As SciDAC program enters its third cycle (in 2011), four SciDAC Institutes have been established with 24 participating institutions and a total annual funding of $15.5 million. The mission of these SciDAC Institutes is to provide intellectual resources in applied mathematics and computer science, expertise in algorithms and methods, and scientific software tools to advance scientific discovery through modeling and simulation. The development of SciDAC tools and resources by the Institutes is intended for computational systems such as those existing and planned for at the Oak Ridge and Argonne Leadership Computing Facilities, the National Energy Research Scientific Computing Center, and similar world-class computing facilities over the next 5 years. Specific goals and objectives for the SciDAC Institutes are:

• Tools and resources for lowering the barriers to effectively use state-of-the-art computational systems;
• Mechanisms for taking on computational grand challenges across different science application areas;
• Mechanisms for incorporating and demonstrating the value of basic research results from Applied Mathematics and Computer Science; and
• Plans for building up and engaging our nation’s computational science research communities.

Although the work of each Institute is not science application-specific, it is application-, architecture-, and Institutes-aware:

Application-aware. One of the primary metrics for the success of the SciDAC Institutes is the extent to which its deliverables are used by application scientists therefore active collaborations among Institutes participants and their domain science partners constitute an integral part of the SciDAC Institutes.

Architecture-aware. The main architectural features of existing and planned computing environments, over the next five years, include heterogeneous nodes (CPUs, GPUs), different memory hierarchies, and varying trade-off costs for computation versus data movement. SciDAC Institutes develop tools and methodologies for coping with and taking full advantage of such architectural complexities.

Institutes-aware. Innovative science projects can be accommodated by the SciDAC Institutes’ pooling of a broad range of computational skills that is otherwise not readily available to DOE domain scientists therefore a key point of SciDAC Institutes is the extent to which Institutes researchers actively collaborate and leverage their expertise in achieving success.

Current SciDAC Institutes

FASTMath – Frameworks, Algorithms, and Scalable Technologies for Mathematics
Topic areas: Structured & unstructured mesh tools, linear & nonlinear solvers, eigensolvers, particle methods, time integration, differential variational inequalities

Institute Director: Lori Diachin (diachin2@llnl.gov)
Lawrence Livermore National Laboratory

QUEST – Quantification of Uncertainty in Extreme Scale Computations
Topic areas: Forward uncertainty propagation, reduced stochastic representations, inverse problems, experimental design & model validation, fault tolerance

Institute Director: Habib N. Najm (hnnajm@sandia.gov)
Sandia National Laboratories

SUPER – Institute for Sustained Performance, Energy and Resilience
Topic areas: Performance engineering (including modeling & auto-tuning), energy efficiency, resilience & optimization

Institute Director: Robert F. Lucas (rflucas@isi.edu)
University of Southern California

SDAV—Scalable Data Management, Analysis and Visualization
Topic areas: Data management, data analysis, data visualization

Institute Director: Arie Shoshani (shoshani@lbl.gov)
Lawrence Berkeley National Laboratory