GPU Accelerated Domain Decomposition

In the real world you often need to solve really big, spatial boundary value problems. There are a couple of examples of this:
- Solutions to general Partial Differential Equations
- Finite Element Methods (solving heat equations, antennae simulations, deformation)
- Fluid Simulations (Eulerian and Legrangian methods)
- Radial Basis Functions (smooth data interpolation)
The basic form for these systems is often the same. $f(x) = \sum_i q_i \Phi(| x - x_i |)$ where $\Phi(r)$ is some smooth kernel function and $x_i$ is an interpolation center / observation site.
The slides for the associated talk are here:
A rather out of date implementation is available on Sourceforge.