Programming Paradigms and Algorithms for Electronic Structure Calculations on clusters of Non-Uniform memory Access Parallel Processors

Cluster computers have emerged as the platform of choice for scientific comptuting. The next generation of these systems will have "fat" nodes that utilize the latest multi-core processor technology. This project will develop software tools and methods for these systems, with an aim of enabling a more productive utilization of these architectures for scientific computation. Our focus is on electronic structure methods, particularly those methods where the cost of the computation scales linear with the number of atoms in the system. Our goal is to develop scalable parallel implementations of these methods that can be used to perform computations on nanoscale systems, such as enzymes and molecular electronic devices.