Local Lipschitz Continuity Of Operator Inverse Square Roots
ProvedFunctionalInequalities.Analysis.locallyLipschitz_inverseSqrt_of_contDiff_ringInverse_readableThis statement establishes local Lipschitz continuity of a principal inverse-square-root operator profile.
Let E be a finite-dimensional real inner-product space and P a real normed space. Let A:P→(E→L[ℝ]E). Assume
- is positive for every p;
- A(p) is a unit for every p; and
- the map is C¹.
Then
is locally Lipschitz in the operator norm, where the inverse square root is the Lean definition inverseSqrt.
This is a reusable regularity statement for parameter-dependent positive invertible operators, including localization covariance profiles.
Formalization Note The parameter space has a normed additive-group and real normed-space structure; no finite-dimensional assumption on P is present.
Original Lean declaration header
Within namespace FunctionalInequalities.Analysis. Shown in its original module context. Proof omitted.
theorem locallyLipschitz_inverseSqrt_of_contDiff_ringInverse {P : Type*} [NormedAddCommGroup P]
[NormedSpace ℝ P] (A : P → (E →L[ℝ] E))
(hpos : ∀ p, (Ring.inverse (A p)).IsPositive)
(hunit : ∀ p, IsUnit (A p))
(hInverse : ContDiff ℝ 1 (fun p => Ring.inverse (A p))) :
LocallyLipschitz (fun p => inverseSqrt (A p))
Module navigation
Module: Analysis/Operator/InverseSqrt.
Supporting theorems: locallyLipschitz_matrixEntries_sqrt.
Previous: locallyLipschitz_matrixEntries_sqrt.
Source excerpt, lines 223–228 · Frozen source 408915f.
import Definitions.Def_KLSModular_FunctionalInequalities_Analysis_Operator_InverseSqrt
variable {E : Type*} [NormedAddCommGroup E] [InnerProductSpace ℝ E] [FiniteDimensional ℝ E]
theorem FunctionalInequalities.Analysis.locallyLipschitz_inverseSqrt_of_contDiff_ringInverse_readable {P : Type*} [NormedAddCommGroup P]
[NormedSpace ℝ P] (A : P → (E →L[ℝ] E))
(hpos : ∀ p, (Ring.inverse (A p)).IsPositive)
(hunit : ∀ p, IsUnit (A p))
(hInverse : ContDiff ℝ 1 (fun p => Ring.inverse (A p))) :
LocallyLipschitz (fun p => inverseSqrt (A p)) := by sorry