Boundedness of the numerical range (Q922618)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Boundedness of the numerical range
scientific article

    Statements

    Boundedness of the numerical range (English)
    0 references
    0 references
    1990
    0 references
    Let A be an \(n\times n\) complex matrix. Define \(W(A)=\{x^*Ax:\) \(x^*x=1\), \(x\in {\mathbb{C}}^ n\}\). Given an \(n\times n\) Hermitian matrix H, denote the minimum (maximum) eigenvalue of H by \(\lambda_ 1(H)(\lambda_ n(H)\) respectively). By identifying the complex plane with \(R^ 2\), it is well-known and very easy to show that \(W(A)\subseteq [\lambda_ 1(H_ 1),\lambda_ n(H_ 1)]\times [\lambda_ 1(H_ 2),\lambda_ n(H_ 2)]\), where \(H_ 1=(A+A^*)/2\) and \(H_ 2=(A- A^*)/2i\). The author proves the above result for real A by a lengthy computation. However, in the theorem, the author incorrectly claims that \([\lambda_ 1(H_ 1),\lambda_ n(H_ 1)]\times [\lambda_ 1(H_ 2),\lambda_ n(H_ 2)]\) is the smallest rectangle containing W(A). A counterexample can be obtained by taking \(A=\left[\begin{matrix} 1&0 \\ 0&-1 \end{matrix}\right]\oplus \left[\begin{matrix} 0&1 \\ -1&0 \end{matrix}\right]\). Also, the author raises a conjecture which can easily be shown to be incorrect when \(n=4\). (Note: the matrix K used in the paper is unitarily similar to \(2H_ 2\oplus 2H_ 2\).)
    0 references
    boundedness of the numerical range
    0 references
    Hermitian matrix
    0 references
    minimum (maximum) eigenvalue
    0 references
    0 references

    Identifiers