Simulating large volumes of granular matter

Master of ScienceDepartment of Computer ScienceDaniel AndresenModern techniques for simulating granular matter can produce excellent quality simulations, but usually involve a great enough performance cost to render them ineffective for real time applications. This leaves something to be desired for low-cost systems and interactive simulations which are more forgiving to inaccurate simulations, but much more strict in regards to the performance of the simulation itself. What follows is a proposal for a method of simulating granular matter that could potentially support millions of particles and several types for each particle while maintaining acceptable frame rates on consumer level hardware. By leveraging the power of consumer level graphics cards, effective data representation, and a model built around Cellular Automata a simulation can be run in real time