Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Corollary 18: singular value transformation by real polynomials

Proved
QSVT19.Main.corollary_18

by mikedeng1 · Oct 18, 2026 · Mathlib 0df444a (Lean v4.33.1)

linear-algebrap2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1quantum-computingquantum-signal-processingsingular-value-transformation

Let UUU be a unitary and Π,Π~\Pi,\widetilde\PiΠ,Π orthogonal projectors on a finite-dimensional Hilbert space. Let n≥1n\ge1n≥1 and let Pℜ∈R[x]P_\Re\in\mathbb R[x]Pℜ​∈R[x] be a polynomial of degree nnn with parity n mod 2n\bmod2nmod2 and ∣Pℜ(x)∣≤1|P_\Re(x)|\le1∣Pℜ​(x)∣≤1 for all x∈[−1,1]x\in[-1,1]x∈[−1,1]. Then there is Φ∈Rn\Phi\in\mathbb R^nΦ∈Rn such that, for every singular value decomposition of Π~UΠ\widetilde\Pi U\PiΠUΠ, Pℜ(SV)(Π~UΠ)={(⟨+∣⊗Π~)(∣0⟩⟨0∣⊗UΦ+∣1⟩⟨1∣⊗U−Φ)(∣+⟩⊗Π)if n is odd,(⟨+∣⊗Π)(∣0⟩⟨0∣⊗UΦ+∣1⟩⟨1∣⊗U−Φ)(∣+⟩⊗Π)if n is even.P_\Re^{(SV)}(\widetilde\Pi U\Pi)=\begin{cases}\bigl(\langle+|\otimes\widetilde\Pi\bigr)\bigl(|0\rangle\langle0|\otimes U_\Phi+|1\rangle\langle1|\otimes U_{-\Phi}\bigr)\bigl(|+\rangle\otimes\Pi\bigr)&\text{if $n$ is odd},\\[2pt] \bigl(\langle+|\otimes\Pi\bigr)\bigl(|0\rangle\langle0|\otimes U_\Phi+|1\rangle\langle1|\otimes U_{-\Phi}\bigr)\bigl(|+\rangle\otimes\Pi\bigr)&\text{if $n$ is even.}\end{cases}Pℜ(SV)​(ΠUΠ)={(⟨+∣⊗Π)(∣0⟩⟨0∣⊗UΦ​+∣1⟩⟨1∣⊗U−Φ​)(∣+⟩⊗Π)(⟨+∣⊗Π)(∣0⟩⟨0∣⊗UΦ​+∣1⟩⟨1∣⊗U−Φ​)(∣+⟩⊗Π)​if n is odd,if n is even.​ Here Pℜ(SV)P_\Re^{(SV)}Pℜ(SV)​ is the singular value transformation of Definition 16, UΦU_\PhiUΦ​ the alternating phase modulation sequence (31), and ∣+⟩=(∣0⟩+∣1⟩)/2|+\rangle=(|0\rangle+|1\rangle)/\sqrt2∣+⟩=(∣0⟩+∣1⟩)/2​ on one ancilla qubit.

Every bounded real polynomial of definite parity can thus be applied to the singular values of a block of a unitary, with one ancilla qubit and nnn uses of UUU and U†U^\daggerU†; this is the identity behind the paper's applications.

Formalization Note The hypothesis n≥1n\ge1n≥1 is added: for n=0n=0n=0 a constant Pℜ≡c≠1P_\Re\equiv c\ne1Pℜ​≡c=1 satisfies the conditions, its transformation is cΠc\PicΠ, while the right-hand side is Π\PiΠ. The ancilla is the first tensor factor; Φ\PhiΦ is chosen before the decomposition.

Preamble
import Definitions.Def_QSVT19_Main_Setting

set_option autoImplicit false

open scoped ComplexOrder
open Polynomial Matrix
Formal statement
namespace QSVT19.Main

/-- Gilyén, Su, Low, Wiebe, arXiv:1806.01838v1, Corollary 18 (Singular value transformation by real polynomials), p. 19: "Let U, Π, Π̃ be
as in Theorem 17. Suppose that P_ℜ ∈ R[x] is a degree-n polynomial satisfying that
• P_ℜ has parity-(n mod 2) and • for all x ∈ [−1, 1]: |P_ℜ(x)| ≤ 1.
Then there exist Φ ∈ R^n, such that
P_ℜ^{(SV)}(Π̃UΠ) = (⟨+| ⊗ Π̃)(|0⟩⟨0| ⊗ U_Φ + |1⟩⟨1| ⊗ U_{−Φ})(|+⟩ ⊗ Π) if n is odd, and
(⟨+| ⊗ Π)(|0⟩⟨0| ⊗ U_Φ + |1⟩⟨1| ⊗ U_{−Φ})(|+⟩ ⊗ Π) if n is even. (33)"

**Formalization Note.** `H_U` is `ι → ℂ` for a finite type `ι`, with inner product
`star v ⬝ᵥ w`; `Π, Π̃` are `Proj, Projt`; "unitary" is `U ∈ Matrix.unitaryGroup ι ℂ` and
"orthogonal projector" is `IsStarProjection` (self-adjoint idempotent). "degree-n" is `natDegree = n`; the parity of `P_ℜ` is that of its image in `ℂ[X]`.
`P_ℜ^{(SV)}(Π̃UΠ)` is Definition 16 applied to `x ↦ P_ℜ(x)` over a singular value decomposition
`S` of `Π̃UΠ` (`svtOdd` for odd `n`, `svtEven` for even `n`); the identity is asserted for every
such decomposition, with `Φ` chosen before it (as `P^{(SV)}` does not depend on the
decomposition). `U_Φ` is `altPhaseSeq`, `⟨+| ⊗ M`, `|+⟩ ⊗ M` and `|0⟩⟨0| ⊗ V + |1⟩⟨1| ⊗ W` are
`plusBra`, `plusKet`, `ctrlSel`, with the ancilla qubit as the first tensor factor. The
hypothesis `1 ≤ n` is added: at `n = 0`, `P_ℜ ≡ c` with `c ≠ 1` satisfies the bullets, its
transformation is `cΠ`, while `Φ ∈ R^0` gives `U_Φ = I` and the right-hand side `Π`. -/
theorem corollary_18 {ι : Type*} [Fintype ι] [DecidableEq ι] (U Proj Projt : Matrix ι ι ℂ) (hU : U ∈ Matrix.unitaryGroup ι ℂ) (hProj : IsStarProjection Proj) (hProjt : IsStarProjection Projt) (n : ℕ) (hn : 1 ≤ n) (PR : ℝ[X]) (hdeg : PR.natDegree = n) (hpar : QAlgorithms.HasParity (PR.map Complex.ofRealHom) n) (hbd : ∀ x ∈ Set.Icc (-1 : ℝ) 1, |PR.eval x| ≤ 1) :
    ∃ Φ : Fin n → ℝ, ∀ S : ProjUnitarySVD U Proj Projt, (Odd n → svtOdd (fun x : ℝ => ((PR.eval x : ℝ) : ℂ)) S = plusBra Projt * ctrlSel (altPhaseSeq U Proj Projt Φ) (altPhaseSeq U Proj Projt (-Φ)) * plusKet Proj) ∧ (Even n → svtEven (fun x : ℝ => ((PR.eval x : ℝ) : ℂ)) S = plusBra Proj * ctrlSel (altPhaseSeq U Proj Projt Φ) (altPhaseSeq U Proj Projt (-Φ)) * plusKet Proj) := by sorry

end QSVT19.Main
Source
Gilyén, Su, Low, Wiebe, arXiv:1806.01838v1, p. 19, Corollary 18, (33)
Human review
  • Endorsed by Shuze Chen · Oct 18, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 18, 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