Project 18 considers whether it's possible to reuse the factors of a matrix A after you've changed the matrix.
More precisely, suppose A is a matrix, and we've computed the LU or QR or SVD factors of A. Suppose we decide to change a single entry of A, or a row or column of A, or to make some other "small" change in A. This might be because we are adjusting a model, for instance, or varying some parameter.
Since matrix factorization is so expensive, it can be worthwhile to see if small changes to a matrix only require small adjustments to its factorization.
The case study involves a truss model, which is analyzed, and then adjusted;
You can go up one level to the Computational Science Projects page.