Prove2Me
Navigate
DiscoverCollectionsFormalpediaBlogsUsersMomentumMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

Unnumbered equation (218): ensemble convex-roof form

Open
QAlgorithms.Papers.Independent.Intakethird100.P22.ensemble_equivalence

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

prove2quantum

All input and output alphabets are finite and nonempty. Let P(a,b∣x,y)P(a,b\mid x,y)P(a,b∣x,y) be a normalized nonnegative bipartite non-signaling device, and let E(P)\mathcal E(P)E(P) be its complete extension, with possibly input-dependent pure-member labels. All Shannon information quantities are in bits, with 0log⁡20=00\log_2 0=00log2​0=0. Write Nsq(P)N_{\rm sq}(P)Nsq​(P) for the maximum over honest settings (x,y)(x,y)(x,y) of the infimum, over Eve's general measurements, of intrinsic information; intrinsic information itself minimizes classical conditional mutual information over arbitrary stochastic channels on Eve.

Let {pi,Pi}\{p_i,P_i\}{pi​,Pi​} range over all finite ensembles of non-signaling devices in the same polytope, with nonnegative normalized weights and ∑ipiPi(a,b∣x,y)=P(a,b∣x,y)\sum_i p_iP_i(a,b\mid x,y)=P(a,b\mid x,y)∑i​pi​Pi​(a,b∣x,y)=P(a,b∣x,y) for every input and output. Then

Nsq(P)=max⁡x,yinf⁡{pi,Pi}∑ipiI(A:B)Pi(⋅∣x,y).N_{\rm sq}(P)=\max_{x,y}\inf_{\{p_i,P_i\}}\sum_i p_i I(A:B)_{P_i(\cdot\mid x,y)}.Nsq​(P)=x,ymax​{pi​,Pi​}inf​i∑​pi​I(A:B)Pi​(⋅∣x,y)​.

The ensemble members need not be extreme, and the infimum precedes the final honest-input maximum.

Eve retains the dice result zzz and extension output eee. Squashed CMI conditions on (z,e)(z,e)(z,e); intrinsic-information channels may depend on both. This is the operational convention of XI.B.3 used by the proof, rather than discarding the dice result in display (7).

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakethird100_P22
Formal statement
noncomputable section

open scoped BigOperators

namespace QAlgorithms.Papers.Independent.Intakethird100.P22

attribute [local instance] Classical.propDecidable Classical.decEq

open QAlgorithms.Papers.Independent.Advance50.P47
  (Box IsCorrelation IsNoSignaling HasLocalModel correlationMixture prBox)

open QAlgorithms.Papers.Independent.Information100.P56 (IsLaw)

open QAlgorithms.Papers.Independent.Information100.P27 (shannon)

open QAlgorithms.Papers.Independent.Gap48.P25 (classicalMI)

section Devices

variable {X Y A B : Type} [Fintype X] [Fintype Y] [Fintype A] [Fintype B]
  [Nonempty X] [Nonempty Y] [Nonempty A] [Nonempty B]

/-- PDF p.45, (218): all-device ensemble convex-roof form; no inf/max exchange. Source convention: Eve keeps the dice
result (p.22, XI.B.3); CE members may depend on (z,e) -/
theorem ensemble_equivalence {p : Box X Y A B} (c : CompleteExtension p) :
    squashedNonlocality c = ensembleForm p := by sorry

end Devices

end QAlgorithms.Papers.Independent.Intakethird100.P22

end
Source
Limitations on device independent key secure against non signaling adversary via the squashed non-locality, arXiv:1903.12154v7, PDF p. 45, Equation (218)
Read-back

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

For finite nonempty sets X,Y,A,BX,Y,A,BX,Y,A,B, a box p(x,y,a,b)p(x,y,a,b)p(x,y,a,b), and a complete extension ccc of ppp, the theorem asserts

max⁡x,yinf⁡dz≥0∑zdz=1I↓ ⁣((a,b,e)↦∑zdzwzepze(x,y,a,b))=max⁡x,yinf⁡n≥1p=∑i<nwipi∑i<nwiIpi(x,y,⋅,⋅)(A:B).\max_{x,y}\inf_{\substack{d_z\ge0\\\sum_zd_z=1}} I_\downarrow\!\left((a,b,e)\mapsto \sum_zd_zw_{ze}p_{ze}(x,y,a,b)\right) = \max_{x,y}\inf_{\substack{n\ge1\\p=\sum_{i<n}w_ip_i}} \sum_{i<n}w_i I_{p_i(x,y,\cdot,\cdot)}(A:B).x,ymax​dz​≥0∑z​dz​=1​inf​I↓​((a,b,e)↦z∑​dz​wze​pze​(x,y,a,b))=x,ymax​n≥1p=∑i<n​wi​pi​​inf​i<n∑​wi​Ipi​(x,y,⋅,⋅)​(A:B).

A valid box has nonnegative entries summing to 111 over both outputs at every input pair, with either party’s marginal independent of the other party’s input. The ensembles on the right range over all finite convex decompositions into valid boxes, allowing zero weights. For a finite law rrr, H(r)=−∑ur(u)log⁡2r(u)H(r)=-\sum_ur(u)\log_2r(u)H(r)=−∑u​r(u)log2​r(u), Ir(A:B)=H(rA)+H(rB)−H(r)I_r(A:B)=H(r_A)+H(r_B)-H(r)Ir​(A:B)=H(rA​)+H(rB​)−H(r), and Ir(A:B∣E)=H(rAE)+H(rBE)−H(rE)−H(r)I_r(A:B\mid E)=H(r_{AE})+H(r_{BE})-H(r_E)-H(r)Ir​(A:B∣E)=H(rAE​)+H(rBE​)−H(rE​)−H(r); I↓(r)I_\downarrow(r)I↓​(r) is the infimum of the latter quantity after every stochastic kernel from EEE to an arbitrary positive finite set. The data of ccc are positive integers Z,EZ,EZ,E and, for each z<Zz<Zz<Z, a minimal ensemble (wze,pze)e<E(w_{ze},p_{ze})_{e<E}(wze​,pze​)e<E​ of ppp, with member labels pzep_{ze}pze​ fixed across zzz. Minimality means that every positive-weight member is extreme among valid boxes and that no ensemble of ppp has support properly contained in its positive-weight support; extremality rules out any nontrivial convex decomposition into distinct valid boxes. Every minimal ensemble of ppp, of any finite size, must agree in total weight at every box with exactly one of the indexed ensembles of ccc.

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