Computer Models of Reality and High-performance Simulation

Multi-scale nature of the real world and multi-dimensional and multi-resolutional character of patterns it creates, involve construction of new methodological and computational tools. We concentrate on discrete particle systems and cellular automata as a common methodological basis for developing such the tools. We elaborated the principles of multi-scale homogeneous computational model based on gridless discrete-particle methods. Our model was employed for simulating thin-film flow evolution, fluid instabilities, creation of colloidal crystals, the growth of colloidal agglomerates, eventually; we attacked the problem of modeling the blood flow in micro-capillaries.

The research on modeling of reality focuses on discrete particle methods as a powerful universal solver for simulating macroscopic phenomena, which are out of scope of standard computational methodologies. Modeling of living organisms, evolving populations, geological phenomena (such as river systems), crowd and traffic simulations, involves both off-lattice discrete-particles and on-lattice cellular automata, Monte-Carlo methods and genetic algorithms. We elaborated unique simulation methodology involving new parallel algorithms and programming tools, exploiting high-performance computational (mutiprocessor systems) and visualization facilities (Power Wall).

Our research was sponsored by many Polish (Polish Committee of Scientific Research KBN) and USgrants (e.g., Energy Research Laboratory Technology Research Program, US Department of Energy). The research is being performed in tight cooperation with:

• University of Minnesota,
• GdañskUniversity of Technology,
• Universityof ColoradoBoulder, Department of Mechanical Engineering

Researchers: W.Alda, K.Boryczko, W.Dzwinel, P. Topa, J.Kitowski

Blood clotting (SGI ALTIX 32 processors) Rivers modeling employing CA