**Location:** MIC-Arena, Maria-Goeppert-Str. 3, 23562 Lübeck

**Title:** Parameter Estimation with Reduced Models

**Abstract: **In most imaging applications, the quantity of interest is reconstructed from indirect measurements. For example, one may be interested in estimating parameters of a partial differential equation (PDE) describing a particular physical process. In the absence of analytical solutions, iterative reconstruction schemes are commonly used. Typically a very large number of solutions to the discretized PDE have to be computed until a satisfactory reconstruction quality is achieved. Thus, one way to increase the efficiency of such reconstruction methods is to reduce computational costs involved with the forward model by, for instance, projecting the PDE into a small dimensional subspace, a technique also referred to as model order reduction.

This talk discusses some state-of-the-art methods for model order reduction techniques and presents an extension that updates the subspace dynamically in the optimization process. The potential of the novel approach is outlined using the DC resistivity problem that arises in applications of geophysical imaging.