Speaker: | Wolfgang Bangerth, Texas A&M University |
Title: | Numerical methods for nonlinear inverse problems |
Time: | 4:00-5:00 pm |
Place: | Milner 317 |
In biomedical imaging, as in many other applications such as geophysics or materials testing, one would like to reconstruct three-dimensional spatially varying material parameters from indirect measurements. If the relationship between parameters and measurable quantities is a partial differential equation, the reconstruction problem is refered to as an "inverse problem".
Realistic nonlinear inverse problems are enormously complicated to solve numerically. In this talk, I will present recent results in the development of methods for such problems. In particular, I will introduce error estimates for inverse problems, and motivate and illustrate results with an application in tumor detection. I will then discuss extensions of the approach to the problem of designing experiments in a way that is optimal for the reconstruction of parameters.
Last revised: 08/29/08 By: abnersg@math