Kentucky-Austria International Cooperative Research
Project Summary
Fast Solvers for Computational Problems Arising in Pharmacy, Life Sciences, Mathematics, Physics, and the Environment
Craig C. Douglas and Robert Lodder (USA)
Gundolf Haase, Ulrich Langer, Walter Zulehner, and Alfio Borzi (Austria)Intellectual Merit of the Proposed Activity. This project will enhance research in computational science on state-of-the-art high performance computers, clusters, and intercontinental GRID computing environments. We will extend parallel methods to solve very large scale problems in computational neurology, science, and engineering using hardware assisted algorithms.
The applications we will investigate include computational neurology including imaging and noninvasive blood flow analysis, ocean modeling, and general computational fluid dynamics. We will develop better algorithms, including ones for algebraic multigrid, parallel methods for solution of partial differential equations, saddle point problems, eigencomponents, optimization methods, and hardware assisted methods. The new algorithms will be incorporated into the applications in order to demonstrate usefulness besides just creativity.
Broader Impacts Resulting from the Proposed Activity. This project will solidify broad based collaborations between faculties at three universities: one U.S. (University of Kentucky) and two Austrian (Johannes Kepler Universitaet Linz and the Karl-Franzens-Universitaet Graz). There are already ties between faculty members at the universities, but not much interaction between students.
We will expose a number of U.S. graduate students to international research opportunities, including minority or Appalachian students (all four students we are considering at the University of Kentucky are one or the other). Typical University of Kentucky students do not have much experience with top flight, internationally oriented universities and are at a serious disadvantage upon graduation. Our project will give the students an edge they cannot hope for by just staying on campus. The experience they will receive from this project will be similar to doing a postdoctoral fellowship before graduation. We will disseminate results and codes through well established web sites, archival journals, and by presenting results at international conferences and workshops.
Expected Mutual Benefit for Participants. We will vastly broaden the horizons of University of Kentucky students, who typically only see local researchers or one day visitors whose schedules do not include more than an hour to talk to students. While visitors sometimes motivate students, there is no replacement for actually working for an extended time with the same visitors and their own research groups.
We will exposes foreign faculty, students, and research staffs to American trained personnel. This will be enhanced by the U.S. members teaching short courses and participating in active seminar series in Austria during the summer months. We expect students to give short, formal talks weekly on progress while in Austria.
Both faculties will personally benefit from exposure to a larger and diverse audience for their research skills and will be able to disseminate results to a much wider audience. We expect a much larger production of multi-author papers for archival journals.
Methodology. The University of Kentucky summer period (3 months) corresponds to the Austrian spring term, when their faculty are in residence and have time to work with visitors. Our students and PIs will perform research in a number of areas outlined in the proposal while sitting in Austria and working directly with the Austrian participants. Further work will be done from a distance during the rest of the calendar year with electronic exchanges through CVS repositories and regular teleweb exchanges.
Numerical Methods
- Algebraic Multigrid
- Domain Decomposition Methods and Parallelization
- Saddle Point Problems
- Eigenvalues
- Constrained Optimization in Applications
- Cache Aware and Hardware Assisted Algorithms
Applications
Other Links
- Research reports
- Software
Craig's Home U. of Wyoming ML-DDDAS Group Preprints Free software Class webs: All
UWyo MA 5490Digital pictures:
Places FamilyJournals:
Computing JACT JOCSScientific Community Web Sites:
DDDAS MGNet Minimus ***** Open Positions *****