On numerical ranges and roots

Existence of the fractional powers is established in Banach algebra setting, in terms of the numerical ranges of elements involved. The behavior of the spectra and (for Hermitian ∗-algebras satisfying some additional hypotheses) the ∗-numerical range under taking these powers also is investigated.     

Containment regions for zeros of polynomials from numerical ranges of companion matrices

Containment regions for the zeros of a monic polynomial are given with the aid of results for containment regions for the numerical range of certain bordered diagonal matrices which are applied to different types of companion matrices of the polynomial.     

Flat portions on the boundary of the numerical ranges of certain Toeplitz matrices

We give criterions for a flat portion to exist on the boundary of the numerical range of a matrix. A special type of Teoplitz matrices with flat portions on the boundary of its numerical range are constructed. We show that there exist 2 × 2 nilpotent matrices A₁,A₂, an n  × n nilpotent Toeplitz matrix N_n, and an n  × n cyclic permutation matrix S_n(s) such that the numbers of flat portions on the boundaries of W(A₁⊕N_n) and W(A₂⊕S_n(s)) are, respectively, 2(n - 2) and 2n.     

Congruence numerical ranges and their radii

The earliest congruence numerical range was introduced by Thompson in 1980. Since then, congruence numerical ranges and their radii have been studied by some authors. The purpose of this article is to discuss several aspects of this subject. We survey some known results and propose some problems for furthur study. A few new results are also presented.     

Congruence numerical ranges and their radii

The earliest congruence numerical range was introduced by Thompson in 1980. Since then, congruence numerical ranges and their radii have been studied by some authors. The purpose of this article is to discuss several aspects of this subject. We survey some known results and propose some problems for furthur study. A few new results are also presented.     

Some results on numerical ranges and factorizations of matrix polynomials

Some new factorization theorems for monic matrix polynomials are obtained. These theorems are based on the numerical range having the number of connected components equal to the degree of the polynomial. For second degree polynomials, sufficient conditions are given for the numerical range to have two connected components.     

Positive block matrices and numerical ranges

Any positive matrix M partitioned in four n-by-n blocks satisfies the unitarily invariant norm inequality $6M6\le 6M_{1,1}+M_{2,2}+\omega I6$, where ω is the width of the numerical range of $M_{1,2}$. Some related inequalities and a reverse Lidskii majorization are given.     

Geometry of higher-rank numerical ranges

We consider geometric aspects of higher-rank numerical ranges for arbitrary N  × N matrices. Of particular interest is the issue of convexity and a possible extension of the Toeplitz-Hausdorff Theorem. We derive a number of reductions and obtain partial results for the general problem. We also conduct graphical and computational experiments.

Added in proof: Following acceptance of this paper, our subject has developed rapidly. First, Hugo Woerdeman established convexity of the higher-rank numerical ranges by combining Proposition 2.4 and Theorem 2.12 with the theory of algebraic Riccati equations. See Woerdeman, H., 2007, The higher rank numerical range is convex, Linear and Multilinear Algebra, to appear. Subsequently Chi-Kwong Li and Nung-Sing Sze followed a different approach that not only yields convexity but also provides important additional insights. See Li, C.-K. and Sze, N.-S., 2007, Canonical forms, higher rank numerical ranges, totally isotropic subspaces, and matrix equations, preprint. See also Li, C.-K., Poon, Y.-T., and Sze, N.-S., 2007, Condition for the higher rank numerical range to be non-empty, preprint.     

Circularity of numerical ranges and block-shift matrices

Let A be a square complex matrix. Several characterizations are found for A to be permutationally similar to a block-shift matrix. One interesting equivalent condition is that the numerical range of every matrix with the same zero pattern as A is a circular disk. Equivalent conditions for the characteristic polynomial of the hermitian part of the matrix e ^iθA to be the same for all real values θ are also obtained.Complex matrices of order four that are unitarily similar to a block-shift matrix are identified. A result of Marcus and Pesce [6] is extended, and an open question of Li and Tsing [4] is also answered partially.     

Circularity of numerical ranges and block-shift matrices

Let A be a square complex matrix. Several characterizations are found for A to be permutationally similar to a block-shift matrix. One interesting equivalent condition is that the numerical range of every matrix with the same zero pattern as A is a circular disk. Equivalent conditions for the characteristic polynomial of the hermitian part of the matrix e^iθA to be the same for all real values θ are also obtained.Complex matrices of order four that are unitarily similar to a block-shift matrix are identified. A result of Marcus and Pesce [6] is extended, and an open question of Li and Tsing [4] is also answered partially.     

On a Compression of Normal Matrix Polynomials

In this article, we study a compression of normal matrices and matrix polynomials with respect to a given vector and its orthogonal complement. The numerical range of this compression satisfies special boundary properties, which are investigated in detail. The characteristic polynomial of the compression is also considered.     

Numerical ranges of cube roots of the identity

The numerical range of a bounded linear operator T on a Hilbert space H is defined to be the subset W(T)={〈Tv,v〉:v∈H,∥v∥=1} of the complex plane. For operators on a finite-dimensional Hilbert space, it is known that if W(T) is a circular disk then the center of the disk must be a multiple eigenvalue of T. In particular, if T has minimal polynomial z³-1, then W(T) cannot be a circular disk. In this paper we show that this is no longer the case when H is infinite dimensional. The collection of 3×3 matrices with three-fold symmetry about the origin are also classified.     

Numerical ranges of weighted shifts

Let A be a unilateral (resp., bilateral) weighted shift with weights wn, n?0 (resp., −∞<n<∞). Eckstein and Rácz showed before that A has its numerical range W(A) contained in the closed unit disc if and only if there is a sequence (resp., ) in [−1,1] such that ²|wn|=(1−an)(1+an+1) for all n. In terms of such an?s, we obtain a necessary and sufficient condition for W(A) to be open. If the wn?s are periodic, we show that the an?s can also be chosen to be periodic. As a result, we give an alternative proof for the openness of W(A) for an A with periodic weights, which was first proven by Stout. More generally, a conjecture of his on the openness of W(A) for A with split periodic weights is also confirmed.     

On the intrinsic and the spatial numerical range

For a bounded function f from the unit sphere of a closed subspace X of a Banach space Y, we study when the closed convex hull of its spatial numerical range W(f) is equal to its intrinsic numerical range V(f). We show that for every infinite-dimensional Banach space X there is a superspace Y and a bounded linear operator such that . We also show that, up to renormig, for every non-reflexive Banach space Y, one can find a closed subspace X and a bounded linear operator T∈L(X,Y) such that .Finally, we introduce a sufficient condition for the closed convex hull of the spatial numerical range to be equal to the intrinsic numerical range, which we call the Bishop-Phelps-Bollobás property, and which is weaker than the uniform smoothness and the finite-dimensionality. We characterize strong subdifferentiability and uniform smoothness in terms of this property.     

Numerical ranges and dilations

Denote by W(A) the numerical range of a bounded linear operator A. For two operators A and B (which may act on different Hilbert spaces), we study the relation between the inclusion relation W(A)⊆W(B) and the condition that A can be dilated to an operator of the form B⊗I. We also investigate the possibilities of dilating an operator A to operators with simple structure under the assumption that W(A) is included in a special region.