Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

§4.1, p. 25 — an ε-biased implementation of a two-outcome POVM {M, I − M} gives the (1, a + 1, ε)-block-encoding U′†(I ⊗ CNOT)U′ of M

Proved
QuantumSDP18.MeasureBE.povm_block_encoding

by mikedeng1 · Oct 15, 2026 · Mathlib 0df444a (Lean v4.33.1)

block-encodingmeasurementp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1povmquantum-information

Let {M,I−M}\{M, I-M\}{M,I−M} be a two-outcome POVM on a finite-dimensional system, so MMM and I−MI - MI−M are positive semidefinite. Suppose the measurement is implemented on a quantum computer as follows: attach aaa ancilla qubits in ∣0⟩⊗a|0\rangle^{\otimes a}∣0⟩⊗a, apply a unitary UUU, and accept when the last ancilla qubit reads ∣0⟩|0\rangle∣0⟩. Suppose that this implementation has bias at most ε≥0\varepsilon\ge 0ε≥0, that is, for every density matrix ρ\rhoρ on the system,

∣Tr⁡((I⊗∣0⟩⟨0∣) U(ρ⊗∣0⟩⟨0∣⊗a)U†)−Tr⁡(Mρ)∣≤ε.(14)\Big|\operatorname{Tr}\Big((I\otimes|0\rangle\langle 0|)\,U\big(\rho\otimes|0\rangle\langle 0|^{\otimes a}\big)U^\dagger\Big) - \operatorname{Tr}(M\rho)\Big| \le \varepsilon. \tag{14}​Tr((I⊗∣0⟩⟨0∣)U(ρ⊗∣0⟩⟨0∣⊗a)U†)−Tr(Mρ)​≤ε.(14)

Let a′=a+1a' = a + 1a′=a+1, U′=U⊗I2U' = U\otimes I_2U′=U⊗I2​ (one fresh qubit), and let CNOT\mathrm{CNOT}CNOT be controlled by the measured qubit and target the fresh qubit. Then

U′†(I⊗CNOT)U′ is a (1,a′,ε)-block-encoding of M,U'^\dagger(I\otimes\mathrm{CNOT})U' \text{ is a } (1, a', \varepsilon)\text{-block-encoding of } M,U′†(I⊗CNOT)U′ is a (1,a′,ε)-block-encoding of M,

that is, it is unitary and ∥M−(⟨0∣⊗a′⊗I) U′†(I⊗CNOT)U′ (∣0⟩⊗a′⊗I)∥≤ε\big\|M - (\langle 0|^{\otimes a'}\otimes I)\,U'^\dagger(I\otimes\mathrm{CNOT})U'\,(|0\rangle^{\otimes a'}\otimes I)\big\| \le \varepsilon​M−(⟨0∣⊗a′⊗I)U′†(I⊗CNOT)U′(∣0⟩⊗a′⊗I)​≤ε in the operator norm (Definition 2).

The result says that any reasonable implementation of a measurement yields, at essentially no extra cost, the block-encoding of its POVM element that the paper's quantum operator input model takes as input.

Formalization Note The system is an arbitrary finite type; a=c+1≥1a = c + 1 \ge 1a=c+1≥1 ancillas, the last of which is measured. The block-encoding predicate is the platform's QAlgorithms.IsBlockEncoding, which takes the ancilla register first, as Definition 2 does; the circuit (written system-first, as on p. 25) is reindexed accordingly, and the all-zero element of the a′a'a′ ancillas is (∣0⟩⊗a,0)(|0\rangle^{\otimes a}, 0)(∣0⟩⊗a,0). The hypothesis ε≥0\varepsilon \ge 0ε≥0 is added; it matters only for a zero-dimensional system, where (14) is vacuous. (14) ranges over all density matrices, as the page's "∀ρ\forall\rho∀ρ" over mixed states does.

Preamble
import Mathlib
import Definitions.Def_QAlgorithms_Defs_BlockEncoding
import Definitions.Def_QuantumSDP18_MeasureBE_Setting
Formal statement
namespace QuantumSDP18.MeasureBE

open scoped ComplexOrder

/-- §4.1, p. 25: if a two-outcome POVM `{M, I − M}` is implemented by attaching `a` ancilla
qubits in `|0⟩`, applying a unitary `U` and accepting when the last qubit reads `0`, with bias at
most `ε` (equation (14)), then `U′† (I ⊗ CNOT) U′` with `U′ = U ⊗ I₂` is a
`(1, a + 1, ε)`-block-encoding of `M` (Definition 2, p. 4; ancilla register moved first). -/
theorem povm_block_encoding {ι : Type*} [Fintype ι] [DecidableEq ι] (c : ℕ)
    (U : Matrix (ι × Anc c) (ι × Anc c) ℂ) (M : Matrix ι ι ℂ) (ε : ℝ)
    (hU : U ∈ Matrix.unitaryGroup (ι × Anc c) ℂ) (hM : M.PosSemidef) (hIM : (1 - M).PosSemidef)
    (hε : 0 ≤ ε)
    (h14 : ∀ ρ : Matrix ι ι ℂ, IsDensity ρ → ‖accProb c U ρ - (M * ρ).trace‖ ≤ ε) :
    QAlgorithms.IsBlockEncoding ((ancZero c, false) : Anc c × Bool)
      (Matrix.reindex (reorder c) (reorder c)
        ((Uprime c U).conjTranspose * cnotLast (ι := ι) c * Uprime c U)) 1 ε M := by sorry

end QuantumSDP18.MeasureBE
Source
van Apeldoorn and Gilyén, Improvements in quantum SDP-solving with applications, arXiv:1804.05058v1, p. 25, §4.1, 'therefore U′†(I ⊗ CNOT)U′ is a (1, a′, ε)-block-encoding of M' (setting pp. 24–25, (14); Definition 2, p. 4)
Human review
  • Endorsed by Shuze Chen · Oct 15, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 15, 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