On the Hu-Hurley-Tam conjecture concerning the generalized numerical range

From MaRDI portal
Publication:1968757





Let \(m,n\in {\mathbb N}\), \(1\leqslant m\leqslant n\), and let \(H\) be a subgroup of the symmetric group \(S_m\) of degree \(m\). The generalized matrix function associated with the principal character of \(H\) on an \(m\times m\) matrix \(B=(b_{ij})\) is defined as \(d^H(B)=\sum_{\sigma\in H}\prod^m_{j=1} b_{j\sigma(j)}\) and the generalized numerical range of an \(n\times n\) matrix \(A\) associated with \(d^H\) is defined as \[ W^H(A)=\{d^H(V^*AV): V\;\text{is} n\times m\;\text{such that} V^V=I_m\}. \] \textit{S.-A. Hu, J. F. Hurley} and \textit{T.-Y. Tam} [Linear Algebra Appl. 180, 121-131 (1993; Zbl 0776.15016)] constructed a class of normal matrices \(A\) with non-convex \(W^H(A)\) and made the following conjecture: Suppose \(H=S_m\), \(2\leqslant m\leqslant n\) with \((m,n)\neq (2,2)\). Let \(A\in M_n({\mathbb C})\) be a normal matrix. Then \(W^H(A)\) is convex if and only if \(A\) is a multiple of a Hermitian matrix. In the present paper the authors give counterexamples showing that the conjecture is not true when \(m<n\) and prove that it is valid under some more restrictive assumptions.











This page was built for publication: On the Hu-Hurley-Tam conjecture concerning the generalized numerical range

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1968757)