Semi-classical description of matter wave interferometers and hybrid quantum systems

This work considers the semi-classical description of two applications involving cold atoms. This is, on one hand, the behavior of a BOSE-EINSTEIN condensate in hybrid systems, i.e. in contact with a microscopic object (carbon nanotubes, fullerenes, etc.). On the other, the evolution of phase space distributions in matter wave interferometers utilizing ray tracing methods was discussed. For describing condensates in hybrid systems, one can map the GROSS-PITAEVSKII equation, a differential equation in the complex-valued macroscopic wave function, onto a system of two differential equations in density and phase. Neglecting quantum dispersion, one obtains a semiclassical description which is easily modified to incorporate interactions between condensate and microscopical object. In our model, these interactions comprise attractive forces (CASIMIR-POLDER forces) and loss of condensed atoms due to inelastic collisions at the surface of the object. Our model exhibited the excitation of sound waves that are triggered by the object’s rapid immersion, and spread across the condensate thereafter. Moreover, local particle loss leads to a shrinking of the bulk condensate. We showed that the total number of condensed particles is decreasing potentially in the beginning (large condensate, strong mean field interaction), while it decays exponentially in the long-time limit (small condensate, mean field inetraction negligible). For representing the physics of matter wave interferometers in phase space, we utilized the WIGNER function. In semi-classical approximation, which again consists in ignoring the quantum dispersion, this representation is subject to the same equation of motion as classical phase space distributions, i.e. the LIOUVILLE equation. This implies that time evolution of theWIGNER function follows a phase space flow that consists of classical trajectories (classical transport). This means, for calculating a time-evolved distribution, one has know the initial distribution and one has to solve the classical equations of motion. Concerning the initial distribution, we have studied a stationary solution of the nonlinear LIOUVILLE equation, the LAMBERT density. We saw that it agrees very well with results from singleparticle quantum mechanics as well as the MAXWELL-BOLTZMANN distribution in the weakly interacting limit. Likewise, in the strongly interacting limit, familiar results of the THOMAS-FERMI approximation are recovered. A distribution that is first prepared in a trap and then released can be described quite conveniently in terms of WIGNER functions. However, propagation in optical potentials associated to the interferometer elements (beam splitter, Pi-half-pulse) do not satisfy the condition of the semiclassical approximation. Nevertheless, one finds discrete before-after mappings that describe the effect of these elements on incident distributions. This leads to several channels of phase space propagation which relate to the interferometer paths and interferences between them. The formalism for WIGNER functions in an interferometer can be translated straightforwardly into a ray tracing algorithm. As mentioned above, this algorithm solves the classical equations of motion and computes time-evolved distributions, using values of the initial distribution. This procedure, in contrast to most analytical solutions, does not require the HAMILTON function to be quadratic (e.g. free propagation, const. acceleration, harmonic oscillator). We compared simulation results to analytic expressions in case of freely propagating GAUSS distribution. They showed perfect agreement, especially for the functional dependence of wave length and contrast of interference fringes on the laser pulse timings