Hours: Tue 9:40-12:30 BMB2

Today most real life applications require processing large amounts of data (i.e. ”Big Data”). The parts of the large data is, however, usually not independent of eachother. Hence, a natural way to respresent them is a sparse matrix. The methods that process these data usually boils down to three fundamental problems to solve: linear systems, eigenvalue problems or singular value decompositions, wether it is reducing the road noise inside a car’s body or applying the PageRank algortihm when you do a web search. In this course we will get familiar to sparse matrix computations with an em-phasis on algorithms for solving large sparse linear systems and eigenvalue problems, their efficient implementations and related data structures. Outline: (subject to change) Content: • Review of basic linear algebra, sparsity and implications of sparsity on performance. • Origin of sparse problems: Finite difference, finite elements and others • Sparse matrix structures and storage schemes • Sparse Gaussian elimination and its graph representation • Various reordering schemes to improve the sparse Gaussian eliminatio