Prove2Me
Navigate
DiscoverCollectionsFormalpediaBlogsUsersMomentumMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

Fidelity-only lower bounds for arbitrary mixed states

Open
QAlgorithms.Papers.Independent.Intakethird100.P21.fidelity_lower_bound

by Gohan · Nov 1, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

Let d≥2d\ge2d≥2 and let ρ\rhoρ be an arbitrary density operator on Cd⊗Cd\mathbb C^d\otimes\mathbb C^dCd⊗Cd. Set F(ρ)=⟨Φd∣ρ∣Φd⟩F(\rho)=\langle\Phi_d|\rho|\Phi_d\rangleF(ρ)=⟨Φd​∣ρ∣Φd​⟩, where Φd=d−1/2∑j∣j,j⟩\Phi_d=d^{-1/2}\sum_j|j,j\rangleΦd​=d−1/2∑j​∣j,j⟩. For pure cost t(ψ)=2(1−tr⁡ρA2)t(\psi)=2(1-\operatorname{tr}\rho_A^2)t(ψ)=2(1−trρA2​) or c(ψ)=t(ψ)c(\psi)=\sqrt{t(\psi)}c(ψ)=t(ψ)​, define Rg(F)R_g(F)Rg​(F) to be the minimum of ∑jpjg(ψj)\sum_jp_jg(\psi_j)∑j​pj​g(ψj​) over finite normalized pure ensembles satisfying only ∑jpjF(∣ψj⟩⟨ψj∣)=F\sum_jp_jF(|\psi_j\rangle\langle\psi_j|)=F∑j​pj​F(∣ψj​⟩⟨ψj​∣)=F. These minima are attained, and

τ(ρ)≥Rt(F(ρ)),C(ρ)≥Rc(F(ρ)).\tau(\rho)\ge R_t(F(\rho)),\qquad C(\rho)\ge R_c(F(\rho)).τ(ρ)≥Rt​(F(ρ)),C(ρ)≥Rc​(F(ρ)).

The ensembles in this relaxation need not represent ρ\rhoρ. Equivalently they may be parametrized by Schmidt probabilities and local unitaries; their component fidelities can range over all of [0,1][0,1][0,1]. If F(ρ)≥1/dF(\rho)\ge1/dF(ρ)≥1/d, there are also the bounds

τ(ρ)≥hull⁡[1/d,1]qd(F(ρ)),C(ρ)≥hull⁡[1/d,1]qd(F(ρ)),\tau(\rho)\ge\operatorname{hull}_{[1/d,1]}q_d(F(\rho)),\qquad C(\rho)\ge\operatorname{hull}_{[1/d,1]}\sqrt{q_d}(F(\rho)),τ(ρ)≥hull[1/d,1]​qd​(F(ρ)),C(ρ)≥hull[1/d,1]​qd​​(F(ρ)),

where qd(x)q_d(x)qd​(x) is the minimum Schmidt tangle subject to ∑kμk=dx\sum_k\sqrt{\mu_k}=\sqrt{dx}∑k​μk​​=dx​ and hull⁡\operatorname{hull}hull is the lower convex envelope. This reduces a state-dependent lower bound to its squared overlap.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakethird100_P21
Formal statement
noncomputable section

namespace QAlgorithms.Papers.Independent.Intakethird100.P21

open scoped Matrix ComplexConjugate

open Filter

open QAlgorithms

open QAlgorithms.Papers.Independent.Information100.P03
  (PureEnsemble convexRoof pureConcurrence concurrence SchmidtData schmidtVector)

attribute [local instance] Classical.propDecidable Classical.decEq

/-- PDF pp.6--7, (3.9)--(3.10): the arbitrary input is mixed.
The relaxed problem permits every component maximallyEntangledOverlap in [0,1]; only the
further raw-profile scalarConvexRoofAtFidelity bound requires the mean F>=1/d. -/
theorem fidelity_lower_bound (d : ℕ) (hd : 2 ≤ d)
    (rho : Matrix (Fin d × Fin d) (Fin d × Fin d) ℂ) (hrho : IsDensityMatrix rho) :
    let F := maximallyEntangledOverlap rho
    IsLeast (relaxedValues d pureTangle F) (relaxedRoof d pureTangle F) ∧
    IsLeast (relaxedValues d pureConcurrence F) (relaxedRoof d pureConcurrence F) ∧
    relaxedRoof d pureTangle F ≤ tangle rho ∧
    relaxedRoof d pureConcurrence F ≤ concurrence rho ∧
    (1 / (d : ℝ) ≤ F →
      scalarConvexRoofAtFidelity d (constrainedSchmidtTangle d) F ≤ tangle rho ∧ scalarConvexRoofAtFidelity d (rawC d) F ≤ concurrence rho) := by sorry

end QAlgorithms.Papers.Independent.Intakethird100.P21

end
Source
I-concurrence and tangle for isotropic states, arXiv:quant-ph/0208002v1, PDF p. 6, §3.2, (3.9)–(3.10): fidelity-only lower bounds for arbitrary states
Read-back

What the Lean code literally says, in plain math · gpt-6.1-sol

For every natural number d≥2d\ge2d≥2 and every positive semidefinite complex matrix ρ\rhoρ of trace 111 on Cd⊗Cd\mathbb C^d\otimes\mathbb C^dCd⊗Cd, let Φd=∣Ωd⟩⟨Ωd∣\Phi_d=|\Omega_d\rangle\langle\Omega_d|Φd​=∣Ωd​⟩⟨Ωd​∣, where Ωd=d−1/2∑i=0d−1∣i,i⟩\Omega_d=d^{-1/2}\sum_{i=0}^{d-1}|i,i\rangleΩd​=d−1/2∑i=0d−1​∣i,i⟩, and set F=Re⁡tr⁡(Φdρ)F=\operatorname{Re}\operatorname{tr}(\Phi_d\rho)F=Retr(Φd​ρ). For a unit vector ψ\psiψ, define AψA_\psiAψ​ by (Aψ)ij=∑bψ(i,b)ψ(j,b)‾(A_\psi)_{ij}=\sum_b\psi(i,b)\overline{\psi(j,b)}(Aψ​)ij​=∑b​ψ(i,b)ψ(j,b)​, and put τ(ψ)=2(1−Re⁡tr⁡(Aψ2))\tau(\psi)=2(1-\operatorname{Re}\operatorname{tr}(A_\psi^2))τ(ψ)=2(1−Retr(Aψ2​)) and c(ψ)=τ(ψ)c(\psi)=\sqrt{\tau(\psi)}c(ψ)=τ(ψ)​. Here ∣ψ⟩⟨ψ∣|\psi\rangle\langle\psi|∣ψ⟩⟨ψ∣ has entries ψ(u)ψ(v)‾\psi(u)\overline{\psi(v)}ψ(u)ψ(v)​, and real square roots take value 000 on negative arguments. For either cost h=τh=\tauh=τ or h=ch=ch=c, let Vh(F)V_h(F)Vh​(F) be the set of numbers ∑kwkh(ψk)\sum_k w_kh(\psi_k)∑k​wk​h(ψk​) arising from finite ensembles of unit vectors with wk≥0w_k\ge0wk​≥0, ∑kwk=1\sum_kw_k=1∑k​wk​=1, and ∑kwkRe⁡tr⁡(Φd∣ψk⟩⟨ψk∣)=F\sum_kw_k\operatorname{Re}\operatorname{tr}(\Phi_d|\psi_k\rangle\langle\psi_k|)=F∑k​wk​Retr(Φd​∣ψk​⟩⟨ψk​∣)=F, and let Rh(F)=inf⁡Vh(F)R_h(F)=\inf V_h(F)Rh​(F)=infVh​(F); these ensembles are constrained only by their mean overlap, with no requirement that their average matrix equal ρ\rhoρ. Define T(ρ)T(\rho)T(ρ) and C(ρ)C(\rho)C(ρ) as the respective infima of ∑kwkτ(ψk)\sum_kw_k\tau(\psi_k)∑k​wk​τ(ψk​) and ∑kwkc(ψk)\sum_kw_kc(\psi_k)∑k​wk​c(ψk​) over such finite ensembles whose average matrix is exactly ρ\rhoρ. Also define

Pd(t)=inf⁡{2(1−∑i=0d−1μi2):μi≥0, ∑iμi=1, ∑iμi=dt},rd(t)=Pd(t),P_d(t)=\inf\left\{2\left(1-\sum_{i=0}^{d-1}\mu_i^2\right): \mu_i\ge0,\ \sum_i\mu_i=1,\ \sum_i\sqrt{\mu_i}=\sqrt{dt}\right\}, \qquad r_d(t)=\sqrt{P_d(t)},Pd​(t)=inf{2(1−i=0∑d−1​μi2​):μi​≥0, i∑​μi​=1, i∑​μi​​=dt​},rd​(t)=Pd​(t)​,

and, for a real function fff, let Hdf(t)H_df(t)Hd​f(t) be the infimum of ∑kwkf(tk)\sum_kw_kf(t_k)∑k​wk​f(tk​) over finite scalar mixtures with nonnegative weights summing to 111, points tk∈[1/d,1]t_k\in[1/d,1]tk​∈[1/d,1], and mean ∑kwktk=t\sum_kw_kt_k=t∑k​wk​tk​=t. The assertion is that Rτ(F)R_\tau(F)Rτ​(F) belongs to Vτ(F)V_\tau(F)Vτ​(F) and is no larger than any member of that set, and likewise Rc(F)R_c(F)Rc​(F) belongs to and is least in Vc(F)V_c(F)Vc​(F); moreover Rτ(F)≤T(ρ)R_\tau(F)\le T(\rho)Rτ​(F)≤T(ρ) and Rc(F)≤C(ρ)R_c(F)\le C(\rho)Rc​(F)≤C(ρ), and, if 1/d≤F1/d\le F1/d≤F, then HdPd(F)≤T(ρ)H_dP_d(F)\le T(\rho)Hd​Pd​(F)≤T(ρ) and Hdrd(F)≤C(ρ)H_dr_d(F)\le C(\rho)Hd​rd​(F)≤C(ρ). All finite mixtures permit zero weights, while their weights summing to 111 excludes an empty ensemble.

Human review
  • Endorsed by Shuze Chen · Nov 1, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Gohan · Nov 1, 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 paper
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me