HR: 0800h
AN: IN21C-1187    [Abstracts]
TI: From StGermain to Underworld: Enabling Community-based code Development in Geodynamics
AU: * Quenette, S M
EM: steve@vpac.org
AF: Victorian Partnership for Advanced Computing (VPAC), 110 Victoria Street, PO Boc 210, Carlton South, Melbourne, Vic 3053 Australia
AU: Moresi, L
EM: louis.moresi@sci.monash.edu.au
AF: Monash University, School of Mathematical Sciences, Monash University, Melbourne, VIC 3800 Australia
AU: Sunter, P D
EM: pds@vpac.org
AF: Victorian Partnership for Advanced Computing (VPAC), 110 Victoria Street, PO Boc 210, Carlton South, Melbourne, Vic 3053 Australia
AU: Hodkinson, L
EM: luke@vpac.org
AF: Victorian Partnership for Advanced Computing (VPAC), 110 Victoria Street, PO Boc 210, Carlton South, Melbourne, Vic 3053 Australia
AU: Lo, A
EM: alan@vpac.org
AF: Victorian Partnership for Advanced Computing (VPAC), 110 Victoria Street, PO Boc 210, Carlton South, Melbourne, Vic 3053 Australia
AU: Hassan, R
EM: raq@vpac.org
AF: Victorian Partnership for Advanced Computing (VPAC), 110 Victoria Street, PO Boc 210, Carlton South, Melbourne, Vic 3053 Australia
AU: Appelbe, B
EM: bill@vpac.org
AF: Victorian Partnership for Advanced Computing (VPAC), 110 Victoria Street, PO Boc 210, Carlton South, Melbourne, Vic 3053 Australia
AU: Turnbull, R
EM: robert.turnbull@sci.monash.edu.au
AF: Monash University, School of Mathematical Sciences, Monash University, Melbourne, VIC 3800 Australia
AB: Each discipline of geophysics has traditionally focused on limited sets of closely related phenomena using methodologies and data sets optimized for its specific area of interest. Why is that? Single discipline, single scale, foundation physics problems are relatively easy to code in Fortran, and hence they eventually become optimized for best performance whilst simultaneously becoming difficult to adapt to new interests. Yet geodynamicists want to break these ``out-of-scope'' barriers, and incorporate signals of interests beyond their immediate phenomena of interest. In turn this often entails a multi physics, multi scale and multi discipline development model. Multi physics is potentially easy to code, but application limited by the choice of numerical technique of the code. Multi scale is a numerical and discretisation issue that is closely related to the fundamental data structures of the code. This is difficult to change, and the ideal is hybrids of optimized solutions at desired scales. Multi discipline is much more focused on people and how they form problem constraints, the language / ontology they use, and their expectation in usability. In summary: facilitating a multi scale, multi physics , multi disciplinary development environment is difficult, complicated and generally not of core interest to a geodynamicist. However, today, with more powerful CPU architectures, we can move away from Fortran style coding with little wall-time cost. We have more powerful numerical techniques and models for constitutive laws, where disciplines beyond those specific to geodynamics such as numerical science, material science and computational science have progressed. Furthermore, more well proven and established libraries are available, when chosen and applied appropriately, lead to less work and for better results. How can we capitalize on this? We propose a multi-level community development model that allows computational scientists, numerical scientists, material scientists and phenomena modelers to develop their core interests and leverage off each other's work. In turn, we aim to produce adaptable code bases where core technologies can be interchanged. That is to provide modelers with the tools to venture beyond their present scope, whilst proving infrastructure builders real problems to test against. Furthermore we aim to match the language to the expectations of the people at those levels. We facilitate this environment through a fundamental framework named StGermain. We demonstrate its application to a spectrum of (long time scale) geodynamics problems encapsulated in the Underworld package. In particular we show three phenomena models from Underworld: mantle convection, roll back, and basin extension modeling. These are all at different scales, incorporate different physics, different work flows and maintained by different people. The methodology and technology is being applied to solve multi scale, multi physics modeling of materials manufacturing problems.
DE: 0545 Modeling (4255)
DE: 0594 Instruments and techniques
DE: 3200 MATHEMATICAL GEOPHYSICS (0500, 4400, 7833)
DE: 4475 Scaling: spatial and temporal (1872, 3270, 4277)
DE: 8100 TECTONOPHYSICS
SC: Earth and Space Science Informatics [IN]
MN: Fall Meeting 2005