Location: The workshop (Oct. 16-17) will take place in the Avaya auditorium of the ACES building (ACE 2.302) on the University of Texas at Austin main campus, which houses the Institute for Computational Engineering and Sciences (ICES). The auditorium is off the main hall on the ground floor. The roundtable (Oct. 18) will take place in the Barrow conference room of the Geological Sciences building (GEO 4.102).
Monday October 16: Computational Challenges in Geodynamics
Computational geodynamicists will present state-of-the-art computational modeling techniques in the areas of seismology, short-term tectonics, long-term tectonics, mantle convection, magma migration, and the geodynamo. They will highlight open computational problems and opportunities for overcoming them.
- 8:00am
- Bus leaves for ACES building from Capitol Place Hotel
- 8:00am - 8:30am
- Continental breakfast
Workshop Overview
- 8:30am - 9:00am
- Omar Ghattas, University of Texas at Austin, Introduction and Logistics
Seismology
- 9:00am - 9:25am
- Jeroen Tromp, Caltech, "Spectral-Element and Adjoint Methods in Seismology"
- 9:25am - 9:50am
- Heiner Igel, Ludwig-Maximilians-Universität (LMU) Munich, "Wave propagation in highly complex models: grid generation, local time stepping, parallelization"
- 9:50am - 10:00am
- Discussion
Break
- 10:00am - 10:30am
Short-Term Tectonics
- 10:30am - 10:55am
- Mark Simons, Caltech, "Going beyond an elastic halfspace"
- 10:55am - 11:20am
- Brad Aagaard, USGS Menlo Park, "Using multi-cycle earthquake simulations to understand crustal dynamics"
- 11:20am - 11:30am
- Discussion
Long-Term Tectonics
- 11:30am - 11:55am
- Luc Lavier, University of Texas at Austin, "Problematic and present limitations in modeling long term lithospheric deformation"
- 11:55am - 12:20pm
- Mousumi Roy, University of New Mexico, "Scientific and computational challenges in simulating fluid-rock interaction and its role in the long-term tectonics of continental lithosphere"
- 12:20am - 12:30pm
- Discussion
Lunch
- 12:30pm - 2:00pm
Mantle Convection
- 2:00pm - 2:25pm
- Shijie Zhong, University of Colorado at Boulder, "Numerical Methods in Finite Element Mantle Convection Code CitcomS"
- 2:25pm - 2:50pm
- Scott King, Purdue University, "Grid refinement and parallel computing for mantle convection and thermochemical convection"
Cancelled: Peter van Keken, University of Michigan, "Thermochemical convection and the use of strong grid refinement: Efficiency, accuracy, solvers, parallelization"
- 2:50pm - 3:00pm
- Discussion
Magma Migration
- 3:00pm - 3:25pm
- Marc Spiegelman, Columbia University, "Computational challenges in Magma Dynamics: Coupling in multi-physics problems"
Break
- 3:45pm - 4:00pm
Geodynamo
- 4:00pm - 4:25pm
- Moritz Heimpel, University of Alberta, "Modeling the Geodynamo: Successes and Challenges"
- 4:25pm - 4:35pm
- Discussion
General discussion on needs and opportunities in computational geodynamics
- 4:35pm - 5:15pm
- Wolfgang Bangerth, Texas A&M University, moderator
Poster Session and Reception (ACES Faculty Lounge, ACE 6.102)
- 5:15pm - 6:45pm
- 6:45pm
- Bus leaves for Capitol Place Hotel from ACES building
Tuesday October 17: Scientific Computing Algorithms and Tools
Scientific computing researchers will address capabilities and barriers associated with scalable large-scale algorithms and tools for geology-aware mesh generation, mesh adaptivity, linear solvers and preconditioners for ill-conditioned problems, nonlinear solvers and time integrators for complex coupled problems, data management, and scientific visualization.
- 8:00am
- Bus leaves for ACES building from Capitol Place Hotel
- 8:00am - 8:30am
- Continental breakfast
PDE Solver Libraries
- 8:30am - 8:55am
- Barry Smith, Argonne National Laboratory, "An Architecture for Composing Solvers"
- 8:55am - 9:20am
- Roscoe Bartlett, Sandia National Laboratories, "Trilinos Overview"
- 9:20am - 9:30am
- Discussion
Data Management and Scientific Visualization
- 9:30am - 9:55am
- Claudio Silva, University of Utah, "Scalable Techniques for Scientific Visualization"
- 9:55am - 10:20am
- Tiankai Tu, Carnegie Mellon University, "Besides Teraflops: Getting More from Big Iron"
- 10:20am - 10:30am
- Discussion
Break
- 10:30am - 11:00am
Linear Solvers
- 11:00am - 11:25am
- Xiaoye Sherry Li, Lawrence Berkeley National Laboratory, "Improving Scalability of Sparse Direct Linear Solvers"
- 11:25am - 11:50am
- Rob Falgout, Lawrence Livermore National Laboratory, "hypre: High Performance Preconditioners"
- 11:50am - 12:00pm
- Discussion
Lunch
- 12:00pm - 1:30pm
Nonlinear Solvers
- 1:30pm - 1:55pm
- John Shadid, Sandia National Laboratories, "An Overview of the Application of Newton-Krylov based Solution Methods for Nonlinear Multiple-time-scale PDE Systems"
- 1:55pm - 2:05pm
- Discussion
Finite-Element Mesh Generation
- 2:05pm - 2:30pm
- Carl Gable, Los Alamos National Laboratory, "Mesh Generation for Geological Applications: Why it is Different than Meshing Planes, Trains, and Automobiles"
- 2:30pm - 2:55pm
- Steve Owen, Sandia National Laboratories, "Mesh Generation and Geometry for Geodynamics Simulations using the CUBIT Toolkit"
- 2:55pm - 3:05pm
- Discussion
Break
- 3:05pm - 3:30pm
Adaptivity
- 3:30pm - 3:55pm
- Wolfgang Bangerth, Texas A&M University, "Techniques for adaptive mesh refinement"
- 3:55pm - 4:20pm
- Abani Patra, State University of New York at Buffalo, "Modeling geophysical mass flows -- mesh(free), model and parameter adaptivity"
- 4:20pm - 4:30pm
- Discussion
Discussion and Wrapup
- 4:30pm - 5:15pm
- 5:30pm
- Bus leaves for hotel from ACES building
Wednesday October 18: CIG Computational Science Roundtable
Location: Barrow conference room, Geological Sciences building (GEO 4.102)
The purpose of the Computational Science Roundtable is to take stock of the workshop discussions and seek community input on the implications for CIG's strategic plan. Members of the SSC and others involved with CIG will lead a discussion on how CIG should move forward.
- 8:00am
- Bus leaves for meeting site from Capitol Place Hotel
- 8:00am - 8:30am
- Continental breakfast
Agenda
- Geodynamics community computational hardware needs, Peter Olson (John Hopkins University), discussion leader**
- How do we help users get CIG software running on systems appropriate for their problems?
- How do we best engage NSF supercomputing centers?
- For developing and deploying CIG software on their systems?
- For facilitating groundbreaking research in geodynamics?
- What types of problems will be solved with terascale computing?
- What types of problems could we solve with petascale systems?
- What are the needs for dedicated geodynamics computational facilities?
- Presentation of the CIG Five-Year Strategic Plan, Peter Olson (Johns Hopkins University), discussion leader**
- Summary of workshop discussions and identification of cross-cutting challenges and solutions, Brad Aagaard (USGS), discussion leader**
- What are the common computational challenges that multiple CIG working groups are encountering?
- What common tools and libraries are available that could be integrated into CIG software to address these challenges?
- Can CIG software be developed/redeveloped using some common components?
- What are the priorities/timelines for integrating these tools into CIG software?
- How should the CIG Five-Year Strategic Plan be updated to reflect the community discussions of the last several days? Omar Ghattas (UT Austin), discussion leader**
Working Lunch
- 12:00pm - 1:00 pm
Wrapup
- 1:00pm - 2:00pm
Adjourn
- 2:00pm
