An efficient discretization to simulate the solution of linear-quadratic stochastic boundary control problem
Abhishek Chaudhary (University of Tubingen, Germany)
| Speaker |
Abhishek Chaudhary (University of Tubingen, Germany)
|
|---|---|
| When |
Feb 28, 2024
from 04:00 PM to 05:00 PM |
| Where | LH-006 (TIFR CAM) |
| Add event to calendar |
vCal
iCal
|
Abstract
In the first part of the talk, we will explore a practical approach to discretizing a linear-quadratic control problem associated with a stochastic heat equation with a non-homogeneous Dirichlet boundary. This leads to a discrete version of Pontryagin's maximum principle, involving the discretization of a backward SPDE. Typically, solving the BSPDE equation requires simulating conditional expectations, but we will introduce a reformulation of the discrete optimality condition that eliminates the need for such simulations. Moving on to the second part of the talk, we will delve into the strong rate of convergence results. Moreover, we also discuss the challenges in analyzing errors, particularly due to the less smooth nature of the optimal control.