Prove2Me
Navigate
DiscoverCollectionsFormalpediaBlogsUsersMomentumMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

II B: removal of constant columns preserves threshold, (8)

Open
QAlgorithms.Papers.Independent.Intakethird100.P30.constantColumns

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

prove2quantum

Let C=(Ck)k∈KC=(C_k)_{k\in K}C=(Ck​)k∈K​ be a nonempty code of distinct nnn-bit words, n≥1n\ge1n≥1. Separate its coordinates into the fixed coordinates FFF and the varying coordinates VVV. Let Dk=Ck∣VD_k=C_k|_VDk​=Ck​∣V​, and let β=Ck0∣F\beta=C_{k_0}|_Fβ=Ck0​​∣F​ for any chosen codeword k0k_0k0​; the latter does not depend on the chosen word. The residual words remain distinct and have no constant coordinates. For the Bob-only filtered operators, explicit restriction of Alice's coordinate register gives

PC(Q)=P{β}(Q)⊗PD(Q),ExtB(P{β}(Q)),QC∗=QD∗(0≤Q≤12).P_C(Q)=P_{\{\beta\}}(Q)\otimes P_D(Q),\qquad \mathrm{Ext}_B(P_{\{\beta\}}(Q)),\qquad Q_C^*=Q_D^* \quad(0\le Q\le\tfrac12).PC​(Q)=P{β}​(Q)⊗PD​(Q),ExtB​(P{β}​(Q)),QC∗​=QD∗​(0≤Q≤21​).

The singleton factor has a one-dimensional Bob register. This retains the full factorization underlying column deletion. A singleton original code can leave a zero-length residual register, which is only a helper case and has threshold zero.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakethird100_P30
Formal statement
noncomputable section

namespace QAlgorithms.Papers.Independent.Intakethird100.P30

open scoped BigOperators Matrix ComplexConjugate ComplexOrder

attribute [local instance] Classical.propDecidable Classical.decEq

variable {I J K L A B : Type}

variable [Fintype I] [Fintype J] [Fintype K] [Fintype L]

variable [Fintype A] [Fintype B]

/-- PDF p.4, (8): deletion of every constant column, with full factorization
under coordinate restriction and the extendable one-dimensional-B factor.
The residual coordinate register may be empty for a singleton code. -/
theorem constantColumns {n : ℕ} (hn : 1 ≤ n) [Nonempty K]
    (C : K → Qubits n) (hC : Function.Injective C) (k0 : K) :
    Function.Injective (deleteConstants C) ∧ NoConstantColumns (deleteConstants C) ∧
    (∀ Q : ℝ, 0 ≤ Q → Q ≤ 1 / 2 →
      BobExtendible (bobState (fun _ : Unit => fun i : FixedCoordinates C => C k0 i.val) Q) ∧
      ∀ (a a2 : Qubits n) (k l : K),
        bobState C Q (a, k) (a2, l) =
          bobState (fun _ : Unit => fun i : FixedCoordinates C => C k0 i.val) Q
            ((fun i => a i.val), ()) ((fun i => a2 i.val), ()) *
          bobState (deleteConstants C) Q
            ((fun i => a i.val), k) ((fun i => a2 i.val), l)) ∧
    bobThreshold C = bobThreshold (deleteConstants C) := by sorry

end QAlgorithms.Papers.Independent.Intakethird100.P30

end
Source
Numerical evidence for bound secrecy from two-way post-processing in quantum key distribution, arXiv:1612.07734v2, PDF p. 4, II B: removal of constant columns preserves threshold, (8)
Read-back

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

Let n≥1n\ge1n≥1, let KKK be a finite nonempty label type with the required decidable equality, let C:K→{0,1}nC:K\to\{0,1\}^{n}C:K→{0,1}n be injective, and choose k0∈Kk_0\in Kk0​∈K. Let F={i:∀k,l, C(k)i=C(l)i}F=\{i:\forall k,l,\ C(k)_i=C(l)_i\}F={i:∀k,l, C(k)i​=C(l)i​} and V={i:∃k,l, C(k)i≠C(l)i}V=\{i:\exists k,l,\ C(k)_i\ne C(l)_i\}V={i:∃k,l, C(k)i​=C(l)i​}; define CV(k)i=C(k)iC_V(k)_i=C(k)_iCV​(k)i​=C(k)i​ for i∈Vi\in Vi∈V, and define the one-label code CF(∗)i=C(k0)iC_F(*)_i=C(k_0)_iCF​(∗)i​=C(k0​)i​ for i∈Fi\in Fi∈F. For any code E:L→{0,1}JE:L\to\{0,1\}^{J}E:L→{0,1}J, put

BE,Q((a,k),(a′,l))=∏i∈J(1−2Q21ai=E(k)i1ai′=E(l)i+Q21ai=ai′1E(k)i=E(l)i).B_{E,Q}((a,k),(a',l)) =\prod_{i\in J} \left( \frac{1-2Q}{2}\mathbf1_{a_i=E(k)_i}\mathbf1_{a'_i=E(l)_i} +\frac Q2\mathbf1_{a_i=a'_i}\mathbf1_{E(k)_i=E(l)_i} \right).BE,Q​((a,k),(a′,l))=i∈J∏​(21−2Q​1ai​=E(k)i​​1ai′​=E(l)i​​+2Q​1ai​=ai′​​1E(k)i​=E(l)i​​).

The theorem asserts that CVC_VCV​ is injective and that every coordinate of CVC_VCV​ takes different values on some pair of labels. For every Q∈[0,1/2]Q\in[0,1/2]Q∈[0,1/2], it asserts that BCF,QB_{C_F,Q}BCF​,Q​ admits a positive semidefinite extension to two copies of its label subsystem with both corresponding partial traces equal to BCF,QB_{C_F,Q}BCF​,Q​, and that, for every a,a′∈{0,1}na,a'\in\{0,1\}^{n}a,a′∈{0,1}n and k,l∈Kk,l\in Kk,l∈K,

BC,Q((a,k),(a′,l))=BCF,Q((a∣F,∗),(a′∣F,∗)) BCV,Q((a∣V,k),(a′∣V,l)).B_{C,Q}((a,k),(a',l)) = B_{C_F,Q}((a|_F,*),(a'|_F,*))\, B_{C_V,Q}((a|_V,k),(a'|_V,l)).BC,Q​((a,k),(a′,l))=BCF​,Q​((a∣F​,∗),(a′∣F​,∗))BCV​,Q​((a∣V​,k),(a′∣V​,l)).

It also asserts b(C)=b(CV)b(C)=b(C_V)b(C)=b(CV​), where b(E)b(E)b(E) is the supremum of {0}\{0\}{0} together with all q∈[0,1/2]q\in[0,1/2]q∈[0,1/2] for which BE,qB_{E,q}BE,q​ has no such positive semidefinite extension; no trace-one condition is imposed on extensions. Either FFF or VVV may be empty: the corresponding products then equal 111, and the assertion about variation on an empty coordinate type is vacuous.

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