Lemma 17, p. 19 — (1 − (1 − x²)^b)/x is ε-close to 1/x on D_{κd} for every integer b ≥ (κd)² log(κd/ε)
ProvedQuantumLinSys.Chebyshev.lemma_17p2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1polynomial-approximationquantum-linear-systemsuniform-approximation
Let , let be an integer (the sparsity of the matrix), let , and let be an integer with
Then the function of (74) is -close to on :
Multiplying by removes the singularity at the origin while changing the function by at most on the domain where the eigenvalues of the rescaled matrix lie; the result is a polynomial, which Lemma 18 expands in Chebyshev polynomials.
Formalization Note (a condition number) and (a sparsity) are made explicit; they make nonempty and keep out of it. The page's "integer " is a natural number: is "a polynomial of degree " (p. 19), and a negative exponent would make singular at .
Preamble
import Mathlib import Definitions.Def_QuantumLinSys_Chebyshev_Setting
Formal statement
namespace QuantumLinSys.Chebyshev
theorem lemma_17 (κ ε : ℝ) (d b : ℕ) (hκ : 1 ≤ κ) (hd : 1 ≤ d) (hε : 0 < ε)
(hb : (κ * d) ^ 2 * Real.log (κ * d / ε) ≤ b) :
∀ x ∈ Dκ (κ * d), |fTamed b x - 1 / x| ≤ ε := by sorry
end QuantumLinSys.Chebyshev
Source
Childs, Kothari and Somma, Quantum algorithm for systems of linear equations with exponentially improved dependence on precision, arXiv:1511.02306v2, p. 19, Lemma 17 (eq. (74))
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.