2006 Workshop on Challenges and Opportunities at the Interfaces of Scientific Computing and Computational Geodynamics
The emergence of terascale computing--and the arrival soon of the petascale era--provide tremendous opportunities for furthering our understanding of geodynamic systems and phenomena through large-scale computer modeling and simulation. However, geodynamics problems involve a number of complex features that present difficulties for scientific computing methods and tools, particularly at large scale. These include significant degrees of nonlinearity, heterogeneity, anisotropy, geometric complexity, multiphysics coupling, and multiscale/multirate behavior.
Accordingly, CIG will sponsor a two-day workshop on October 16-17, 2006 at the University of Texas at Austin that brings together computational geodynamicists and scientific computing experts to identify and assess challenges and opportunities at the interfaces of frontier computational geodynamics problems and scalable numerical and geometric algorithms and software.
The goals of the workshop are
- to identify the scientific computing issues and obstacles encountered in computational geodynamics simulations as they continue to scale up in size and complexity;
- to assess the prospects of state-of-the-art scientific computing algorithms and software in addressing the complexities of computational geodynamics problems; and
- to define directions for scientific computing research that meet the challenges presented by geodynamics problems.
The workshop will include talks by application scientists discussing contemporary modeling techniques, while highlighting open computational science problems at the frontiers of the areas of short-term crustal dynamics, long-term deformation, geodynamo, global seismology, mantle convection, and magma migration. Complementing the geodynamics presentations, scientific computing researchers will present talks that address challenges in geology-aware large-scale mesh generation and adaptation, linear solvers and preconditioners for ill-conditioned problems, nonlinear solvers and operator splitting techniques for complex coupled problems, data formats and management, and parallel scientific visualization. Issues in software frameworks and quality assurance of open-source libraries supporting complex large-scale computations will also be discussed.
Immediately following the workshop, on October 18, the CIG Science Steering Committee (SSC) will host a computational science roundtable to focus the workshop discussions on CIG's software development roadmap. Members of the SSC and others involved with CIG will describe how they think CIG should move forward, followed by open discussion. The SSC would then use the results of this roundtable discussion in the development of next year's CIG Strategic Plan. All are welcome to participate.
Computational geodynamicists and scientific computing researchers are invited to attend and participate in the workshop discussions, and present a poster at the poster session.