Asking same question as Running cor() (or any variant) over a sparse matrix in R, but that solution only computes all-pairs, cor(x), is there a way to add the y argument to this computation, cor(x,y)?
Using a sparse cor(x,y) would allow an easy port to %dopar% to make it even faster by chunking across column sets like https://gist.github.com/LW6EGE/651feb519983b2fa1fadf52f4e44bc55.
