The inverse spectral problem for indefinite strings (Q295806)

A famous result in inverse spectral theory going back to M. G. Krein is the observation of the one-to-one mapping between the set of all vibrating strings and the class of Stieltjes functions by means of the Weyl-Titchmarsh function \(m\) associated with the string. In the present terminology the behavior of the string is characterized by the eigenvalue equation \[ -u'' = z \, u \, \omega \quad \text{on } [0,L) \] subject to certain boundary conditions, where \(\omega\) is a non-negative Borel measure on \([0,L)\) (the mass distribution of the string). Since the 1970s a number of approaches to indefinite generalizations was presented, so far allowing only settings with finitely many ``negative masses''. It was already observed that in order to cover a suitable class of functions \(m\) the eigenvalue equation must be generalized to \[ -u'' = z \, u \, \omega + z^2 u \, \nu \] with a non-negative Borel measure \(\nu\). The present paper seems to present the first ''completely indefinite'' approach allowing an arbitrary real-valued distribution \(\omega \in H^{-1}_{\mathrm{loc}}[0,L)\) (including signed measures). Then, using a left-definite framework the ''indefinite string'' given by \((L, \omega, \nu)\) is realized by a self-adjoint relation in a Hilbert space and an associated Weyl-Titchmarsh function \(m\) is constructed belonging to the class of Herglotz-Nevanlinna functions. The main result of the paper states that the mapping \((L, \omega, \nu) \rightarrow m\) defines a bijection between the set of all ``indefinite strings'' and this class. The essential tool for the proof is an inverse result of de Branges for Hamiltonian systems. The paper is accomplished by a continuity result for the mapping \((L, \omega, \nu) \rightarrow m\). Finally, Krein's classical result is embedded in the present theory.