1. Physics Steven Gottlieb, NCSA/Indiana University Lattice QCD: focus on one area I understand well. A central aim of calculations using lattice QCD is to determine the underlying parameters of the Standard Model (SM) by stripping away the effects of the strong interactions, thus providing accurate determinations of the masses of the up, down, strange, charm and bottom quarks, the strong coupling constant, and the values of the weak transition coupling between quarks—i.e., the elements of the Cabibbo-Kobayashi-Maskawa (CKM) matrix. These calculations are essential to interpret high energy experiments done around the world that cost several hundred million dollars per year. NCSA Strategic Planning Presentation (April 20,2010)

2. NCSA has expertise in supporting codes for lattice QCD due to its support for Blue Waters. QCD PACT (petascale problem and NSF benchmark) Sugar’s PRAC team In addition, the Innovative Systems Lab has ported QCD code to the IBM Cell B.E. and to NVIDIA’s C for CUDA. Within the Physics Department, Aida El-Khadra is a leader in the field. However, John Kogut has left the department for the DOE, and John Stack has moved into administration. NCSA Strategic Planning Presentation (April 20,2010)

3. Bottlenecks/Issues to Achieving Objectives To improve control of systematic errors to the level required by experiments, calculations taking about 100 petaflops-years are desired. Developing applications for an exascale architecture may require collaboration between disciplinary experts, computer scientists and applied mathematicians. New algorithms are desired, if not required. Training of students and postdocs in advanced science and computational science will be essential. (Common theme in exascale reports.) Lattice QCD community has been involved with designing high performance computers (QCDOC, BlueGene, QPACE, etc.) and would likely get involved in co-design efforts. NCSA Strategic Planning Presentation (April 20,2010)

4. Cyberinfrastructure Challenges in Reaching the Objectives Generally require very high floating point and network performance, but not as much memory as other applications. Lattice generation does not require that much I/O bandwidth or disk storage. file sizes about 900 GB to 14 TB archival storage 2PB to 28 PB External networking requirements would be modest, if analysis can be done on same machine as lattice generation. Some help with parallel I/O strategy might be helpful, though this may be worked out on Blue Waters. Again, an interdisciplinary team may be required to deal with performance and algorithmic issues. NCSA Strategic Planning Presentation (April 20,2010)

5. Scientific Grand Challenges Workshop Series http://extremecomputing.labworks.org http://extremecomputing.labworks.org NCSA Strategic Planning Presentation (April 20,2010) Ten workshops, so far. Most relevant for physics are: High Energy Physics, Dec. 9-11, 2008 Nuclear Physics, Jan. 26-28, 2009 Fusion Energy Sciences, Mar. 18-20, 2009 Nuclear Energy, May 11-12, 2009 Basic Energy Sciences, Aug. 13-25, 2009 Breakout sessions listed for these workshops on next slides.

6. High Energy: Cosmology and Astrophysics Simulation High Energy Theoretical Physics Accelerator Simulation Astrophysics Data Handling, Archiving and Mining Experimental Particle Physics Nuclear Physics: Nuclear structure and nuclear reactionsNuclear astrophysicsNuclear forces and cold QCDHot and dense QCDAccelerator physics

7. Nuclear Energy: Integrated Nuclear Energy Systems Materials Behavior Verification, Validation and Uncertainty and Risk Quantification System Integration Basic Energy Sciences: Correlation and Time DependencePhotovoltaic FundamentalsEnergy StorageDynamics

8. Fusion Energy Sciences: Burning Plasma/ITER Science Advance Physics Integration Plasma-Material Interaction Science High Energy Density Laboratory Physics/Laser-Plasma Interactions Basic Plasma Science/Magnetic Reconnection Physics Scalable Algorithms Data Analysis, Management and Visualization Mathematical Formulations Programming Models, Frameworks and Tools