Objective: The broad objective of this project is to solve both relativistic gravitational field and hydrodynamic equations in the study of relativistic astrophysics and gravitational wave astronomy. Approximate solutions must remain faithful to the true solution by conserving certain quantities such as energy and enforcing constraints such as the Hamiltonian (a fundamental expression in formulating the solution). A particular objective of this project has been to develop high-performance codes to conserve physical quantities and enforce constraints. These codes will enable astrophysicists to study stellar core collapse and supernovae and the formation of compact objects such as black holes and neutron stars.
Accomplishments: In the past year, the project has developed numerical codes that successfully solve the vacuum Einstein field equations in full 3D. This code (actually a family of codes) is written in portable data parallel Fortran and MPI and runs efficiently on a variety of massively parallel and parallel/vector machines. Numerical procedures for conserving important quantities and for enforcing constraints have been combined with this code.
Significance: The significance is threefold. First, software to aid in the opening of new frontiers in astronomy. Applied to the coalescence of binary systems, such as two black holes in orbit around each other, this code and others in the family can yield information, for example, on the Hubble constant and gamma ray bursts. Second, extensive experience and results in the development of high performance on distributed-memory machines (e.g., Thinking Machines Corp. CM-5, CRAY T3D, IBM SP-2, Intel Paragon), distributed-shared-memory machines (e.g., Convex Exemplar), and traditional shared-memory vector supercomputers (e.g., CRAY C90). Third, the development of algorithms to set up and solve the equations expressing the conservation/constraint of physical quantities. This is important in obtaining approximate solutions that make sense.
Status/Plans: A general-purpose, parallel, scalable code for relativistic astrophysics is under development. New ideas for incorporating techniques from the computational fluid dynamics community will be explored together with advanced solution techniques to conserve physical quantities and enforce constraints.
Points of Contact:
Paul Saylor