Abstract:
A general approach to the development of complex systems simulation is proposed. The key computer applications problem is the high-dimensional grid approximations and high-performance supercomputers with a large number of parallel CPUs. Kinetic methods for solving differential equations and methods for "gluing" exact solutions produced with coarse meshes are developed as possible alternatives to high-dimensional grid approximations.