Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Unnumbered further strict separations in Section VII

Open
QAlgorithms.Papers.Independent.Support50.P16.further_entanglement_separations

by Goku · Oct 23, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

For each of the following inequalities there is a normalized bipartite density matrix on some nonzero finite-dimensional local registers that witnesses strictness. Different inequalities may have different states and dimensions:

Ec>Er∞,Esq,c∞>Er∞,Er∞>Er,→∞,Er∞>Kd,EI>Er,↔∞,Esq>Er,→∞,Esq>Kd,EI>Er,Esq>Er,↔∞.\begin{gathered} E_c>E_r^\infty,\quad E_{\mathrm{sq},c}^\infty>E_r^\infty,\quad E_r^\infty>E_{r,\to}^\infty,\quad E_r^\infty>K_d,\\ E_I>E_{r,\leftrightarrow}^\infty,\quad E_{\mathrm{sq}}>E_{r,\to}^\infty,\quad E_{\mathrm{sq}}>K_d,\\ E_I>E_r,\qquad E_{\mathrm{sq}}>E_{r,\leftrightarrow}^\infty. \end{gathered}Ec​>Er∞​,Esq,c∞​>Er∞​,Er∞​>Er,→∞​,Er∞​>Kd​,EI​>Er,↔∞​,Esq​>Er,→∞​,Esq​>Kd​,EI​>Er​,Esq​>Er,↔∞​.​

There is also an entangled density matrix with Ed=0<Er,→∞E_d=0<E_{r,\to}^\inftyEd​=0<Er,→∞​. Here Ec,Ed,KdE_c,E_d,K_dEc​,Ed​,Kd​ are the operational LOCC formation, distillation, and secret-key rates; ErE_rEr​ is relative entropy of entanglement; Er,↔E_{r,\leftrightarrow}Er,↔​ is its LOCC-measurement version; and Er,→E_{r,\to}Er,→​ optimizes the one-way measured version after arbitrary finite-output LOCC operations. The measures Esq,Esq,c,EIE_{\mathrm{sq}},E_{\mathrm{sq},c},E_IEsq​,Esq,c​,EI​ are, respectively, half the extension infima of conditional mutual information, its classical-flag restriction, and I(AA′;BB′)−I(A′;B′)I(AA';BB')-I(A';B')I(AA′;BB′)−I(A′;B′). All regularizations are actual per-copy limits in bits. These directional witnesses do not assert the conjectural converse directions mentioned on p.15.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Support50_P16
Formal statement
namespace QAlgorithms.Papers.Independent.Support50.P16

noncomputable section

open QAlgorithms Filter

open scoped Matrix ComplexConjugate ComplexOrder ENNReal

attribute [local instance] Classical.propDecidable Classical.decEq

local notation "tau" => QAlgorithms.Papers.Independent.Information100.P76.maximallyMixed

local notation "RE" => QAlgorithms.Papers.Independent.Information100.P54.relativeEntanglement

local notation "RER" => QAlgorithms.Papers.Independent.Information100.P54.regularizedEntanglement

local notation "SQ" => QAlgorithms.Papers.Independent.Dependency50.P19.squashed

local notation "CMI" => QAlgorithms.Papers.Independent.Dependency50.P19.conditionalMutual

local notation "MI" => QAlgorithms.Papers.Independent.Dependency50.P19.mutualInformation

local notation "LOCC" => QAlgorithms.Papers.Independent.Information100.P21.FiniteLOCC

local notation "power" => QAlgorithms.Papers.Independent.Dependency50.P42.dimensionedPower

local notation "index" => QAlgorithms.Papers.Independent.Dependency50.P42.copyIndex

-- Accepted matrix coordinates (not P50.Bipartite, which is a vector space).

local notation "Bipartite" => QAlgorithms.Papers.Independent.Dependency50.P50.BipartiteMat

-- Positive outcome counts follow already from a POVM on a nonzero register.

local notation "FiniteMeasurement" => QAlgorithms.Papers.Independent.Information100.P04.Measurement

local notation "relative" =>
  ((fun _ _ rho => RE rho) : ExtendedEntanglementFunctional)

-- Reuse the accepted operational/variational quantity through a typed notation.

local notation "squashed" =>
  ((fun _ _ rho => ENNReal.ofReal (SQ rho)) : ExtendedEntanglementFunctional)

local notation "originalMarginal" => QAlgorithms.Papers.Independent.Information100.P04.principalMarginal

-- Exact accepted partial trace of the same AA'BB' extension.

local notation "ancillaryMarginal" => QAlgorithms.Papers.Independent.Information100.P04.ancillaryMarginal

local notation "distillable" =>
  ((fun _ _ rho => ENNReal.ofReal (QAlgorithms.Papers.Independent.Gap48.P32.twoWayDistillable rho)) : ExtendedEntanglementFunctional)

local notation "cost" =>
  ((fun _ _ rho => ENNReal.ofReal (QAlgorithms.Papers.Independent.Dependency50.P42.alternativeCost
    (fun _ _ X Y => traceNorm (X-Y)) rho)) : ExtendedEntanglementFunctional)

/-- Unnumbered further separations, PDF pp.15--16: seven displayed pairs,
the two known directions highlighted on p.15, and a bound-entangled witness
separating E_r,rightArrow^infinity from E_d, as printed on p.16.
Every existential has its own state and finite dimensions. -/
theorem further_entanglement_separations :
    StrictSeparation cost (fun _ _ rho => regularized relative rho) ∧
    StrictSeparation (fun _ _ rho => regularized classicalSquashed rho)
      (fun _ _ rho => regularized relative rho) ∧
    StrictSeparation (fun _ _ rho => regularized relative rho)
      (fun _ _ rho => regularized updated rho) ∧
    StrictSeparation (fun _ _ rho => regularized relative rho) key ∧
    StrictSeparation conditionalEntanglement (fun _ _ rho => regularized (measuredEntanglement .LOCC) rho) ∧
    StrictSeparation squashed (fun _ _ rho => regularized updated rho) ∧
    StrictSeparation squashed key ∧
    StrictSeparation conditionalEntanglement relative ∧
    StrictSeparation squashed (fun _ _ rho => regularized (measuredEntanglement .LOCC) rho) ∧
    (∃ a b : ℕ+, ∃ rho : Bipartite a b, IsDensityMatrix rho ∧
      rho ∉ separable a b ∧ Regularizes updated rho ∧
      distillable a b rho = 0 ∧ 0 < regularized updated rho) := by sorry

end

end QAlgorithms.Papers.Independent.Support50.P16
Source
Relative entropy and squashed entanglement, arXiv:1210.3181v2, PDF p. 16, Unnumbered further strict separations in Section VII
Read-back

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

Write C,R,Q,DC,R,Q,DC,R,Q,D for the functions named cost, relative, squashed, and distillable, respectively; their definitions are not included in the supplied code. For a functional FFF, write

F∞(ρ)=lim inf⁡r→∞F(ρ[r+1])r+1,F^\infty(\rho)=\liminf_{r\to\infty} \frac{F\bigl(\rho^{[r+1]}\bigr)}{r+1},F∞(ρ)=r→∞liminf​r+1F(ρ[r+1])​,

where ρ[n]\rho^{[n]}ρ[n] denotes the operation named power, whose definition is likewise not supplied, and the local dimensions in the numerator are ar+1,br+1a^{r+1},b^{r+1}ar+1,br+1 when ρ\rhoρ has local dimensions a,ba,ba,b. Say that FFF regularizes at ρ\rhoρ when this sequence actually converges to F∞(ρ)F^\infty(\rho)F∞(ρ); these values and limits lie in [0,∞][0,\infty][0,∞], and finiteness is not required. Let SSS be the functional obtained by taking one half of the real infimum of ∑ipiI(ρi)\sum_i p_i\mathcal I(\rho_i)∑i​pi​I(ρi​) over all finite ensembles with pi≥0p_i\ge0pi​≥0, ∑ipi=1\sum_i p_i=1∑i​pi​=1, density matrices ρi\rho_iρi​, and ∑ipiρi=ρ\sum_i p_i\rho_i=\rho∑i​pi​ρi​=ρ, and then applying x↦max⁡(x,0)x\mapsto\max(x,0)x↦max(x,0); here I\mathcal II is the symbol MI, whose definition is absent. A density matrix means a positive semidefinite complex matrix of trace 111. Let Ec(ρ)E_c(\rho)Ec​(ρ) be the infimum over separable density matrices σ\sigmaσ of the supremum over measurements in class ccc of

K(p,q)={max⁡ ⁣(0,∑xpx(log⁡2px−log⁡2qx)),∀x, px>0⇒qx>0,+∞,otherwise,px=Re⁡tr⁡(Mxρ),qx=Re⁡tr⁡(Mxσ).K(p,q)= \begin{cases} \displaystyle \max\!\left(0,\sum_xp_x\bigl(\log_2p_x-\log_2q_x\bigr)\right), &\forall x,\ p_x>0\Rightarrow q_x>0,\\ +\infty,&\text{otherwise}, \end{cases} \qquad p_x=\operatorname{Re}\operatorname{tr}(M_x\rho),\quad q_x=\operatorname{Re}\operatorname{tr}(M_x\sigma).K(p,q)=⎩⎨⎧​max(0,x∑​px​(log2​px​−log2​qx​)),+∞,​∀x, px​>0⇒qx​>0,otherwise,​px​=Retr(Mx​ρ),qx​=Retr(Mx​σ).

Separable density matrices here are finite sums ∑iwiAi⊗Bi\sum_i w_i A_i\otimes B_i∑i​wi​Ai​⊗Bi​, with wi≥0w_i\ge0wi​≥0, ∑iwi=1\sum_iw_i=1∑i​wi​=1, and density matrices Ai,BiA_i,B_iAi​,Bi​. The measurement type is named FiniteMeasurement, and its full definition is not supplied. Its one-way class consists of measurements whose effects have the form

Mx=∑i,j: f(i,j)=xAi⊗Bi,j,M_x=\sum_{i,j:\,f(i,j)=x}A_i\otimes B_{i,j},Mx​=i,j:f(i,j)=x∑​Ai​⊗Bi,j​,

for finite families with Ai≥0A_i\ge0Ai​≥0, ∑iAi=1\sum_iA_i=1∑i​Ai​=1, Bi,j≥0B_{i,j}\ge0Bi,j​≥0, and ∑jBi,j=1\sum_jB_{i,j}=1∑j​Bi,j​=1 for every iii, and an outcome map fff. Its LOCC class consists of measurements for which there exists a map LLL satisfying the superoperator predicate named LOCC and (L(X))(0,x),(0,x)=tr⁡(MxX)(L(X))_{(0,x),(0,x)}=\operatorname{tr}(M_xX)(L(X))(0,x),(0,x)​=tr(Mx​X) for every complex matrix XXX and outcome xxx; that superoperator predicate is not defined in the supplied code. Define U(ρ)U(\rho)U(ρ) as the supremum of EoneWay(L(ρ))E_{\mathrm{oneWay}}(L(\rho))EoneWay​(L(ρ)) over all positive finite output dimensions and all maps LLL satisfying that same predicate. Let J(ρ)J(\rho)J(ρ) be the infimum, over positive finite ancillary dimensions and density matrices ω\omegaω satisfying originalMarginal(ω)=ρ(\omega)=\rho(ω)=ρ, of

max⁡ ⁣(0,I(ω)−I(ancillaryMarginal⁡ω)2).\max\!\left(0,\frac{\mathcal I(\omega)-\mathcal I(\operatorname{ancillaryMarginal}\omega)}2\right).max(0,2I(ω)−I(ancillaryMarginalω)​).

The ancillary marginal traces out the two principal factors; the defining equation for originalMarginal is not supplied. Let Kkey(ρ)K_{\mathrm{key}}(\rho)Kkey​(ρ) be the supremum of all nonnegative real rates rrr with the following property: for every ϵ,δ>0\epsilon,\delta>0ϵ,δ>0, all sufficiently large positive integers nnn admit positive integers d,s,td,s,td,s,t, a density matrix τ\tauτ on Cs⊗Ct\mathbb C^s\otimes\mathbb C^tCs⊗Ct, unitaries V0,…,Vd−1V_0,\ldots,V_{d-1}V0​,…,Vd−1​ on that space, and a finite adaptive local Kraus protocol taking the nnn-copy matrix of ρ\rhoρ to within trace norm ϵ\epsilonϵ of the private matrix whose entries, after grouping each party’s key and ancillary indices, are d−1(ViτVj∗)αβ,α′β′d^{-1}(V_i\tau V_j^*)_{\alpha\beta,\alpha'\beta'}d−1(Vi​τVj∗​)αβ,α′β′​ when both row keys equal iii and both column keys equal jjj, and zero otherwise, with r−δ≤log⁡2(d)/nr-\delta\le\log_2(d)/nr−δ≤log2​(d)/n. Such a protocol is built recursively from identity maps and finite complete local Kraus families, with a subsequent protocol chosen for each Kraus outcome and the branch outputs summed; trace norm here is Re⁡tr⁡X∗X\operatorname{Re}\operatorname{tr}\sqrt{X^*X}RetrX∗X​. The theorem asserts that each of the following nine strict inequalities has a witness consisting of positive integers a,ba,ba,b and a bipartite density matrix ρ\rhoρ, at which all four functionals R,S,U,ELOCCR,S,U,E_{\mathrm{LOCC}}R,S,U,ELOCC​ regularize:

R∞(ρ)<C(ρ),R∞(ρ)<S∞(ρ),U∞(ρ)<R∞(ρ),Kkey(ρ)<R∞(ρ),ELOCC∞(ρ)<J(ρ),U∞(ρ)<Q(ρ),Kkey(ρ)<Q(ρ),R(ρ)<J(ρ),ELOCC∞(ρ)<Q(ρ).\begin{gathered} R^\infty(\rho)<C(\rho),\\ R^\infty(\rho)<S^\infty(\rho),\\ U^\infty(\rho)<R^\infty(\rho),\\ K_{\mathrm{key}}(\rho)<R^\infty(\rho),\\ E_{\mathrm{LOCC}}^\infty(\rho)<J(\rho),\\ U^\infty(\rho)<Q(\rho),\\ K_{\mathrm{key}}(\rho)<Q(\rho),\\ R(\rho)<J(\rho),\\ E_{\mathrm{LOCC}}^\infty(\rho)<Q(\rho). \end{gathered}R∞(ρ)<C(ρ),R∞(ρ)<S∞(ρ),U∞(ρ)<R∞(ρ),Kkey​(ρ)<R∞(ρ),ELOCC∞​(ρ)<J(ρ),U∞(ρ)<Q(ρ),Kkey​(ρ)<Q(ρ),R(ρ)<J(ρ),ELOCC∞​(ρ)<Q(ρ).​

The dimensions and state may be different for every inequality. In addition, there exist positive integers a,ba,ba,b and a bipartite density matrix ρ\rhoρ that is not separable in the finite-mixture sense above, such that UUU regularizes at ρ\rhoρ, D(a,b,ρ)=0D(a,b,\rho)=0D(a,b,ρ)=0, and 0<U∞(ρ)0<U^\infty(\rho)0<U∞(ρ). All infima, suprema, logarithms, and divisions retain the total operations and codomains used in the code.

Human review
  • Endorsed by Shuze Chen · Oct 24, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Goku · Oct 24, 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