It turns out, and this is kinda crazy, but when you delete a row and a column from a matrix, it stops having the same eigenvectors. So, like, if you're trying to build something using the eigenvectors of the adjacency matrix of a graph (don't think about it too hard, it's weird), and you want the end product to be robust to the loss of a vertex in
(
Read more... )