High-Performance Computations in Multi-agent Simulation Problems of Percolation Cluster’s Behavior

Аннотация

Use of supercomputers for solving problems of multi-agent percolation clusters growth simulation is considered. An improved version of Hoshen–Kopelman multiprocessing algorithm of percolations cluster multiple labeling is offered. Possibility of detection of the latent periods of epidemics spreading dependencies from probability of an infection of units of population representatives is shown, and a way of threshold values formation when local epidemics can evolve into large-scale pandemics is presented.