Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 3 (seed paper result): bounded-rank reversibility and the pure-state normal form

Proved
QAlgorithms.Papers.Independent.Gap48.P45.boundedRankReversibility_complement

by Goku · Oct 22, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

All physical Hilbert spaces are nonzero complex separable Hilbert spaces, with finite dimensions included. States are positive trace-class operators of trace one, channels are completely positive and trace preserving on the entire trace class, and logarithms are natural. Inner products are conjugate-linear in the first argument. Write HBΦH_B^\PhiHBΦ​ for the closed span of all output supports, dΦ=dim⁡HBΦd_\Phi=\dim H_B^\PhidΦ​=dimHBΦ​, m0(Φ)=dim⁡ker⁡(Φ∗∣B(HBΦ))m_0(\Phi)=\dim\ker(\Phi^*|_{\mathcal B(H_B^\Phi)})m0​(Φ)=dimker(Φ∗∣B(HBΦ​)​), and mr=min⁡{m0(Φ)+r2,dΦ}m_r=\min\{m_0(\Phi)+r^2,d_\Phi\}mr​=min{m0​(Φ)+r2,dΦ​}; dimensions retain infinity. Let S={ρi}i∈LS=\{\rho_i\}_{i\in L}S={ρi​}i∈L​ be a complete finite or countable family with rank⁡ρi≤r\operatorname{rank}\rho_i\le rrankρi​≤r, where r≥1r\ge1r≥1. If one channel recovers every member after Φ\PhiΦ, every fixed complement Φ^\widehat\PhiΦ admits a countable Kraus family VaV_aVa​ satisfying

#{a}≤∣L∣mr,rank⁡Va≤r.\#\{a\}\le |L|m_r,\qquad \operatorname{rank}V_a\le r.#{a}≤∣L∣mr​,rankVa​≤r.

The complement is rrr-partially entanglement breaking; every ancillary output is in fact a countable convex mixture of normalized pure states of Schmidt rank at most rrr. If r=1r=1r=1, write ρi=∣vi⟩⟨vi∣\rho_i=|v_i\rangle\langle v_i|ρi​=∣vi​⟩⟨vi​∣ with ∥vi∥=1\|v_i\|=1∥vi​∥=1. For every strictly positive probability law πi\pi_iπi​, compute ρˉ=∑iπi∣vi⟩⟨vi∣\bar\rho=\sum_i\pi_i|v_i\rangle\langle v_i|ρˉ​=∑i​πi​∣vi​⟩⟨vi​∣ and wi=πiρˉ−1/2viw_i=\sqrt{\pi_i}\bar\rho^{-1/2}v_iwi​=πi​​ρˉ​−1/2vi​. These inverse images exist, {wi}\{w_i\}{wi​} is Parseval, and there are mutually orthogonal vectors ψik\psi_{ik}ψik​ within each row with ∑k∥ψik∥2=1\sum_k\|\psi_{ik}\|^2=1∑k​∥ψik​∥2=1 such that

Φ^(X)=∑i⟨wi∣X∣wi⟩∑k=1m1∣ψik⟩⟨ψik∣.\widehat\Phi(X)=\sum_i\langle w_i|X|w_i\rangle\sum_{k=1}^{m_1}|\psi_{ik}\rangle\langle\psi_{ik}|.Φ(X)=i∑​⟨wi​∣X∣wi​⟩k=1∑m1​​∣ψik​⟩⟨ψik​∣.

This result provides a necessary structural condition on the complementary channel and its exact pure-family normal form.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Gap48_P45
Formal statement
noncomputable section

open Filter MeasureTheory

open scoped BigOperators ENNReal Topology

attribute [local instance] Classical.propDecidable Classical.decEq

attribute [local instance] IsStarNormal.instContinuousFunctionalCalculus

namespace QAlgorithms.Papers.Independent.Gap48.P45

open QAlgorithms.RealRegisters (Hilbert)

open QAlgorithms.Papers.Independent.Dependency50.P33
  (State IsTraceClass IsTraceClassChannel traceSize traceClosure rankOne productCoeff tensorApply)

open QAlgorithms.Papers.Independent.Frontier18.P01 (Channel Complementary traceFirst traceSecond)

open QAlgorithms.Papers.Independent.Frontier18.P07
  (GeneralizedEnsemble averageCoefficients extendedEntropyNats extendedRelativeEntropyNats ensembleHolevo pureStates finiteEnsembleCapacity)

open QAlgorithms.Papers.Independent.Frontier18.P18
  (operatorCoeff boundedOperator conjugateCoeff ReversibleOn IsometricallyEquivalent
   IsParsevalFrame IsKrausRepresentation)

open QAlgorithms.Papers.Independent.Claimed50.P16 (tensorRankLE)

open QAlgorithms.Papers.Independent.Next46.P31 (positiveSquareRoot)

open QAlgorithms.Papers.Independent.Next46.P42 (minimumOutputEntropy)

/-- Theorem 3, PDF p.8, both clauses. Complementarity uses the SAME isometry.
Input count and dimensions retain infinity. No probability law is a hypothesis
of the rank assertion; every positive law is allowed for the pure assertion. -/
theorem boundedRankReversibility_complement
    {I J E L : Type} [Countable I] [Countable J] [Countable E] [Countable L]
    [Nonempty I] [Nonempty J] [Nonempty E]
    (Phi : Channel I J) (Psi : Channel I E) (hComp : Complementary Phi Psi)
    (rho : L → State I) (r : ℕ) (hr : 0 < r)
    (hComplete : CompleteFamily (Set.range rho))
    (hRank : ∀ i, operatorRank (boundedOperator (rho i).coefficients) ≤ r)
    (hRev : ReversibleOn Phi.val (Set.range rho)) :
    HasBoundedRankKraus Psi.val r (dimensionCount L * rankMultiplicity Phi.val r) ∧
    PartiallyEntanglementBreaking Psi.val r ∧ CountableSchmidtOutputs Psi.val r ∧
    (r = 1 → ∀ (v : L → Hilbert I), (∀ i, ‖v i‖ = 1 ∧ (rho i).coefficients = rankOne (v i)) →
      ∀ pi : L → ℝ, PositiveLaw pi → HasWhitenedForm Psi.val pi v (rankMultiplicity Phi.val 1)) := by sorry

end QAlgorithms.Papers.Independent.Gap48.P45

end
Source
Reversibility conditions for quantum channels and their applications, arXiv:1203.0262v3, PDF p. 8, Theorem 3 (seed paper result): bounded-rank reversibility and the pure-state normal form
Read-back

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

Let I,J,E,LI,J,E,LI,J,E,L be countable sets, with I,J,EI,J,EI,J,E nonempty, and put HA=ℓ2(A;C)H_A=\ell^2(A;\mathbb C)HA​=ℓ2(A;C). Write Ru(a,b)=u(a)u(b)‾R_u(a,b)=u(a)\overline{u(b)}Ru​(a,b)=u(a)u(b)​. A coefficient matrix XXX is called nuclear here if it has a representation Xab=∑nun(a)vn(b)‾X_{ab}=\sum_n u_n(a)\overline{v_n(b)}Xab​=∑n​un​(a)vn​(b)​, with un,vn∈HAu_n,v_n\in H_Aun​,vn​∈HA​ and ∑n∥un∥∥vn∥<∞\sum_n\|u_n\|\|v_n\|<\infty∑n​∥un​∥∥vn​∥<∞; let τ(X)\tau(X)τ(X) be the infimum of these sums, regarded as elements of [0,∞][0,\infty][0,∞], with infimum ∞\infty∞ when there is no representation. A state is such a nuclear matrix for which every finite quadratic form ∑a,b∈sz(a)‾Xabz(b)\sum_{a,b\in s}\overline{z(a)}X_{ab}z(b)∑a,b∈s​z(a)​Xab​z(b) is real and nonnegative, the real diagonal is summable, and its sum is 111; identify states with their coefficient matrices. A channel here preserves nuclear matrices, is additive and complex homogeneous on them, and, for every n≥0n\geq0n≥0, its entrywise amplification

(Φ⊗id)(X)(j,a),(j′,b)=Φ((X(i,a),(i′,b))i,i′)j,j′(\Phi\otimes\mathrm{id})(X)_{(j,a),(j',b)} =\Phi\bigl((X_{(i,a),(i',b)})_{i,i'}\bigr)_{j,j'}(Φ⊗id)(X)(j,a),(j′,b)​=Φ((X(i,a),(i′,b)​)i,i′​)j,j′​

takes states on I×{0,…,n−1}I\times\{0,\ldots,n-1\}I×{0,…,n−1} to states on J×{0,…,n−1}J\times\{0,\ldots,n-1\}J×{0,…,n−1}. Let Φ:I→J\Phi:I\to JΦ:I→J and Ψ:I→E\Psi:I\to EΨ:I→E be channels that are complementary in the following specific sense: there is a linear isometry V:HI→HJ×EV:H_I\to H_{J\times E}V:HI​→HJ×E​ such that, for every state XXX, Φ(X)\Phi(X)Φ(X) and Ψ(X)\Psi(X)Ψ(X) are respectively the sums over the second and first diagonal coordinates of the matrix

CV(X)p,q=∑a∑b(Vea)(p)Xab(Veb)(q)‾,C_V(X)_{p,q}=\sum_a\sum_b(Ve_a)(p)X_{ab}\overline{(Ve_b)(q)},CV​(X)p,q​=a∑​b∑​(Vea​)(p)Xab​(Veb​)(q)​,

where eae_aea​ is the coordinate unit vector. For any matrix XXX, let BXB_XBX​ be a chosen bounded operator with coordinate matrix XXX, if one exists, and zero otherwise; write b(T)ab=(Teb)(a)b(T)_{ab}=(Te_b)(a)b(T)ab​=(Teb​)(a), rk⁡(T)\operatorname{rk}(T)rk(T) for the complex dimension of its range, converted to N∪{∞}\mathbb N\cup\{\infty\}N∪{∞}, and dim⁡A=∣A∣\dim A=|A|dimA=∣A∣ with every infinite cardinal converted to ∞\infty∞. Define

T(X,C)=∑ab(CBX)aa,DΦ(C)=B(T(Φ(Eba),C))a,b,T(X,C)=\sum_a b(CB_X)_{aa},\qquad D_\Phi(C)=B_{(T(\Phi(E_{ba}),C))_{a,b}},T(X,C)=a∑​b(CBX​)aa​,DΦ​(C)=B(T(Φ(Eba​),C))a,b​​,

where EbaE_{ba}Eba​ is the matrix unit. Let OΦO_\PhiOΦ​ be the closed linear span of the ranges of BΦ(X)B_{\Phi(X)}BΦ(X)​ over all input states, let QΦQ_\PhiQΦ​ be its orthogonal projection, and set

mr=min⁡ ⁣(dim⁡Cspan⁡{C:QΦCQΦ=C, DΦ(C)=0}+r2, dim⁡COΦ),m_r=\min\!\left( \dim_{\mathbb C}\operatorname{span}\{C:Q_\Phi C Q_\Phi=C,\ D_\Phi(C)=0\}+r^2,\ \dim_{\mathbb C}O_\Phi \right),mr​=min(dimC​span{C:QΦ​CQΦ​=C, DΦ​(C)=0}+r2, dimC​OΦ​),

again converting dimensions to N∪{∞}\mathbb N\cup\{\infty\}N∪{∞}. Suppose (ρi)i∈L(\rho_i)_{i\in L}(ρi​)i∈L​ is a family of states whose bounded-operator ranges together have dense linear span HIH_IHI​, every BρiB_{\rho_i}Bρi​​ has rank at most a positive integer rrr, and there is a channel R:J→I\mathcal R:J\to IR:J→I with R(Φ(ρi))=ρi\mathcal R(\Phi(\rho_i))=\rho_iR(Φ(ρi​))=ρi​ for every iii. Then there is a countable index set KKK and bounded operators Ak:HI→HEA_k:H_I\to H_EAk​:HI​→HE​, with dim⁡K≤(dim⁡L)mr\dim K\leq(\dim L)m_rdimK≤(dimL)mr​ and rk⁡(Ak)≤r\operatorname{rk}(A_k)\leq rrk(Ak​)≤r, such that ∑kAk∗Akx=x\sum_k A_k^*A_kx=x∑k​Ak∗​Ak​x=x for every xxx, and, for every nuclear XXX, the net over finite F⊆KF\subseteq KF⊆K satisfies

τ ⁣(Ψ(X)−∑k∈Fb(AkBXAk∗))⟶0.\tau\!\left(\Psi(X)-\sum_{k\in F}b(A_kB_XA_k^*)\right)\longrightarrow0.τ(Ψ(X)−k∈F∑​b(Ak​BX​Ak∗​))⟶0.

Moreover, for every countable nonempty set AAA and every state XXX on I×AI\times AI×A, (Ψ⊗idA)(X)(\Psi\otimes\mathrm{id}_A)(X)(Ψ⊗idA​)(X) belongs to the τ\tauτ-closure of the real convex hull of matrices RuR_uRu​, where ∥u∥=1\|u\|=1∥u∥=1 and u(e,a)=∑t=0r−1ft(e)gt(a)u(e,a)=\sum_{t=0}^{r-1}f_t(e)g_t(a)u(e,a)=∑t=0r−1​ft​(e)gt​(a) for some ft∈HE,gt∈HAf_t\in H_E,g_t\in H_Aft​∈HE​,gt​∈HA​; membership in this closure means approximation within every positive real tolerance in τ\tauτ. The same output also has a decomposition with real weights wn≥0w_n\geq0wn​≥0, ∑nwn=1\sum_n w_n=1∑n​wn​=1, and vectors unu_nun​ of norm 111, each having the displayed rrr-term tensor representation, such that the matrices wnRunw_nR_{u_n}wn​Run​​ sum to the output in the following coefficient trace sense: each summand and the output are nuclear coefficient matrices; their associated bounded operators, every finite partial sum, and its difference from the target admit norm-convergent nuclear expansions ∑q∣aq⟩⟨bq∣\sum_q|a_q\rangle\langle b_q|∑q​∣aq​⟩⟨bq​∣ with ∑q∥aq∥∥bq∥<∞\sum_q\|a_q\|\|b_q\|<\infty∑q​∥aq​∥∥bq​∥<∞; and the infimum of these costs for the finite-partial-sum differences tends to zero. The unit-vector requirement applies even to zero-weight terms. Finally, if r=1r=1r=1, then for every family of unit vectors vi∈HIv_i\in H_Ivi​∈HI​ with ρi=Rvi\rho_i=R_{v_i}ρi​=Rvi​​, and every real family πi>0\pi_i>0πi​>0 with unordered sum 111, put X=∑iπiRviX=\sum_i\pi_iR_{v_i}X=∑i​πi​Rvi​​, SX=cfc⁡(z↦Re⁡z)(BX)S_X=\operatorname{cfc}(z\mapsto\sqrt{\operatorname{Re}z})(B_X)SX​=cfc(z↦Rez​)(BX​), and let wiw_iwi​ be a chosen solution of SXwi=πiviS_Xw_i=\sqrt{\pi_i}v_iSX​wi​=πi​​vi​, wi⊥ker⁡SXw_i\perp\ker S_Xwi​⊥kerSX​, or zero if no such solution exists. These vectors satisfy those equations and ∑i⟨wi,x⟩wi=x\sum_i\langle w_i,x\rangle w_i=x∑i​⟨wi​,x⟩wi​=x for every xxx. There are vectors ξi,p∈HE\xi_{i,p}\in H_Eξi,p​∈HE​, indexed by p∈{n∈N:n<m1}p\in\{n\in\mathbb N:n<m_1\}p∈{n∈N:n<m1​}, with ∑p∥ξi,p∥2=1\sum_p\|\xi_{i,p}\|^2=1∑p​∥ξi,p​∥2=1 and ⟨ξi,t,ξi,p⟩=0\langle\xi_{i,t},\xi_{i,p}\rangle=0⟨ξi,t​,ξi,p​⟩=0 for p≠tp\ne tp=t, and matrices σi\sigma_iσi​, such that ∑pRξi,p=σi\sum_pR_{\xi_{i,p}}=\sigma_i∑p​Rξi,p​​=σi​ in the coefficient trace sense just specified and, for every nuclear YYY,

Ψ(Y)=∑iT(Y,BRwi)σi\Psi(Y)=\sum_i T(Y,B_{R_{w_i}})\sigma_iΨ(Y)=i∑​T(Y,BRwi​​​)σi​

in that same sense. No nonemptiness assumption on LLL is supplied separately; the dense-span hypothesis excludes an empty family because III is nonempty. Infinite sums in normed spaces are the total infinite sums, taking value zero when the family is not summable.

Human review
  • Endorsed by Shuze Chen · Oct 22, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Goku · Oct 22, 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