Scientific Computing in Optimization, Simulation and Inversion

The goal of this research collaboration is to strengthen our research
in scientific computing and algorithm development with emphasis on
nonlinear and combinatorial optimization, simulation, and inversion.
Among the most promising algorithms today are those based on various
splitting techniques for subdivision of the problem as well as the
algorithm, and there is a significant overlap between the splitting
techniques currently in use within the above areas.
In this project we will coordinate the algorithm development within
our specific research areas and thus be able to draw collectively
upon progress in the individual areas. The focus of our research will
lie on the following areas: 1) new splitting techniques for
branch-and-bound algorithms in optimization, 2) space-mapping
techniques for complex optimization problems, 3) application of domain
decomposition and approximation theory in simulation algorithms,
4) preconditioning techniques (based on domain decomposition and
multilevel algorithms) for inversion algorithms, 5) methods for
including prior knowledge/side constraints in linear and nonlinear
inversion algorithms.