- Christian Clason, Universtät Graz
"Computational methods for inverse problems in Banach spaces"
inverse problems were predominantly posed in Hilbert spaces. Although
this is indeed the correct setting for many physical models (e.g.,
those involving energy terms), it is just as often simply due to
convenience and numerical tractability, and a Banach space setting
would be more natural, e.g., for parameter identification problems for
partial differential equations, image processing using
total-variation-type penalties, or inverse problems subject to
non-Gaussian noise. While the regularization theory for inverse
problems in Banach spaces has seen tremendous progress in the last
decade, computational methods for their solution have been less well
studied. However, their development is a crucial issue since they need
to reflect the relevant structural properties of the Banach space
setting even after discretization.
The goal of this
minisymposium therefore is to bring together researchers working in
this area, whose contributions range from the development and analysis
of general Banach space algorithms to use in concrete applications.