Time-dependent variational approach for boson systems (Q1290462)

The time-dependent variational principle of \textit{R. Balian} and \textit{M. Vénéroni} [Phys. Rev. Lett. 47, 1353, 1765(E) (1981), Ann. Phys. (N.Y.) 164, 334 (1985)] is applied to many-boson systems. The purpose of the paper is to use the Balian-Vénéroni variational principle in order to build several approximation schemes for dynamical boson systems when one is interested in prediction of the expectation values as well as the correlations and fluctuations of one-body operators. Bose-Einstein condensation of dilute atomic gases in laser traps at ultralow temperatures and dynamical behavior of the condensate under the time-dependent perturbations constitute the physical background of the investigation. By suitably restricting the trial spaces for the two variational operators \(\mathcal{D}(t)\) and \(\mathcal{A}(t)\) the different approximation schemes can be obtained. The operators \(\mathcal{D}(t)\) and \(\mathcal{A}(t)\) are related correspondingly to a density operator and a measured observable. If \(\mathcal{D}(t)\) belongs to the class of exponential operators of quadratic polynomials (EQP) in the boson creation and annihilation operators and \(\mathcal{A}(t)\) belongs to the class of one-body operators, then the variational derivation of the time- dependent Hartree-Fock-Bogoliubov equations for \(\mathcal{D}(t)\) and the `dual-RPA' equations for \(\mathcal{A}(t)\) can be obtained. In order to go beyond this mean-field-like approximation, and to be able to compute in a consistent way correlations and fluctuations of single particle observables, both \(\mathcal{D}(t)\) and \(\mathcal{A}(t)\) are considered to be EQP operators. Then the resulting equations become coupled and, due to the boundary conditions inherent to the variational principle, lead to a non-standard two-time (or mixed) boundary value problem. If \(\mathcal{D}(t)\) and \(\mathcal{A}(t)\) degenerate into diadics or projectors on pure states, the coupled equations allow the determination of transition probabilities.