This project is going to be working to create functions to perform 1-rank updates, downdates and additions to a given SVD. This is based off of work by Matthew Brand of Ohio State University. His work that we are focused on, is using a thin decomposition of a matrix which is all that is needed for many purposes. The main objective for creating a SVD update algorithm is to create a means to avoid the high cost of recomputing the SVD and high memory usage involved in storing large datasets. This is greatly used to break down huge data sets into smaller, more managable sets.
References: http://www.merl.com/papers/docs/TR2006-059.pdf ; http://www.stat.osu.edu/~dmsl/thinSVDtracking.pdf
Created By: Taylor Steiger James Pak