Constructing quantum observables and self-adjoint extensions of symmetric operators. I (Q946540)

The authors aim to give a proper choice of the domain for a quantum-mechanical observable that makes it a self-adjoint (SA) operator. They derive a theorem based on the first and second von Neumann Theorem. Main Theorem: Let \(\widehat f\) be a symmetric operator with a domain \(D_f\) in a Hilbert space \(H\), \(\widehat f\subseteq\widehat f^+\), where \(\widehat f^+\) is the adjoint operator of \(\widehat f\). Deficient subspaces of \(\widehat f: \chi_{\overline z}= \{\xi_z:\widehat f^+\xi_z= z\xi_z\}\) and \(\chi_z= \{\xi_{\overline z}:\widehat f^+\xi_{\overline z}= \overline z\xi_{\overline z}\}\), where \(z= x+ iy\), \(y> 0\). Deficiency indices of \(\widehat f: m_+= \dim\chi_{\overline z}\) and \(m_-= \dim_z\). (a) The operator \(\widehat f\) has SA extensions \(\widehat f_U=\widehat f^+_U\), \(\widehat f\subseteq\widehat f_U\), iff \(m_+= m_-= m\). (b) If \(m_+= m_-=0\), the operator \(\widehat f\) is essentially SA. (c) If \(m\neq 0\), there exists an \(m^2\)-parameter family \(\{\widehat f_U\}\) of SA extensions. Each SA extension \(\widehat f_U\) is defined by an isometric mapping \(\widehat U:\chi_{\overline z}\to \chi_z\), and \(D_{fU}= D_f+ (\widehat I+\widehat U)\chi_{\overline z}\). (d) The SA extension \(\widehat f_U\) can be defined as an SA restriction of the adjoint \[ \widehat f^+: D_{fU}= \{\xi_U\in D_{f^+}: \omega_*((\widehat I+ \widehat U)\eta_z, \xi_U)= 0,\,\forall \eta_z\in \chi_{\overline z}\}, \] where \[ \omega_*(\eta_*,\xi_*)= \langle\eta_*, \widehat f^+\xi_*\rangle- \langle\widehat f^+\eta_*, \xi_*\rangle,\;\xi_*,\;\eta_*\in D_{f^+}. \] Their results are applied to the differential operator \(\widehat p\) from \(id/dx\) in Hilbert spaces \(L^2(a, b)\). If \((a, b)=(-\infty,\infty)\), \(m_+= m_-= 0\). If \((a, b)\to [a, b]= [0,\ell]\), \(\ell<\infty\), \(m_+= m_-= m= 1\) and the domain \(D_{fU}\) of the \(\widehat f_U\) in the 1-parameter family \(\{f_{U(\theta)}\}\), \(\widehat Ue_+=\widehat U(\theta) e_+= e^{i\theta} e_-\), is given by using (d) for \(\widehat f= \widehat p\), where \(e_+= e^{-\kappa x}\) and \(e_-= e^{-\kappa(\ell-x)}\). [For part two, see the review (2007; Zbl 1152.81025).]