Adiabatic transition probability for a tangential crossing (Q2642343)

For time-depending Schrödinger equation \[ ih\frac {d}{d\,t}\psi(t)={\mathcal H}(t,\,\varepsilon)\psi(t),\tag{*} \] where \[ {\mathcal H}(t,\,\varepsilon)= \left( \begin{matrix} V(t)& \varepsilon\\ \varepsilon& -V(t) \end{matrix} \right) \] and \(\varepsilon,\; h\) are small positive parameters and \(V(t)\) is a real-valued function for real \(t\), the problem of the calculation of the scattering matrix and the transition probability is solved. The following assumptions on the function \(V(t)\) are imposed. (A) V(t) is real-valued on \(\mathbb R\) and there exist two real numbers \(0<\theta_0<\pi/2\) and \(\rho>0\) such that \(V(t)\) is analytic in the complex domain \[ {\mathcal S}=\{ t\in\mathbb C;|\text{Im}\,t|<|\text{Re}\,t|\tan\theta_0\}\cup \{|\text{Im}\,t<\rho\}. \] (B) There exist two real non-zero constants \(E_r,\) \(E_l\) and \(\sigma>0\) such that \[ V(t)=\begin{cases} E_r+O(|t|^{-\sigma}) & \text{as } \text{Re}\, t\to +\infty \text{ in }\mathcal S,\\ E_l+O(|t|^{-\sigma}) &\text{as } \text{Re}\, t\to -\infty \text{ in }\mathcal S. \end{cases} \] (C) \(V(t)=0\) if and only if \(t=0.\) Under the conditions (A) and (B) there exists four Jost solutions \(\psi^r_+, \psi^r_-,\psi^l_+\) and \(\psi^l_- \) to (*) uniquely defined by the following asymptotic conditions: \[ \begin{alignedat}{2} \psi^r_+(t)&\sim \exp\left[+\frac{i}{h}\sqrt{E_r^2+\varepsilon^2 t}\right] \binom{-\sin\theta_r}{\cos\theta_r} &&\quad \text{as } \text{Re}\,t\to +\infty\text{ in } {\mathcal S}, \\ \psi^r_-(t)&\sim \exp\left[-\frac{i}{h}\sqrt{E_r^2+\varepsilon^2 t}\right] \binom{\cos\theta_r}{\sin\theta_r } &&\quad \text{as } \text{Re}\to +\infty \text{ in } {\mathcal S}, \\ \psi^l_+&\sim \exp\left[+\frac{i}{h}\sqrt{E_l^2+\varepsilon^2 t}\right] \binom{-\sin\theta_l}{\cos\theta_l} &&\quad \text{as }\text{Re}\,t\to -\infty \text{ in } {\mathcal S},\\ \psi^l_-&\sim \exp\left[-\frac{i}{h}\sqrt{E_l^2+\varepsilon^2 t}\right] \binom{\cos\theta_l}{\sin\theta_l} &&\quad \text{as } \text{Re}\,t\to -\infty \text{ in } {\mathcal S}, \end{alignedat} \] where \(\tan 2\theta_r = \varepsilon/E_r\) and \(\tan 2\theta_l = \varepsilon/E_l\) \((0<\theta_r\), \(\theta_l<\pi/2).\) The pairs of Jost solutions \((\psi_+^r(t),\,\psi_-^r(t))\) and \((\psi_+^l(t),\,\psi_-^l(t))\) are orthonormal bases on \(\mathbb C^2\) for any fixed \(t.\) The scattering matrix \(S\) is defined as the change bases of Jost solutions \[ (\psi^l_+,\,\psi_-^l) =(\psi_+^r,\,\psi_-^r)S(\varepsilon,\, h), \] where \[ S(\varepsilon,\, h)=\left(\begin{matrix} s_{11}(\varepsilon,\, h)& s_{12}(\varepsilon,\, h)\\ s_{21}(\varepsilon,\, h) & s_{22}(\varepsilon,\, h) \end{matrix}\right) \] is an unitary matrix independent of \(t.\) The transition probability \(P(\varepsilon,\, h)\) is defined by \[ P(\varepsilon,\, h) = |s_{21}(\varepsilon,\, h)|^2. \] Let \(n\in \mathbb N\) be a positive integer such that \(V^{(k)}(0)=0\) for \(0\leq k< n.\) Then there exist \(2n\) simple turning points \(T_j(\varepsilon),\;\overline{T}_j(\varepsilon)\), that is \(V(T_j(\varepsilon))^2+\varepsilon^2=0\) and \(V(\overline{T}_j(\varepsilon))^2+\varepsilon^2=0.\) By definition \[ A_j(\varepsilon)=2\int_{0}^{T_j(\varepsilon)}\sqrt{V(t)^2+\varepsilon^2}. \] The main result of the paper is the proof of the following asymptotic formulae for \(P(\varepsilon,\, h)\) when \(\varepsilon\) and \(h\) are both small. Theorem. For \(V(t)\) assume (A), (B), (C). Then for \(n=1\) there exist \(\varepsilon_0 > 0\) such that we have \[ P(\varepsilon,\, h)= \text{exp}\, \left[ -\frac{2\text{Im}A_1(\varepsilon)}{h} \right] (1+ O(h))\quad \text{as } h\to 0 \] uniformly for \(\varepsilon \in (0,\, \varepsilon_0 ).\) For \(n\geq 2\) there exist \(\varepsilon_0 > 0\) such that, for all \(\varepsilon \in (0,\, \varepsilon_0 ),\) we have \[ P(\varepsilon,\, h)= \left|\exp \left[ \frac{i}{h}A_1(\varepsilon)\right]+ (-1)^{n+1}\exp \left[ \frac{i}{h}A_n(\varepsilon)\right]\right| \left(1+ O\left(\frac{h}{\varepsilon^{n+1/n}}\right)\right) \] as \(\frac{h}{\varepsilon^{n+1/n}}\to 0.\) The basic tool for the solution of these problems is the exact WKB method for \(2\times 2\) systems introduced by Fujiié, Lasser and Nedelec.