Operators with numerical range in a conic domain (Q2373353)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Operators with numerical range in a conic domain |
scientific article |
Statements
Operators with numerical range in a conic domain (English)
0 references
19 July 2007
0 references
For a bounded linear operator \(A\) on a complex Hilbert space \(H\), the numerical range \(W(A)\) of \(A\) is the subset \(\{\langle Ax,x\rangle: x\in H,\| x\|= 1\}\) of the plane. This paper is concerned with the problem of finding, for a convex open subset \(\Omega\) of the plane, the smallest constant \(C(\Omega)\) such that for every \(A\) with \(\overline{W(A)}\subseteq\Omega\) and every rational function \(r\) with poles off \(\Omega\), the inequality \(\| r(A)\|\leq C(\Omega)\sup_{z\in\Omega}|r(z)|\) holds. A result of von Neumann says that if \(\Omega\) is a half-plane, then \(C(\Omega)= 1\). For bounded \(\Omega\), \textit{B.}\ and \textit{F.\,Delyon} [Bull.\ Soc.\ Math.\ Fr.\ 127, No.\,1, 25--41 (1999; Zbl 0937.47004)] showed the finiteness of \(C(\Omega)\) and the second author obtained in [J.~Funct.\ Anal.\ 244, No.\,2, 668--690 (2007; Zbl 1116.47004)] that \(C(\Omega)\leq 11.08\). In the paper under review, the authors consider convex domains with conic boundaries. They show that if \(\Omega\) is an elliptic disc with eccentricity \(e\), then \(C(\Omega)\leq 2+(2/\sqrt{4- e^2})\). In the limiting case, they obtain that \(C(\Omega)\leq 2+ (2/\sqrt{3})\) for \(\Omega\) with boundary a parabola. This improves a previous estimate by the second author. When \(\Omega\) is a convex domain whose boundary is a branch of a hyperbola with asymptotics forming the angle \(2\alpha\) \((0<\alpha<\pi/2)\), they prove that \(C(\Omega)\leq 2(1- (\alpha/\pi))+ u(\alpha)\), where \(u(\alpha)\) is a certain improper integral involving \(\alpha\). The limiting case gives an estimate for the constant of the sector formed by two lines at angle \(2\alpha\), which improves a previous estimate when \(0<\alpha< 0.22\pi\). The proofs for both main results follow the general approach of the second author for bounded \(\Omega\) and make some meticulous refinements by taking full advantage of the conic boundary.
0 references
numerical range
0 references
field of values
0 references
spectral sets
0 references