An optimal increasing estimate of the iterated Maslov-type indices (Q5955905)

The authors study the linear Hamiltonian system on \(\mathbb{R}^{ 2n}\), given by \(y'=JB(t)y\), where \(B(t)\) is a symmetric, \(T\)-periodic and continuous matrix and \(J\) is defined in the standard way. For the above Hamiltonian system a Maslov-type index has been defined by amongst others Conley, Zehnder and the second author. It consists of a pair of numbers \((i_T (B)\), \(\nu_T(B))\), representing respectively the index part and the nullity part. As a \(T\)-periodic matrix is also \(kT\)-periodic, the numbers \((i_{kT}, \nu_{k T})\) can be defined in a similar way. It was shown by the second author in [Pac. J. Math. 187, 113-149 (1999; Zbl 0924.58024)] that \(\lim_{k\to+ \infty} {i_{kT} \over k}\) exists and equals a finite real number, called the mean Maslov-type index and denoted by \(\widehat i_T\). The main result of the paper then shows that for all natural numbers \(k\), we have the inequality: \(k\widehat i_T-n\leq i_{kT}\leq k\widehat i_T+n-\nu_{kT}\). Note that this inequality has been extented by the same authors in a more recent paper, see \textit{C. Liu} and \textit{Y. Long}, J. Differ. Equations 165, 355-376 (2000; Zbl 0961.37017).