February 15, 2019 - 1:30pm

**Title**: Hierarchical Matrix Compression in PDE Constrained Optimization by Tucker Hartland

**Abstract**: Effective preconditioners can dramatically reduce the computational cost of solving a linear system of equations. In this talk, we will discuss hierarchical off-diagonal low rank (HODLR) approximations and their application - as preconditioners - in partial differential equation (pde) constrained optimization. We will motivate why the Hessian has off-diagonal low rank structure, as well as pde discretization issues that should be addressed to most effectively exploit the underlying HODLR structure.

**Time**: Refreshments at 1:30PM, Talk at 2:00PM