Eigenvalue Grid and Cluster Computations, Using Task Farming Computing Paradigm and Data Persistency
Abstract
Recent progress has made possible to construct high performance distributed computing environments, such as computational grids and cluster of clusters, which provide access to large scale heterogeneous computational resources. Exploration of novel algorithms and evaluation of performance is a strategy research for the future of computational grid and cluster scientific computing for many important applications. We adapted the well-known parallel iterative Lanczos method to compute Hermitian eigenvalues of large sparse matrices for a GRID platform and for a cluster of clusters worldwide deployed between France and Japan. Parts of the proposed GRID algorithm use an efficient task-farming computing paradigm, with data persistency scheduling strategies.