NEWDFTFUNCT · Development of new non-empirical DFT functionals

Nowadays the application of computational chemistry extends to the realm of bioinorganic, organic and nanotechnology molecules, whose size impedes the use of purely ab initio methods, leaving density functional theory (DFT) as the only recourse to model chemical reactions. The accuracy of DFT calculations depends critically on the quality of functionals used and, as a result, in the last years there has been a craving for new density functionals. Many functionals have been designed, often based on the parameterization of existing ones. While this research brings accurate functionals for very particular purposes, it does not pave the way for long-lasting all-purpose ones. A universal functional should be constructed from physical constraints or model systems, which cover a wide range of molecular situations.Harmonium, a model system alike to an ordinary atom, permits an easy tuning of electron correlation effects, thus being a formidable test bed for DFT methods. The researcher has designed an algorithm to calculate highly accurate energies and wavefunctions of few-electron harmonium systems, which are used to calibrate a large number of DFT functionals. From the benchmark calculations obtained from few-electron harmonium he will design a new version of Müller's functional driven by the amount of electron-correlation introduced in the system. He will also construct a non-empirical DFT functional by imposing appropriate physical constraints. Finally, in order to improve the performance of the functional a hybridized form of latter will be constructed. The designed DFT functionals will not be just an evolution of the current ones, but a new type. Thus, they could solve some of the critical problems of the current functionals.There are very few groups in Spain working in the development of new functionals. This project will pursue this goal employing a very genuine strategy, and holds the promise to provide relevant results for the development of DFT methods.