SUsh4lTTED TO WORKSHOP ON OPENm AF'PLICATIONS AND TOOLS OpenMP-based Frameworks for Interoperable Structured Adaptive Methods Dinshaw S. Balsara National Center for Supercomputing Applications and Center for Simulation of Advanced Rockets University of Illinois at Urbana-Champaign dbalsara@ncsa.uiuc.edu Charles D. Norton National Aeronautics and Space Administration Jet Propulsion Laboratory, California Institute of Technology High Performance Computing Systems and Applications Group and Center for Space Mission Information and Software Systems nortonc@bryce.ipl.nasa._pov Extended Abstract Parallel adaptive mesh refinement (AMR) is an important numerical technique that leads to the efficient solution of many physical and engineering problems. While some AMR libraries have been designed, there are many advantages to considering alternative approaches based on language paradigms and standards. Furthermore, it is even more desirable to develop a framework that allows one to easily compose solutions to new problems where solvers and multi-grid methods can interoperate freely. This kind of framework-oriented design is even amenable to the features of computational power grid processing. This abstract describes recent work where a very general approach to AMR has been devised by combining the best aspects of object-oriented programming using modern aspects of Fortran 90/95, the parallelizing features of OpenMP, and solvers designed for interoperability. The approach combines efficiency, portability, and maintainability for the application scientist, building onresults described in Balsara and Norton [2]. Our approach, based entirely on well-defined standards, reduces programming complexity, preserves the investment inexisting Fortran-based solvers, and benefits from years of compiler optimization techniques. Our very general approach is scalable, efficient, and complete, integrating emerging trends in high performance computing with the desire to create interoperable frameworks that simplify the modification of simulationsto new problems. Our work has been applied toBalsara's RIEMANN framework, see Balsara [ 11 and references therein.