Hi all,
Abstract: In this thesis, we developed different methods for the training and the optimization of a Density Functional Theory (DFT) exchange and correlation functionals, vxc, which relies on the minimization of the electronic density difference obtained by subtracting the approximated DFT density with a reference one evaluated on a quantum computer. The determination of accurate and universal DFT vxc functional represents an unsolved problem in computational chemistry and despite many attempts (based on theoretical as well as heuristic approaches) a general solution that is applicable in all physical contexts is still lacking. The new generation of DFT functionals are build combining different energy contributions and energy corrections (such as exchange, correlation but also van der Waals energy terms with different short and long range behaviors) weighted by a number of parameters that are usually fitted to minimize the errors on a set of properties (mainly relative energies) in a training set of molecules and solid state systems.

In this work, motivated by the central role of the density in the Hohenberg-Kohn theorem, we investigate the possibility to use directly the electronic density of molecular systems to optimize the DFT functionals, hoping for a greater universality of the new functional resulting from this optimization.