Point interactions: \(\mathcal{PT}\)-Hermiticity and reality of the spectrum. (Q1609425)

Since the past a few years it has been discovered that some Hamiltonians \(L = -\frac{d^2}{dx^2} + V(x)\) in one dimension with the so-called \({\mathcal {PT}}\)-symmetry can be non-selfadjoint and have the real spectrum at the same time. This fact has been observed first through numerical computations by \textit{C. M. Bender} and \textit{S. Boettcher} and their collaborators, e.g. in [Phys. Rev. Lett. 80, 5243--4246 (1998; Zbl 0947.81018)] and approved through analytic calculations by M. Znojil and his collaborators, e.g. in \textit{G. Lévai} and \textit{M. Znojil} [J. Phys. A, Math. Gen. 33, No. 40, 7165--7180 (2000; Zbl 0963.81012)]. Here \({\mathcal P}\) stands for parity operation and \({\mathcal T}\) for time reversal operation. In the paper under review, the authors deal with this problem for the second derivative operators in one dimension with general point interactions possessing \({\mathcal {PT}}\)-symmetry, which are exactly solvable. In this paper, \({\mathcal P}\) is the space parity operator defined by \(({\mathcal P}\psi)(x) = \psi(-x)\), and \({\mathcal T}\) the complex conjugation defined by \({\mathcal T}\psi = \bar{\psi}\), for a function \(\psi(x)\) in one-dimensional space. The Schrödinger operator \(L\) in \(L^2({\mathbf R})\) as above is called \textit{\({\mathcal {PT}}\)--symmetric} if \(L^* \supseteq {\mathcal P}L{\mathcal P}\), and called \textit{\({\mathcal {PT}}\)-selfadjoint} if \(L^* = {\mathcal P}L{\mathcal P}\). As the main results, they have characterized the exactly solvable \({\mathcal {PT}}\)-selfadjoint second derivative operators with point interactions at the origin and determined their spectrum. They have also constructed \({\mathcal {PT}}\)-selfadjoint point interactions concentrated at different two points on the real line. It is further proved, although the relation between the two properties, the \({\mathcal {PT}}\)-symmetry of the Hamiltonian and the reality of its spectrum, was expected to be rather close, that one of these properties does not imply the other, but that there exist Hamiltonians with point interactions which are \({\mathcal {PT}}\)-symmetric and have real spectrum. Since the operators concerned are not selfadjoint operators, the classical von Neumann theory cannot be applied without modification, so that instead they use the method of boundary conditions, and show how to modify the notion of \({\mathcal {PT}}\)-symmetry in application to the extension theory of linear operators.