Does the built in function singular value decomposition (svd) do the correlation matrix for you?

I am doing some research on proper orthogonal decomposition using the snapshot method where I use svd. I am wondering if this function already does the correlation matrix for you or would I need to do that before I use svd?

Answer 1

[John D'Errico]

The SVD does NOT perform the re-scaling involved in a correlation matrix. It just uses the array you give it, with no preemptive scaling done at all. If you want that, you need to scale your data in advance.