Optimization of C++ Numerical Simulation Algorithms Using Multithreading Methods

Abstract

The main methods of numerical simulation (finite difference method, finite element method, Monte Carlo method, Runge–Kutta method) are presented. The main parameters used to optimize numerical modeling algorithms in terms of code execution time and efficient use of processor resources are considered. The main disadvantages of multithreading related to data synchronization, deadlocks and race conditions and methods for eliminating them based on the use of mutexes and atomic operations using the Monte Carlo method as an example were analyzed.