• Progress

• People

• Goals

• Developers

• Login

• Documentation

• Credits

• Machines

• Resources

• VMR Status

• Portal Master

• Project Master

• Web Master

The ASC effort is envisioned to make possible the following scenario:

An astronomy graduate student at Harvard, working on neutron star (NS) structure, obtains a library implementing an improved equation of state (EOS). She is excited to study its effect on the structure of near-critical, rotating NS configurations.

She logs in to the web-based, community-developed ASC, uploads her new EOS routine, and requests that the NS structure be computed with numerical modules that solve the full nonlinear Einstein equations with elliptic constraint equations under the stationarity condition for the rotating NS, given appropriate input parameters (e.g., spin, central density, resolution, etc.) and her EOS. The ASC assembles the required code, estimates resource requirements, locates an appropriate set of resources (in this case a collection of workstations at Harvard), and performs the computation.

ASC data analysis routines show that the maximum stable NS mass is significantly altered, and hence potentially important impact on the population of stellar-mass black holes (BH) in the galaxy. The student and her colleagues at Harvard and Berkeley decide to perform a more ambitious dynamic simulation. They use the web-based interface to the ASC to compose, in less than 30 minutes, a tailor-made simulation code comprising some ten different community-developed modules that enable a massively parallel simulation based on the Einstein equations, including a visualization module to allow real-time time monitoring by the collaborators at Harvard and Berkeley.

Again, they request the ASC to initiate the simulation. This time, there is some negotiation involved, as it turns out that the requested resolution would exceed their budget; hence, they incorporate an adaptive mesh refinement (AMR) module and resubmit the request. The computation begins, using this time a cluster of 512 NT workstations at NCSA. The scientists watch as the core of the star collapses to form a black hole after a small amount of accretion, ejecting matter and generating a gravitational wave burst. The AMR modules attempt to resolve both the inner collapsed region and the outer wave region, nearly exhausting the memory of the NT cluster. A warning is given, and possible options are suggested: continue refining, but only on a user-specified subdomain (e.g., core or outer region), terminate the simulation, or request additional resources. It determines that a large SGI cluster at NCSA plus a 1024 node IBM SP at SDSC can handle the simulation and that they are now available. The colleagues choose to acquire the additional resources, using the high speed connections between NCSA and SDSC, and the simulation is redistributed across both machines and continued.