The Scientific Discovery through Advanced Computing (SciDAC) Initiative of the U.S. Department of Energy is dependent upon a level of cooperation in algorithm and software development among distributed projects with diverse scientific goals that is unprecedented in its size and complexity. SciDAC program goals, especially the goal of high scalability and performance for the production codes expected to enable breakthrough scientific discovery, will not be met by a mere a priori negotiation on abstract interfaces (e.g., matrix and vector data structures, or nonlinear function calls) Instead, close collaboration and mutual education of participants by each other will be required, as code interfaces evolve under influences that suggested during algorithm-application collaborations.
This workshop seeks to expose to the scalable iterative algorithm community to some application areas being addressed under SciDAC represent new challenges for scalable solvers. Three members of applications SciDAC teams will describe their algorithmic needs and state of the art. They will be introduced by the coordinator of the "Terascale Optimal PDE Simulations" integrated software infrastructure center, also created under SciDAC, in which many members of the Copper Mountain community participate.
List of speakers: