New mathematical methods in quantum Monte Carlo (QMC) simulation will be applied and developed to obtain more accurate properties of Earth materials than is possible using current electronic structure methods. QMC is one among the most precise known techniques to study realistic materials in physics and chemistry and provide a significant gain in precision compared traditional density functional theory (DFT) approaches. This will bring electronic structure quantum Monte Carlo methodology to a qualitatively higher level of applicability to complex materials and lead to increased accuracy. The new QMC techniques will be applied to forefront problems in the properties of Earth materials in order to obtain accurate equations of state, phase transitions, and elasticity of solid materials that are of high interest in geophysics. One significant limitation of today's QMC methods is the high computational demand, which currently makes applications to larger systems including solid solutions prohibitively expensive. A substantial portion of the QMC computation is spent on forming and evaluating a Slater determinant, which is constructed from one-particle orbitals. The team plans to develop and apply different localization transformations in order to obtain a sparse determinant. Two linear algebra methods will be developed for their efficient evaluation. First, Krylov's method for the iterative evaluation of a determinant will be incorporated into QMC. Secondly, the trace of the determinant will be calculated with Monte Carlo methods. Both techniques will further the goal of obtaining an order-N QMC techniques that are more efficient and applicable to a wider range of materials, well beyond the current possibilities.
Computational mineral physics is part of the large effort to use computer simulations to predict and understand properties of the Earth. New mathematical techniques will be derived to make advance quantum Monte Carlo simulations more precise and significantly more efficient. The properties of important Earth materials at high pressure will be predicted with unprecedented accuracy. This work will also lead to a more precise description of a number of fundamental phase transitions in solid deep Earth. This project is a close collaboration between a mathematician, mathematical physicists, and geophysicists. It will bring new applied math methods into geoscience, with broad applicability to all Earth materials. The broader impact will include the development of new computational techniques applicable to many areas as well as the education of graduate students and post-docs in state-of-the-art materials simulations, teaching of new computational techniques in graduate level classes and during two workshops that will be organized.
