Modern Parallel Optimization Techniques in Scientific Computing Applications

There exist significant, well established code bases in the scientific computing and research communities. Many of these code bases have been updated to work in parallel across clusters, but do not take advantage of modern parallelism techniques. The goal of this research is to apply these techniques to existing code bases while maintaining tool portability across individual workstations, lab compute clusters, and leadership class supercomputers; while preserving maintainability and extensibility of the original code base