AN AUTOMATED PARALLEL COMPUTING SYSTEM IN THE GEON GRID: APPLICATIONS TO MULTISCALE CRUSTAL DEFORMATION IN THE WESTERN UNITED STATES

In the past decade supercomputing power has become available to most researchers in the form of affordable Beowulf clusters and other parallel computer platforms. However, to take full advantage of such computing power requires developing parallel algorithms and related codes, a task that is often too daunting for geoscience modelers whose main interest is in geosciences. As part of the GEON effort, we have been developing an automated parallel computing system built on open-source algorithms and libraries. Users interact with this system by specifying the partial differential equations, solvers, and model-specific properties using a high-level modeling language in the input files. The system then automatically generates the finite element codes that can be run on distributed or shared memory parallel machines. This system is dynamic and flexible, allowing users to address a large spectrum of problems in geosciences. We demonstrate this modeling system with a suite of geodynamic models that simulate multiscale crustal deformation in the western United States, ranging from timescale-dependent faulting in the San Andreas Fault system to strain localization and active deformation in the western US cordillera. We show that this system may facilitate integration of high-performance computing with distributed data grids in the emerging geoscience cyber-infrastructures.