Renormalized oscillation theory for Hamiltonian systems (Q2400521): Difference between revisions

The well-known oscillation theorem for Sturm-Liouville problems can be formulated for the Dirichlet problem in the limit-point case on an interval \([a,\infty )\) as follows: for \(\lambda \in \mathbb{R}\) the dimension of the range of the spectral projection associated to the interval \((-\infty ,\lambda _0)\) equals the oscillation count of a solution of the Dirichlet problem for the parameter value \(\lambda _0\). Various generalizations, such as to Dirac operators, are also known. In this papers, the authors generalize the oscillation theorem to the case of Hamiltonian systems and any interval \((\lambda _0,\lambda _1)\) outside the essential spectrum, and if \(\lambda _0\) or \(\lambda _1\) are real numbers, respectively, then they do not belong to the spectrum of the operator. More precisely, let \(\mathcal I\) be a closed interval of the form \([a,b]\), \([a,\infty )\), or \((-\infty ,\infty )\), where \(a<b\) are real numbers. For a set \(X\) let \(X^{k\times l}\) denote the set of \(k\times l\) matrices with entries in \(X\). Let \(m\in \mathbb{N}\), \(0_m\) and \(I_m\) be the \(m\times m\) zero and identity matrices, \[ J=\begin{pmatrix} 0_m&-I_m\\I_m&0\end{pmatrix}, \] let \(1\leq r\leq 2m\), let \(W\in (L^1_{\text{loc}}(\mathcal I))^{r\times r}\), \(B\in (L^1_{\text{loc}}(\mathcal I))^{2m\times 2m}\), \(0<W(x)\), \(B(x)=B(x)^*\) for (almost all) \(x\in \mathcal {I}\), and introduce \(0 \leq A(x)\), \(C(x)\), \(E_r\in \mathbb{C}^{2m\times 2m} \) as \[ A(x)=\begin{pmatrix} W (x)& 0 \\0&0\end{pmatrix},\quad C(x)=\begin{pmatrix} W (x)^{-1}&0\\0&I_{2m-r}\end{pmatrix}, \quad E_r =\begin{pmatrix} I_r&0\\0&0\end{pmatrix}. \] The Hamiltonian system is the differential equation \[ J\Psi'(z,x)=[zA(x)+B(x)]\Psi (z,x),\quad x\in \mathcal{I},\;z\in \mathbb{C}, \] where \(\Psi (z,\cdot )\in (AC_{\text{loc}}(\mathcal{I}))^{2m\times \ell}\) with \(1\leq \ell\leq2m\). It is assumed that the limit case prevails at infinite endpoints. If boundary conditions are imposed at a finite endpoint or an interior point for a restriction operator, then they are of the form \(\gamma ^*JG(a)=0\) where \(\gamma =\begin{pmatrix} \gamma _1&\gamma _2 \end{pmatrix}^{\intercal}\in \mathbb{C}^{2m\times m}\) is such that \[ \begin{pmatrix} \gamma _1&-\gamma _2\\\gamma _2&\gamma _1 \end{pmatrix} \text{ is a unitary }\mathbb{C}^{2m\times 2m} \text{ matrix.} \] Let \(T\) be the operator in \((L^2(I))^{2m}\) associated with the Hamiltonian system and boundary conditions at finite endponts as described above. Under some mild technical conditions, the following is shown: Theorem. Let \(\lambda _0,\lambda _1\in \mathbb{R}\setminus \sigma (T)\) such that \((\lambda _0,\lambda _1)\cap \sigma _{\text{ess}}(T)=\emptyset \). Then \[ N((\lambda _0,\lambda _1);T)=\sum _{x\in \text{int}(\mathcal{I})}N(\Psi _+(\lambda _0,x,x_0)^*J\Psi _-(\lambda _1,x_,x_0)). \] Here, \(N((\lambda _0,\lambda _1);T)\) is the sum of the multiplicities of the at most finitely many eigenvalues of \(T\) in \((\lambda _0,\lambda _1)\), \(\Psi _\pm(\lambda _0,x,x_0)\) are certain Weyl-Titchmarsh solutions of the Hamiltonian system and \(N(\Psi _+(\lambda _0,x,x_0)^*J\Psi _-(\lambda _1,x,x_0))\) is the nullity of the \(m\times m\) matrix \(\Psi _+(\lambda _0,x,x_0)^*J\Psi _-(\lambda _1,x,x_0)\).