Solving Poisson's equation on a multiprocessor system in problems of incompressible fluid flow simulation (Q1769022)

The recent development of multiprocessor computer systems permits one to carry out numerical simulation of complicated physical phenomena, including hydrodynamic flows, in the 3D statement. Special algorithms are required to use many processors effectively. Such algorithms must not only adequately describe the physical picture but also take account of the peculiarities of computer systems. In particular, it is of interest to consider algorithms with linear (or almost linear) growth of throughput with the number of processors. The construction of mathematical models of hydrodynamic flows (as well as a number of other problems, including electronic optics, the analysis of the influence of charged particles on surfaces, and others) often requires the solution of a boundary value problem for the Poisson equation. Various numerical methods were developed for such problems with regard to the symmetry of the corresponding system of linear equations [\textit{Kh. D. Ikramov}, Numerical methods for symmetric linear systems (in Russian), Moscow (1988; Zbl 0636.65017), p. 56]. In the present paper, we use the method of [\textit{I. A. Belov, S.A. Isaev} and \textit{V. A. Korobkov}, Problems and methods for the analysis of separated flows of an incompressible fluid (in Russian), Leningrad (1989), p. 47], which is a modification of the Thomas method, for the solution of the Poisson equation for problems in the two- and three-dimensional space statement. Of its advantages, we note the simplicity of multiprocessor implementation, the resulting highly efficient parallelization, and easy adaptation to various types of boundary conditions. In the present paper, as an example of the use of the above-mentioned method for the solution of the Poisson equation, we consider the viscous incompressible flow simulation in a cavity and flow simulation for a viscous electrically conducting fluid in a cylinder in the presence of a magnetic field.