Center for Simulation of Dynamic Response of Materials

Computational and Computer Science

In order to directly support the scientific goals described above, the mission of the Computational Science group
will continue to be two-fold.

1. Conduct research in both the algorithms and the computational infrastructure required in order to achieve
   integrated simulation capabilities of the level of sophistication and fidelity required by the multiscale research
   program discussed below and
2. Investigate the broader issues associated with conducting high-fidelity credible scientific investigations using
   diverse, distributed computing environments.

During the first five years of this project, the presence of Pyre, our integration framework, has facilitated a process whose end result is the distillation of our development efforts into reusable components and an evolutionary pattern has emerged. First, modules are prototyped as direct extensions of the existing ones in order to provide a proof of concept and explore, in a realistic setting, algorithms, data structures and the various interdependencies with other modules. Second, responsibility for all aspects of the module that are either under the direct control of the end user or are not performance critical is transferred to the integration framework and the new modules are now viewed as framework services. This step permits the decoupling of the specification of a service, i.e. its high level interface, from its implementation. Finally, the remaining lower level or performance critical aspects are re-engineered using algorithms and data structures that are optimal for the particular module. It is our experience that this process yields components with near optimal design characteristics.

The end of the first five years will find us with a number of efforts at each stage of this process. Our long-term goal is to produce a simulation framework and a family of integrated reusable components that when put together constitute a state of the art computational facility. We believe that the end result will be highly usable by experts and non-experts alike and will serve as a model for conducting scientific investigations regardless of area of concentration.

All material on these pages is Copyright © 2003 ASC/CSDRM.
Please report problems with this page to: webmaster@cacr.caltech.edu