Nayak's theorem for compact operators

Abstract

Let $A$ be an $m\times m$ complex matrix and let $\lambda _1, \lambda _2, \ldots , \lambda _m$ be the eigenvalues of $A$ arranged such that $|\lambda _1|\geq |\lambda _2|\geq \cdots \geq |\lambda _m|$ and for $n\geq 1,$ let $s^{(n)}_1\geq s^{(n)}_2\geq \cdots \geq s^{(n)}_m$ be the singular values of $A^n$. Then a famous theorem of Yamamoto (1967) states that $$\lim _{n\to \infty}(s^{(n)}_j )^{\frac{1}{n}}= |\lambda _j|, ~~\forall \,1\leq j\leq m.$$ Recently S. Nayak strengthened this result very significantly by showing that the sequence of matrices $|A^n|^{\frac{1}{n}}$ itself converges to a positive matrix $B$ whose eigenvalues are $|\lambda _1|,|\lambda _2|,$ $\ldots , |\lambda _m|.$ Here this theorem has been extended to arbitrary compact operators on infinite dimensional complex separable Hilbert spaces. The proof makes use of Nayak's theorem, Stone-Weirstrass theorem, Borel-Caratheodory theorem and some technical results of Anselone and Palmer on collectively compact operators. Simple examples show that the result does not hold for general bounded operators.