Colloquium: Dr. Anouar Benali, Argonne National Labor.

 Towards predictive simulations of molecules and solids using quantum Monte Carlo methods.

ABSTRACT:

In the past decades, quantum Monte Carlo methods, and particularly, diffusion Monte Carlo (DMC) have proven quite accurate at reproducing, and in some cases, predicting properties of a wide class of solids and molecules. Solving stochastically the many-body Schrodinger equation,  allows to explicitly take into account from strong correlations to van der Waals interactions, in both solids and molecules.  Thanks to the significant advances in high performance computing (specially at the dawn of exascale computing), the accuracy of the method depends on a limited number of approximations trivially controlled, and on the fixed-node approximation, also controllable but at a higher cost. In this seminar, we will present the current advances,  development and limitations of DMC as implemented in the open source code QMCPACK with examples of applications on material science (bulks, defects and layered materials) and on quantum chemistry (MOF, barrier reactivity, formation energies).