During my visit, I focused on a new hybrid domain decomposition method combining the FETI and FETI-DP methods
that delivers better parallel scalability on a very large number of processors. I also worked on its C++ implementation using the Intel MKL library,
and on the parallelization of the local coarse problems using Intel Xeon Phi accelerators.
Additional Information
Education:
Ph.D. Applied Mechanics, 2009, VSB Technical University of Ostrava, Ostrava, Czech Republic Research Scientist. National Supercomputing Center IT4Innovations, Ostrava, Czech Republic