Implementation of the decomposition method in domains in AZTRAN

This paper describes the implementation of the decomposition method in domains to the neutron transport code AZTRAN (Aztlan Transport). The code is a tool developed for the AZTLAN platform, which is written in Fortran 90 for the neutron analysis of nuclear reactors. AZTRAN numerically solves the transport equation with time dependence with Cartesian geometry in discrete ordinates. The architecture of computers today allows codes to run in parallel using their components such as processors or GPUs in order to reduce computation time. Because of this, an effort has been made to parallel codes that solve physical models such as the heat or diffusion equation, in which the domain decomposition method has been implemented. The method consists of decomposing the spatial domain into sub-domains, which will be distributed in processors so that each process simultaneously resolves the sub-domain and at the end of approximating the flow in each sub-domain the boundaries are updated. This method was implemented in the stationary two-dimensional version of the AZTRAN code using the message passing interface (Mpi), in which the Benchmark C5/G7-2D was simulated in steady state, which presented encouraging results that motivate a fully version parallel three-dimensional. (Author)