The Research Of Virtual Heart Parallel Computing Method Based On Cluster System

Virtual heart is using the strong ablity of computer on computing and graphic display to simulate the character of anatomy and electrophysiology of realistic heart. The mathematical model of virtual heart electrophysiology includes two aspects: the model of single myocardial cell which consists of a set of ordinary diferential equations and the model of whole heart tissue is a reaction-diffusion equation. With the increase of cell model number and the degree of complex of action potential the computing scale of virtual heart silmulation has become a bottleneck in the development of virtual Heart Study. In order to improve the speed of virtual heart simulation, the paper numerical methods and fast parallel computing has carried out two studies.First, the characteristic of the action potential is relatively independent with the reaction-diffusion process of heart tissue. So the variable adaptive time step numerical method is introduced. The precision and efficiency is analysised by compared with the Runge-Kutta method.Second, the three-dimensional cardiac tissue model contains millions of cardiac cells. Two domain decomposition methods which are the standard deviation of the weight adjustment method and dichotomy-method are proposed for parallel computing. The speedup performance is also analysised.Finally, in the 32 CPU cluster environment, using MPI technology to implement the variable adaptive time step method. The cardiac electrophysiological simulation is also reaseached.The results of simulation show that the methods proposed in this paper improve computational efficiency up to an order of magnitude. And it can simulate the ventricular electrical conduction process correctly. For the virtual heart of the research provides an efficient way.