Electronic: Ab initio simulation of strongly correlated materials: methods and applications
During last several decades, the ab-initio approach to simulations of materials has proven to be very fruitful. This approach is based on the density functional theory (DFT) and has allowed for theoretical modeling of a wide range of properties of solid-state materials without use of any adjustable parameters. However, the DFT-base methods for electronic structure simulations have inherent limitations due to their quasi-one-electron treatment of the electron-electron interaction. These limitations become severe in the case of so-called strongly correlated materials (transition-metal, rare-earth and actinide-base compounds), where a strong inter-atomic Coulomb repulsion between localized electrons leads to complicated many-body phenomena like Mott metal-insulator transitions, heavy-fermion behavior and high-temperature superconductivity.
During recent years, the progress in theoretical understanding of correlation effects has led to formulation of an approach that combines the ab initio DFT-base methods for the electronic structure simulation with a correct treatment of local many-body effects by the so-called dynamical mean field theory (DMFT). This advanced approach, however, requires a complete solution of the many-body problem, at least for a single atomic site. Numerically exact approaches to this problem are based on quantum Monte-Carlo algorithms and remain quite computationally expensive. Up to date, the majority of such studies have been still restricted to simple systems due to the associated computational cost.
Within our project, we apply the LDA+DMFT method in large-scale computations, involving real transition metal systems based on Fe, Ni, Co and Mn, as well as their alloys and compounds. These simulations are aimed at predicting magnetic and structural properties of such systems at elevated temperature and pressure, which is important for a wide range of technological applications as well as for the fundamental solid-state physics and geophysical science. Transition metals are the main ingredients in many novel functional materials, including high-performance steels and materials for spintronics applications.
For example, within this theoretical framework we study the phase stability of iron at temperature and pressure conditions ranging from ambient to that expected for the inner Earth core. We investigate both magnetic and spectroscopic properties. Such applications of the LDA+DMFT method naturally require a continuous method development involving both the physics and computational algorithms that are designed to take a full advantage of the currently available facilities. In particular, an excellent scalability of our quantum Monte-Carlo (QMC) method in parallel computations makes it a good candidate for petaflop simulations. The implementation work during last several years have allowed to combine this advanced QMC many-body technique with a highly accurate first-principle electronic structure approach (the linear adjumented plain-wave method, LAPW). The current development effort is directed towards further significant increase in the method performance in order to make ab intio predictions of elastic properties of real materials accessible within this technique.
I. A. Abrikosov, L. V. Pourovskii, M. Ekholm