Make slides and/or notes connecting column correlation with eigenvalues

[davidrosenberg]

One basic argument: use variational characterization of minimum eigenvalue. Then look at what happens when you plug in the difference between x1 and x2 (two correlated variables). this gives an upper bound on the lowest eigenvalue.