D. Martín de Diego, S. Ferraro, R. T. Sato Martin de Almagro
In this talk we will introduce a parallelization strategy for some navigation problems that leverages the capabilities of multicore CPUs and GPUs (graphics cards). In particular, we design an iterative method for discrete variational methods appropriate for boundary value problems and its convergence properties. We apply this method to the classical Zermelo’s navigation problem, a time optimal control problem where the trajectories are affected by a drift vector field (time-dependent wind or water currents) and a fuel-optimal navigation problem. We show that our methods quickly give us a set of local optimal trajectories fixed initial and final conditions.
Palabras clave: Discrete variational calculus, optimization, navigation problems
Programado
Sesión Invitada RSME-SEIO. Investigación matemática aplicada a la sostenibilidad de la industria pesquera y transporte marítimo
8 de junio de 2022 16:00
A11
