topbar
| Outline of Heart Simulator | Research Subjects | Simulation of Diseases and Development of New Therapies |
| Development of Therapeutic Devices | Experiments on Cells and Virtual Cells | Experimental Validation | Computational Science | etc. |

Introduction of Research
Computational Science 4/4

<=PREV | NEXT=>
In the simulation of electrical and mechanical phenomena of the heart, we eventually encounter problems with repeatedly solving large-scale simultaneous equations with millions or tens of millions of dimensions. The part of the simulation program which solves the simultaneous equations is called Solver. Large-scale simultaneous equations are usually solved using iterative methods, but because of the properties of the physical phenomenon and the employed numerical method, the mathematical characteristics of the coefficient matrix change significantly. A strategy to develop an optimal solver is therefore important. As a single processor is no longer enough to deal with the whole heart, the domain is decomposed into many sections, and parallel computations are performed with communication between neighboring subdomains. This research is both mathematically interesting and of practical importance. Eventually, collaboration will be required between computational science and computer science. The figure on the right shows the decomposed domains for massive parallel computations.

Computational Science



All Rights Reserved, Copyright(C)2013, UT-Heart Laboratory