TABLE 5.2 Support for Astrophysics and Cosmology Theory


Budget (million $)

NSF-AST Astronomy and Astrophysics Research Grants


NSF other (AAPF, CAREER, Cyberinfrastructure, and others)


NSF-PHY Astrophysics and Cosmology Theory


NSF-PHY Physics Frontier Centers


NASA Astrophysics Division Astrophysics Theory Program


NASA Astrophysics Division Great Observatories Guest Observer Programs


DOE Scientific Discovery through Advanced Computing


DOE High Energy Physics Theory


DOE Nuclear Physics Theory


understand the formation of structure in the universe, the explosion of massive stars, the evolution of our solar system over billions or trillions of years, and how a complex experiment works. They are also essential to processing astronomical images whose sizes now exceed 1 billion bytes (a gigabyte) into data that are usable by the astronomical community. The largest codes may have in excess of a million lines and run on supercomputers that have more than 100,000 cores, generating data sets that occupy 1 trillion bytes (a terabyte) of storage. These codes are now an indispensible part of the astronomical enterprise. However, they often require teams—scientists, computer professionals, applied mathematicians, and algorithm specialists—to create, maintain, and constantly develop them.

NSF, NASA, and DOE have made substantial investments in high-performance computing (HPC) over the past decade, making available close to a petaflop of sustained computing power to the astrophysics community. Such facilities enable cutting-edge theoretical calculations and analyses that push the astrophysics frontier. Future progress in supercomputer power will come from further parallelization, with the largest systems evolving from 104 to 105 processor cores today to perhaps 108 to 109 cores by the end of the decade.6

These capabilities will enable qualitatively new physical modeling.7 Exploiting the new computer systems will require new software codes and sustained support for focused research groups. At the same time, strategic balance should be maintained between investment in HPC and hardware resources for individual investi-


Such large increases in processing capability carry implications for the amount of power and cooling that will be necessary. On the presumption that the total power usage cannot increase significantly in a “green” computing future, major advances in chip design and special-purpose software will be necessary.


Simulations in cosmological structure formation, galaxy formation, stellar evolution, supernova explosions, gamma-ray bursts, star formation, planet formation, and high-energy particle acceleration are just a few example areas.

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement