A parallel algorithm for constructing approximate attainable sets of nonlinear control systems

The paper investigates the effectiveness of shared memory parallel programming approach for constructing approximate attainable sets of nonlinear control systems in a finite-dimensional Euclidean space. In this study, we propose a parallel iterative algorithm for constructing approximate attainable sets employing a regular Cartesian grid for spatial discretization. The proposed algorithm has been designed for implementation on SMP systems and handles such issues as data decomposition, threads synchronization and distribution of work between multiple threads. Numerical experiments on a system with two quad-core processors confirmed a high efficiency of shared memory parallel programming approach for applying grid-based methods to construct approximate attainable sets.