Data e ora inizio evento:
Data e ora fine evento:
Sede:
Dipartimento di Matematica Guido Castelnuovo, Università Sapienza Roma
Aula:
Sala di Consiglio
Speaker ed affiliazione:
Luca Saluzzi, GSSI
The classical Dynamic Programming (DP) approach to optimal control problems is based on the characterization of the value function as the unique viscosity solution of a Hamilton-Jacobi-Bellman (HJB) equation. The DP scheme for the numerical approximation of viscosity solutions of those equations is typically based on a time discretization which is projected on a fixed space triangulation of the numerical domain. In this talk, we will discuss a new approach for finite horizon optimal control problems where we compute the value function on a tree structure built directly by the time discrete dynamics avoiding the use of a space triangulation to solve the HJB equation. This allows to drop the cost of the space interpolation and guarantees a perfect matching with the discrete dynamics. We will also provide error estimates for the algorithm if the dynamics is discretized with an Euler method. Furthermore, this approach has been extended to high-order schemes and we will show some examples of second order approximation schemes. Finally we will show the effectiveness of the method for the control of PDEs, considering the coupling of the method with the Proper Orthogonal Decomposition. This is a joint work with Maurizio Falcone and Alessandro Alla (PUC).