Many-body-based DFT treatment of fermions in optical lattices

Abstract: With recent advances in the field of ultra cold atoms one can, by trapping atoms at low temperatures by laser beams, simulate systems which can be adequately described by single-band lattice Hamiltonians. Also, due to the high parameter tunability of the experimental setups, lattice disorder can be introduced in a con- trolled fashion in these systems. This thesis considers disordered/ordered interacting fermion lattice systems in equilibrium in one and two dimensions, subject to trap- ping parabolic potentials, where different levels of description, ranging from exact where possible (in 1D, Density Matrix Renormalization Group, DMRG), to approxi- mate (based on several local-density approximations within the framework of Lattice Density Functional Theory) are used. The exchange-correlation potentials consid- ered come from Many-Body Approximations obtained using Green’s functions, as well as from an exact local-density approximation (LDA) based on the Bethe-Ansatz (BALDA) in 1D. Both one- and two-dimensional systems were studied in equilib- rium, essentially looking at the ground state density profiles. Furthermore, in 2D, a pseudo-dynamics representing the trap-opening in the complete adiabatic limit was also studied. For one-dimensional systems, it is found that, in general, BALDA yields very good results compared to DMRG, except for the low density limit, but DMRG can describe features that none of the LDA:s considered can reproduce. It is also found that the strength of the external potential affects the impact that the exchange-correlation potential has on the system. Many of these features translate to the 2D case. However, a new aspect emerges in two dimensions, related to the competition of disorder and interaction. Here an important outcome is that, on opening the trap at an ideally adiabatic rate, different MBA:s (and thus different LDA:s) provide different minimal vs maximal expansion radii of the particle cloud, as a result of the interplay of disorder and interaction, and the underlying square lattice structure. For the 2D results, exact benchmarks were not available, and our findings may thus need further validation, by e.g. considering several disorder configurations or, ideally, by performing full Green’s function calculations. These considerations are summarised in our conclusions and outlook remarks, where possible directions for future investigation are highlighted.