We are developing mathematical models and effective numerical
algorithms for simulating physical systems at interactive rates. Such
simulations are now critically important in several industrial
applications, ranging from medical simulation for training to computer
animation for games and feature films.
Interactive simulation has its roots in traditional scientific
computing, but it differs from the usual approaches in important ways.
One aspect is the prevalence of real time constraints in driving the
simulation. Our models employ real time I/O from a human user or from
sensors. A second aspect is the central role of inexpensive parallel
processors that have been recently developed for computer games and 3D
graphics. These new computer architectures require new types of
mathematical models and simulation algorithms.