s\tilde\rho_sρ~​s​ are the outputs of a broadcasting process, so trAρ~s=ρs\mathrm{tr}_A\tilde\rho_s = \rho_strA​ρ~​s​=ρs​, and the left side is written as FB(ρ0,ρ1)=∑btr(ρ~0(1⊗Eb))tr(ρ~1(1⊗Eb))F_B(\rho_0,\rho_1) = \sum_b \sqrt{\mathrm{tr}(\tilde\rho_0(1\otimes E_b))}\sqrt{\mathrm{tr}(\tilde\rho_1(1\otimes E_b))}FB​(ρ0​,ρ1​)=∑b​tr(ρ~​0​(1⊗Eb​))​tr(ρ~​1​(1⊗Eb​))​ for an optimal measurement {Eb}\{E_b\}{Eb​} of ρ0,ρ1\rho_0,\rho_1ρ0​,ρ1​, which equals F(ρ0,ρ1)F(\rho_0,\rho_1)F(ρ0​,ρ1​). Together with the companion bound for trB\mathrm{tr}_BtrB​ (4.74) it shows that a partial trace cannot decrease fidelity.

Differences from the source:

  • generalization: the outputs ρ~s\tilde\rho_sρ~​s​ of a broadcasting process on two NNN-dimensional systems → any positive semidefinite operators on ABABAB, with AAA and BBB of any finite dimensions; reason: the source derives the inequality as the general partial-trace property ("this demonstrates the said partial trace property"); broadcasting outputs are such operators, so it specializes to the source's.
  • reading: FB(ρ0,ρ1)F_B(\rho_0,\rho_1)FB​(ρ0​,ρ1​), the optimal-POVM representation of F(ρ0,ρ1)F(\rho_0,\rho_1)F(ρ0​,ρ1​) with ρs=trAρ~s\rho_s = \mathrm{tr}_A\tilde\rho_sρs​=trA​ρ~​s​ → F(trAρ~0,trAρ~1)F(\mathrm{tr}_A\tilde\rho_0, \mathrm{tr}_A\tilde\rho_1)F(trA​ρ~​0​,trA​ρ~​1​); reason: the source says FB(ρ0,ρ1)F_B(\rho_0,\rho_1)FB​(ρ0​,ρ1​) "stands for" F(ρ0,ρ1)F(\rho_0,\rho_1)F(ρ0​,ρ1​) in that representation.
  • encoding: the trace tr⋅\mathrm{tr}\sqrt{\cdot}tr⋅​ of (4.56) → the real part of the trace, square roots by the continuous functional calculus of the real square root; reason: Lean convention; on positive semidefinite arguments the trace is real and the square root is the positive one, so no instance changes.
Preamble
import Mathlib
import Definitions.Def_QAlgorithms_Defs_QuantumInformation
import Definitions.Def_WildeQIT_partialTrace

open scoped ComplexOrder
Formal statement
namespace FuchsThesis.Broadcast

/-- Fuchs, *Distinguishability and Accessible Information in Quantum Theory*,
arXiv:quant-ph/9601020v1, §4.3, eqs. (4.75)–(4.76), p.117 (PDF p.126): fidelity cannot decrease
under the partial trace over `A`: for positive operators `ρ̃₀, ρ̃₁` on `AB`,
`F(tr_A ρ̃₀, tr_A ρ̃₁) ≥ F(ρ̃₀, ρ̃₁)`.

**Formalization Note.** `F` is `QAlgorithms.rootFidelity` = (4.56); `AB` is indexed by `a × b`
(first factor `A`), and `tr_A` is `WildeQIT.partialTraceLeft`. In the source `ρ̃ₛ` are the
broadcast states with `tr_A ρ̃ₛ = ρₛ`, and the left side is written through an optimal POVM for
`ρ₀, ρ₁`; the inequality is the general partial-trace property, stated here for every pair of
positive semidefinite operators on `AB` with `A`, `B` of any finite dimensions. -/
theorem eq_4_76 {a b : Type} [Fintype a] [DecidableEq a] [Fintype b] [DecidableEq b]
    (ρt0 ρt1 : Matrix (a × b) (a × b) ℂ) (h0 : ρt0.PosSemidef) (h1 : ρt1.PosSemidef) :
    QAlgorithms.rootFidelity ρt0 ρt1 ≤
      QAlgorithms.rootFidelity (WildeQIT.partialTraceLeft ρt0) (WildeQIT.partialTraceLeft ρt1) := by sorry

end FuchsThesis.Broadcast
Source
Fuchs, Distinguishability and Accessible Information in Quantum Theory, arXiv:quant-ph/9601020v1, §4.3, eqs. (4.75)–(4.76), p.117 (PDF p.126)
Human review
  • Endorsed by Shuze Chen · Oct 31, 2026

    Confirmed by the moderator at approval.

  • Endorsed by iris · Oct 31, 2026

    Confirmed by the mission captain (proposal self-audit).

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