Acquisition of a Supercomputer for Theoretical Geophysics

点击翻译按钮获取中文摘要

9975753CohenThe acquisition of a 16 processor SGI/Cray SV1 supercomputer with peak performance of 19.2 GFLOPS will allow us to do important theoretical geophysics computations that would otherwise not be tractable. We also will upgrade and increase the capacity of our RAID 5 network file server to maintain our input and output data sets and codes. We will use the facility to study: (1) the behavior of transition metal ions and iron bearing minerals at high pressures; (2) diffusion and rheology of perovskite and magnesiowustite; (3) high-pressure high-temperature elasticity and anelasticity of perovskite, magnesiowustite, iron, and other important Earth phases; and (4) convective and subduction transport in the mantle and effects of realistic rheologies on mantle convection. Major outside users currently include Lars Stixrude and his group (Univ. Michigan), and S. King (Purdue Univ.). This facility will be unparalleled in ability to provide significant blocks of access, especially for large memory jobs. This facility is crucially needed for our research, which would be crippled without it. The upgrade would also enhance and sustain the investment NSF has already made in our Cray system. Research problems that will become possible are: (1) Diffusion, rheology, and finite temperature elasticity and anelasticity of orthorhombic silicate perovskite (with NSF sponsored post-doc Frederic Marton); (2) Effects of ferrous and ferric iron on properties of magnesiowustite and perovskite (with NSF sponsored post-doc Stephen Gramsch); (3) High quality molecular dynamics simulations of viscosity and melting curve of Fe (with major outside user Prof. L. Stixrude); (4) Convection modeling with realistic rheologies obtain via mineral physics (with major outside user: Prof. S. King, Purdue University).***