An approach to distributed interactive simulation and visualization of complex systems using cluster computing

When dealing with complex systems, interactive, realtime simulations require significant computational capabilities that can be provided by cluster computing. Current cluster computing based techniques are mostly focused on batch jobs. However, it is possible to use clusters so that an application can run and directly communicate with the remote client(s). Direct communication enables, without loss of accuracy or frame rate, real time visualization of and interaction with much larger models compared to a single machine implementation. The degree of coupling between the dependent variables in the model determines the degree of parallelization that can be achieved by evaluating the solution for each dependent variable in parallel. A distributed mass-spring simulation system was developed to serve as an open platform that can be used to improve the scalability of the simulation computation. Several techniques are used to improve scalability, both in terms of the problem size and number of clients. The developed system provides support for large scale mass-spring simulations to leverage available cluster computing and visualization resources. It can be applied to a wide range of problems related to de-formable solids including many biologically related like human organ modeling and medical animation where realtime feedback is required.