Documentation

LeanPool.OperatorTheory.Operator.Crouzeix.NormalProductSymmetrized

Universal circle symmetrization implies the normal auxiliary product bound #

For any operator, the sharp symmetrized estimate on every polynomial forces an enclosing circle's center into the closed numerical range. Indeed, if the center were outside, an affine polynomial separator would have value one at the center and norm at most r < 1 on the closed numerical range. Applying symmetrization to its powers makes the functional-calculus term tend to zero by the unconditional Crouzeix--Palencia bound, while the circle auxiliary remains the identity, a contradiction.

The aligned-circle normal product theorem then supplies the literal sharp auxiliary product estimate.

Main declarations #

Universal sharp symmetrization on an enclosing circle, controlled by a compact nonempty convex set, forces the circle center into that set. No independent polynomial-calculus bound is needed: if a separator power B is close to -1, its square is close to 1, while symmetrization of the squared power would make it close to -1.

theorem circle_center_mem_of_compactConvex_of_globalBound_of_symmetrization {E : Type u} [NormedAddCommGroup E] [InnerProductSpace ℂ E] [CompleteSpace E] [Nontrivial E] (A : E →L[ℂ] E) {K : Set ℂ} (hcompact : IsCompact K) (hnonempty : K.Nonempty) (hconvex : Convex ℝ K) (c : ℂ) {R : ℝ} (hA : ‖A - c • 1‖ < R) (C : ℝ) (_hC : 0 ≤ C) (_hcalc : ∀ (q : Polynomial ℂ), ‖(Polynomial.aeval A) q‖ ≤ C * polynomialSupNorm q K) (hsymm : ∀ (q : Polynomial ℂ), ‖(Polynomial.aeval A) q + star (crouzeixPolynomialAuxiliaryOperator A (SmoothJordanDomain.ball c R ⋯) q)‖ ≤ 2 * polynomialSupNorm q K) :
c ∈ K

A finite polynomial-calculus bound on a compact nonempty convex set, together with universal sharp symmetrization on an enclosing circle controlled by the same set, forces the circle center into that set. No normality or particular value of the finite constant is needed.

A finite global polynomial-calculus bound, together with universal sharp symmetrization on an enclosing circle, forces the circle center into the closed numerical range. No normality or particular value of the finite constant is needed.

Universal sharp symmetrization on an enclosing circle forces its center into the closed numerical range of an arbitrary operator.

Universal circle symmetrization makes the functional calculus uniformly power-bounded on the ideal of polynomials vanishing at the circle center. If p(c) = 0, every positive power of p(A) has norm at most twice the corresponding power of the sup norm on the compact control set.

Universal circle symmetrization preserves the operator/scalar coupling for every positive power after an arbitrary shift b. This is stronger than separately bounding the two summands and retains the cancellation that is relevant to the sharp product problem.

Universal circle symmetrization uniformly power-bounds the centered functional calculus. For every polynomial p, positive powers of p(A) - p(c)I are controlled by twice the corresponding power of the sup norm of p - p(c) on the compact control set.

Universal circle symmetrization alone confines the spectrum of the centered value p(A) - p(c)I to the disk whose radius is the sup norm of p - p(c) on the compact control set. No independent global polynomial-calculus estimate is assumed; the zero Hilbert space has empty spectrum.

Equivalently, universal circle symmetrization confines the spectrum of p(A) to the disk centered at p(c) with radius sup_K |p - p(c)|, including the zero-space case.

If the circle center belongs to the compact control set, the coupled shifted moments force spectrum p(A) into the sharp disk around every scalar b, with radius sup_K |p-b|. The proof keeps the shifted scalar power inside the moment and lets its norm be absorbed asymptotically; the zero Hilbert space is handled by its empty spectrum.

If the circle center belongs to the compact control set, universal circle symmetrization gives a sharp factor-one spectral-radius bound after every scalar shift.

On a compact nonempty convex control set, universal circle symmetrization supplies the center membership needed by the arbitrary-shift spectral localization theorem; the zero Hilbert space has empty spectrum.

Universal circle symmetrization over a compact nonempty convex control set gives a sharp factor-one spectral-radius bound after every scalar shift.

Universal circle symmetrization over a compact nonempty convex control set gives the sharp factor-one spectral-radius bound for every polynomial value. This is a spectral conclusion and therefore also covers the zero Hilbert space.

Universal circle symmetrization controlled by the closed numerical range places spectrum p(A) in every disk centered at b with radius sup_{closure W(A)} |p-b|; the zero Hilbert space has empty spectrum.

Universal circle symmetrization controlled by the closed numerical range gives a sharp factor-one spectral-radius bound after every scalar shift.

Universal circle symmetrization controlled by the closed numerical range gives the sharp factor-one spectral-radius bound for every polynomial value, including on the zero Hilbert space.

Universal circle symmetrization controlled by a compact nonempty convex set forces that set to contain the spectrum of A. An exterior spectral value would yield a polynomial separator equal to one there and strictly contractive on K, contradicting the corresponding zero-centered shifted spectral disk. The zero Hilbert space has empty spectrum and is handled separately.

The sharp auxiliary product bound holds without any normality assumption whenever the center value is small: if |p(c)| ≤ (sqrt 2 - 1) m, the same-polynomial symmetrized estimate bounds ‖p(A)‖ by 2m + |p(c)|, and the identity (sqrt 2 - 1)^2 + 2 (sqrt 2 - 1) = 1 closes the product estimate.

Universal circle symmetrization applied to p - C b controls p(A) after an arbitrary scalar shift. The symmetrized operator is exactly p(A) + (p(c) - 2b)I, giving the displayed quantitative estimate.

Universal circle symmetrization controlled by a compact nonempty convex set makes that set a 3-polynomial spectral set. Spectrum containment comes from shifted moments and polynomial separation; the norm estimate is the zero-shift bound 2m + |p(c)| ≤ 3m after center alignment. The zero Hilbert space is discharged separately.

Shifting p by an arbitrary scalar before applying universal symmetrization gives a family of sharp-product criteria. The shift b controls p(A) through p(A) + (p(c) - 2b)I; the displayed scalar inequality is exactly what is needed after the triangle estimate.

The midpoint shift b = p(c) / 2 eliminates the scalar residual in the shifted symmetrization estimate. Thus the displayed bound on sup_K |p - p(c)/2| alone suffices for the literal sharp product.

Centering p gives an adaptive sharp-product criterion without normality. Universal symmetrization bounds p(A) - p(c)I by twice the sup norm of p - p(c); hence the displayed scalar inequality suffices for the literal auxiliary product bound.

The sharp auxiliary product bound also holds without normality in the near-constant regime. If the sup norm of p - p(c) is at most m - |p(c)|, universal symmetrization bounds the centered operator by twice that variation, and (m - |p(c)|)^2 ≥ 0 closes the exact product estimate.

For an arbitrary operator, universal sharp symmetrization on an enclosing circle implies an auxiliary product bound with the global 1 + sqrt 2 Crouzeix--Palencia factor. The universal family first aligns the circle center with the closed numerical range; scalar evaluation is then sharp, while polynomial evaluation uses the unconditional global bound.

Universal sharp symmetrization on an enclosing circle implies the literal sharp auxiliary product bound for every polynomial whose individual value p(A) is star-normal. The ambient operator need not be star-normal.

In the star-normal branch, universal sharp symmetrization on an enclosing circle implies the literal sharp auxiliary product bound for every polynomial.