![]() |
ИСТИНА |
Войти в систему Регистрация |
ИСТИНА ИНХС РАН |
||
Work proposed examines the problem of the super-computer simulation of the thermodynamic equilibrium of the microsystems, which contain substances, which are found in different states of aggregation at different temperatures. Together with the problems of numerical simulation is examined the possibility of the interpretation of the results of the calculations in the form of 3D visualization of the calculated region, which reflects, besides the positions of particles, different parameters, for example temperature, the distribution of pressure and particle trajectories. As an example we examined gas-metall microsystem, that consists of approximately 10^7 particles, where as the gas is used nitrogen, as the metal - nickel. This choise is caused by the fact that this microsystem is the basis of many technical applications. In particular it is used in the installations of the supersonic cold gas-dynamic spraying of nanos-particle on the surface of promising materials. During the first stage studies was calculated the state of equilibrium of nitrogen-nickel microsystem. For the simulation of the process of the establishment of thermodynamic equilibrium in this system it was used molecular dynamics approach. The chosen numerical algorithm was based on finite-difference Verle scheme. For accelerating the calculations parallel algorithm and its realization within the MPI and OpenMPI framework is proposed. Using developed program, the process of the establishment of thermodynamic equilibrium was studied both in the clean components (nitrogen and nickel) at several temperatures, including room temperature, and in the nitrogen-nickel system. In the numerical experiments were defined both the optimum parameters of calculation procedure, including strategy of execution in parallel with the use of processors with different architecture and physical parameters of the process being simulated. Also in this work was developed a software complex for visualization and postprocessings of the results of molecular dynamics simulation both in the active phase of calculations and after the completion of calculations. Developed software makes it possible to use both the personal computer and clusters, thereby significantly increasing the speed of processing and visualization of large data sets. As the tools of the development of this software complex were chosed Python, numba and ParallelPython, those making it possible to considerably accelerate calculations without resorting to low-level programming.