MELIORA: A Finite-Element Mixed Stochastic-Deterministic Density Functional Theory Code for Accurate First-Principles Simulations of Warm Dense Matter

Density functional theory (DFT) has recently been the most successful approach to first-principles microscopic simulations of not only material properties under ambient conditions, but also matter under extreme conditions, such as those encountered in the warm dense matter regime (WDM). However, DFT simulations of the WDM regime are computationally challenging due to three main reasons: 1) explicit inclusion of all electrons is required at high densities and temperatures where pseudopotentials do not provide the required accuracy, 2) the low density regime requires large simulation cells, which means large number of grid points required to accurately sample the electronic density and 3) at high temperatures, a large number of partially occupied electronic states above the Fermi-Dirac (FD) level needs to be take into account. Here, we present MELIORA, a finite-element (FE) all-electron DFT-based molecular dynamics code designed for efficient WDM simulations. MELIORA uses adaptive mesh refinement to optimize the real space sampling of the electron density and employs mixed stochastic-deterministic DFT in order to avoid explicit treatment of the many partially FD occupied bands encountered at high temperatures. Furthermore, the FE basis provides a block-diagonal matrix representation of the Kohn-Sham Hamiltonian, thus opening the door to recently developed linear-scaling diagonalization techniques. Preliminary results from MD simulations of warm dense H and C will be presented.