The article considers the issue of modeling the kinetics of charged particles in the interelectrode space during plasma-arc synthesis of carbon nanostructures in the presence of a catalyst using a modified numerical method of large particles. The main focus is on adapting standard molecular dynamics algorithms to work with a large number of particles and their interactions at the macroscopic level, as well as developing effective methods to speed up the simulation process using personal computers. A modification of the numerical method is based on the adaptive distribution of large particles depending on the collision activity in local regions of the interelectrode space. In areas with a high collision frequency, the weight of macroparticles decreases and their number increases to ensure the accuracy of describing a large number of interactions, while in areas with a low collision frequency, the opposite is true. A method for balancing the load of computing cluster workstations and calculating the required number of cores is proposed, based on the assignment of computing resources depending on the number of large model particles in the calculation area. It is established that the developed methodology and algorithms for processing large amounts of data based on parallel calculations necessary for solving multidimensional nonlinear problems using the resources of a distributed computing system with processor load balancing can minimize the processor calculation time. The maximum acceleration was obtained by 266 times compared to the sequential algorithm, while the parallel algorithm using a GPU allows calculations to be accelerated only by 153 times. The work has the potential to further improve methods for the synthesis of carbon nanostructures and the development of high-performance computational tools in the field of nanotechnology.
Kalach et al. (Sat,) studied this question.