The many-electron quantum theory is the base for a first-principles description and understanding of properties of atoms, molecules, clusters, and materials. Breakthrough in electronic-structure methods and algorithms can have far-reaching influence on chemistry, physics, and materials science where the first-principles computation is becoming an indispensable tool.
Kohn-Sham density-functional theory within its local-density approximation (LDA) and generalized gradient approximations (GGAs) is presently the "standard model" for materials science. These approximations are usefully accurate for a large variety of applications ranging from molecules to solids, surfaces, and clusters in physics, chemistry, biology and the nano-sciences. However, they fail in a number of well-documented situations, among which the most prominent are van der Waals (vdW) bonded systems, materials with strong correlations, certain surface adsorption problems, and chemical reaction barrier heights to name a few. Some of these difficulties can be overcome or at least alleviated by employing the so-called hybrid functionals, which mix a fraction of Hartree-Fock exchange with semi-local exchange. However, in some situations, even hybrid functionals are not sufficiently accurate, or have the necessary ingredients to capture the right physics.
The objectives of the Max Planck Partner Group is to develop computational approaches and softwares to overcome the intrinsic deficiencies of conventional approximations in density-functional theory (DFT). The driving force for this development was the increasing need for higher accuracy and reliability of electronic-structure simulations in chemistry, physics, and materials science. The recently proposed Renormalized Second-order Perturbation Theory (rPT2) (Phys. Rev. B 88, 035120 (2013)) is an outcome of our previous investigations along this line.
Based on what we have achieved so far, our future objective is to develop and implement advanced electronic-structure methods that are both sufficiently accurate and numerically tractable for materials science. From the methodology point of view, what we would like to have is a method that is generally better than rPT2, reaching an accuracy level comparable to CCSD(T), but with lower-scaling and thus lower computational cost. From the application point of view, we would like to extend the reach of rPT2 and its further development to condensed-matter physics and materials science. This requires an efficient implementation of rPT2 and beyond for periodic systems, which is the first goal to be achieved in this project. Starting from this, our long-term goal is to develop advanced and generally applicable computational quantum mechanical methods and the corresponding software package that can be used in chemistry, condensed-matter physics, materials science, and biophysics.