Lemma 1.1 — the inverse of is continuous at a point where is invertible
ProvedGlobalInexactNewton.Backtracking.lemma_1_1inexact-newtonnonlinear-equationsp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1
Let be a finite-dimensional real normed space and continuously differentiable. Let be a point at which the derivative is invertible. Then for every there is such that, for every with , is invertible and
where the norm on the left is the operator norm induced by the norm of .
This is the analytic tool that keeps uniformly bounded near a limit point at which is invertible, in the proofs of Theorems 3.3 and 6.1.
Formalization Note is ContinuousLinearMap.inverse of fderiv ℝ F y; invertibility is ContinuousLinearMap.IsInvertible. The neighbourhood is the open ball. Continuous differentiability is the paper's standing assumption (p. 395).
Preamble
import Mathlib import Definitions.Def_GlobalInexactNewton_Backtracking_Method open Filter Topology
Formal statement
namespace GlobalInexactNewton.Backtracking
/-- Eisenstat–Walker (1994), p. 395, Lemma 1.1: if `F'(x)` is invertible, then for every `ε > 0`
there is `δ > 0` such that `F'(y)` is invertible and `‖F'(y)⁻¹ - F'(x)⁻¹‖ < ε` whenever
`y ∈ N_δ(x)`. Standing assumption (p. 395): `F` is continuously differentiable. -/
theorem lemma_1_1 {E : Type*} [NormedAddCommGroup E] [NormedSpace ℝ E] [FiniteDimensional ℝ E]
(F : E → E) (hF : ContDiff ℝ 1 F) (x : E) (hx : (fderiv ℝ F x).IsInvertible)
(ε : ℝ) (hε : 0 < ε) :
∃ δ > 0, ∀ y ∈ Metric.ball x δ, (fderiv ℝ F y).IsInvertible ∧
‖(fderiv ℝ F y).inverse - (fderiv ℝ F x).inverse‖ < ε := by sorry
end GlobalInexactNewton.Backtracking
Source
Eisenstat and Walker, Globally Convergent Inexact Newton Methods, SIAM J. Optim. 4(2) (1994), p. 395, Lemma 1.1
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.