Self-adjoint extensions in quantum mechanics. General theory and applications to Schrödinger and Dirac equations with singular potentials. (Q665526)

The authors show here how quantization problems associated with the correct definition of observables and their spectral analysis can be studied, at the first step. Let \(\widehat f\) be a symmetric operator with domain \(D_f\) and adjoint\(\widehat f^+\), \(\widehat f\subseteq\widehat f^+\). They give self-adjoint (s.a.) extension \(\widehat f_U\) in Main Theorem. That is \(\widehat f_U\xi_U=\overline{\widehat f}\xi+ z\xi_z+\overline z\widehat U_z\), \(\xi_U=\underline\xi+ \xi_x+\widehat U\xi_z\), \(\widehat U:\chi_{\overline z}\to \chi_z\) (isometric map), \(\xi\in D_{\overline f}\), \(\xi_z\in\chi_{\overline z}= \ker\widehat f^+= \{\xi_z;\widehat f^+\xi_z= z\xi_z\}\). They arrive at the problem: (i) \(\vec H=\vec H+V(x)=-d^2_x+ V(x)\), \(V(x)=\overline V(x)\), \(\vec p=-id_x\); (ii) \( (\vec H-W)\psi(x)= 0\), \(x\in(a, b)\subset\mathbb R\), \(W\): complex constant, which is called the one-dimensional (stationary) Schrödinger equation. A guiding functional \(\phi(F;z)\) for an s.a. \(2\times 2\) matrix operator \(\widehat h_e\), associated with s.a. \(\vec h\) is given by \(\phi(F;z)= \int_{\mathbb R^+} U(r;z)F(r)\,dr\), \(F\in D_r(\mathbb R_+)\subset D(h_e)\), \(U=(u/\nu)\), \(U(r; z)F= uf+\nu g\), \((\vec h-z) U(r; z)= 0\). 1. \(\phi(F;z)\) is entire in \(z\), 2. If \(\phi(F_0;z_0)= 0\), \(F_0\in D\), \(z_0\in\mathbb R\), then there exist \(\psi\in D\) such that \((\vec h-z_0)\psi= F_0\), 3. \(\phi(\vec h_e F;z)= \int U(r; z)\cdot\vec h\,F\,dr= z\phi(F; z)\), \(F\in D\). If 1, 2, 3 hold, \(\phi\) is simple. They obtain \(U_\lambda= \rho(\lambda) U(r;\lambda)\), \(\lambda\in\Delta\), \(U_n= Q_nU(\lambda_n)\), \(n\in\mathbb N\), given by the Green's function \[ G: F(r)= \int_{\mathbb R_+} G(r,r';z)\psi(r')\,dr',\;\text{Im}\,z> 0; \] \[ U(c; \lambda)\otimes U(c;\lambda)\sigma'(\lambda)= \pi^{-1}\text{Im\,}G(c- 0,c+0; \lambda+ i0),\;\sigma'(\lambda)= \rho^2(\lambda)+ \sum_{n\in\mathbb N} Q^2_n\delta(\lambda- \lambda_n). \] Calogero solvable Hamiltonian is \(\vec H=-d^2_x+\lambda x^{-2}\), and the problem (ii) for the s.a. extension \(\vec H\) (in \(L^2(\mathbb R_+)\)) with \(V(x)=\alpha x^{-2}\), \(\alpha<-1/4\), is solved. The s.a. extension is given by using the asymmetry form \[ \Delta_{H^+}(\psi_*)= \lim_{\varepsilon\to 0,L\to\infty} (\overline\psi_*' \psi_*- \overline\psi_* \psi_*')|^L_\varepsilon,\;\psi_*\in D^*_H(\mathbb R_+). \] They extend the Calogero problem to the operators \(\vec H\) with various potentials \(V(x)\) including \(V(x)= g_1 x^{-1}+ g_2 x^{-2}\) and \(\delta\)-like potential, and try to give the exact solution. As Paradox 5, they treat \(ihd_t\psi(t, x)=\vec H\psi(t, x)\), \(\psi(t, x)=\vec U(t)\psi(x)= \exp(-iWt/h)\psi(x)\) from \(\vec U(t)= \exp(-it\vec H/h)\). Next they study the Dirac equation with Coulomb field: \(i\partial_t\psi(x)= \vec H\psi(x)\), \(x= (x_0, r)\), \(r= (x_k,k= 1,2,3)\), \(x_0= t\), \(\vec H=\alpha\vec p+ m\beta- gr^{-1}\), \(\vec p=-i\nabla\), \(\nabla= (\partial_x, \partial_y, \partial_z)\), \(r=|r|\), \(q=-q)1q_2\). That is, \(H= \text{diag}(m-qr^{-1}, -m-gr^{-1})+ \text{antidiag}(\sigma\vec p,\sigma\vec p)\). For an electron in a hydrogen-like atom, we have \(q=Z\widetilde\alpha\), \(\widetilde\alpha= e^2/(hc)= Z^{-1}_c= 1/137\), \(0<Z\leq 118\doteq(3/4)^{1/2} Z_c= Z_s\). Pauli matrices \(\sigma= (\sigma^1, \sigma^2, \sigma^3)\): \[ \sigma^1= \begin{pmatrix} 0 & 1\\ 1 & 0\end{pmatrix},\quad \sigma^2= \begin{pmatrix} 0 & -i\\ i & 0\end{pmatrix},\quad \sigma^3= \begin{pmatrix} 1 & 0\\ 0 &-1\end{pmatrix}. \] They derive the radial equation from \(\widehat H:=\vec H\) with the rotational invariance. That is, let \(\psi(r)= \sum_{j,M,\zeta}\psi_{j,M,\zeta}(r)\); \(\zeta=\pm 1\), \(j= 1/2,3/2,\dots, M=-j\), \(-j+1,\dots, j\), \[ \psi_{j,M,\zeta}(r)= r^{-1}\times^t (\Omega_{j,M,\zeta}(\theta,\phi) f,\;i\Omega_{j,M,-\zeta}(\theta, \phi)g), \] where \[ \begin{multlined}\Omega_{j,M,\zeta}(\theta, \phi)= ((j-M)!/(4\pi(j+ M)!))^{1/2}\exp(i(M-1/2)\phi)\times^t\\ ([M- \zeta\,j- (1+\zeta)/2] P_{j+\zeta/2}^{M-1/2}(\cos\theta), e^{i\phi} P_{j+\zeta/2}^{M+ 1/2}(\cos\theta)).\end{multlined} \] The equation is \(\widehat h_{j,M,\zeta} F(r)=\vec h_{j,\zeta} F(r)\), \(W\): complex constant, \(F(r)=^t(f(r), g(r))\in D(\mathbb R_+)\oplus D(\mathbb R_+)\), where \[ \vec h_{j,\zeta}F(r)= (-i\sigma^2 d_r+\kappa r^{-1}\sigma^1- qr^{-1}+ m\sigma^3) F(r),\;\kappa=\zeta(j+ 1/2). \] They discuss a nonuniqueness of the above s.a. Dirac Hamiltonians. Finally they treat the Dirac operators \(\vec H=\alpha(\vec p-qA)+ m_e\beta\), with Aharonov-Bohm field \(A= A_{AB}\) and Magnetic-Solenoid field \(A= A_{MSF}\). \(A^1_{AB}= -(\Phi/2\pi)\rho^{-1}\sin\phi\), \(A^2_{AB}= (\Phi/2\pi)\rho^{-1}\cos\phi\), \(A^0_{AB}= A_{AB}^3= 0\), \(x_1= x=\rho\cos\phi\), \(x_2= y=\rho\sin\phi\), \(\rho= (x^2+ y^2)^{1/2}\), \(A_{MSF}^1= A^1_{AB}- Bx_2/2\), \(A_{MSF}^2= A^2_{AB}+ Bx_1/2\), \(A^0_{MSF}= A_{MSF}^3= 0\), \(\gamma= e|B|/ch> 0\), \(\varepsilon_B= \text{sgn\,}B\). By ``\(s=\pm 1\), \(M=(m_e^2+ p_z^2)^{1/2}\)'', \[ \psi\rightleftarrows\psi_s(p_z, \rho)\rightleftarrows \chi_s(p_z,\rho)= \sum_{\ell\in\mathbb Z}(2\pi \rho)^{-1/2} S_\ell(\phi) F(s,\ell, p_z; \rho) \] is given. They study the radial equation \([vec h(s,\ell)- W]F(\rho)=0\) of them. That is, \[ \vec h(s,\ell)= i\sigma^2\partial_\rho+ \varepsilon(\gamma\rho/2+ \rho^{-1}\kappa_\ell)\,\sigma^1- sM\sigma^3, \] \(\kappa_\ell= \ell+ \mu-1/2\), \(0\leq \mu< 1\), \(|\kappa_\ell|\leq 1/2\), is solved by using the s.a. Dirac operators.