Performs matrix residualisation over estimated canonical vectors. There are three types: basic (subtracts scaled estimated latent variable from data), null (uses the null space of the estimated canonical vector to construct a new matrix) and LV (uses SVD to residualise).
Useful links