10:15 AM-12:15 PM
Room 242
Wave propagation problems are often computationally demanding both in memory and in CPU because of the size of the physical domains and the long simulation times. Computing the RCS of a full airplane, or simulating a 3D seismic survey is currently at the limit of what is feasible on large supercomputers. Parallel computing is a way of obtaining the necessary computing power to solve these problems. The speakers will discuss their experience on different applications. They will present algorithms that address specific difficulties due to the oscillatory nature of the problems. Because of the diverse application fields, they will present a wide spectrum of techniques, covering both time and frequency domain methods, and addressing both theoretical and practical implementation issues. They will discuss some important algorithms used in 3D seismic migration and in computational electromagnetism and will present a survey of domain decomposition methods adapted to wave problems, and discuss how the new programming language Java can be used in a large-scale geophysical application.
Organizer: Michel KernWP98 Homepage | Updates| Overview | Program | Speaker Index | Registration | Inns & Hotels | Dorms | Transportation