
Performance improvement of particle-in-cell method for numerical modelling of open magnetic system
Author(s) -
Marina Boronina,
Igor Chernykh,
E. A. Genrikh,
В. А. Вшивков
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1640/1/012014
Subject(s) - gyroradius , diamagnetism , computational physics , particle in cell , physics , plasma , computation , realization (probability) , electron , magnetohydrodynamics , ion , particle (ecology) , atomic physics , magnetic field , computer science , mathematics , quantum mechanics , algorithm , statistics , oceanography , geology
The work is connected with numerical simulation of plasma dynamics in open magnetic trap in diamagnetic regime. The hybrid particle-in-cell model we develop allows to perform numerical experiments for the high ratio of Larmor radius of the ions and the one of the electrons due to the combining of the kinetic description for the ions and MHD description for the electron plasma component. The disadvantage of the model is the stability condition and the corresponding requirements for the time step. In practice a doubling of the grid nodes in each direction leads to the decrease of the time step x6 times. For the characteristic times of the plasma processes 10 2 reciprocal ion cyclotron frequencies the computations required few days for grids 100x500. The processing of the particle data takes more than 85% of the total computation time, thus its effective realization yields significant gain in performing. In our algorithm we combine the dynamic load balance and vectorized computations of densities and current densities. We present the results of numerical experiments on its basis including the highly non-uniform particle distribution in the domain and the increasing of the particle number due to the beam injection.