Orthogonal SVD — orthonormalize named axes via QR, fill the rest from SVD

Mirrors rENA's orthogonal_svd() in ena.rotate.by.mean.R: the named axes in the output are the orthonormalized Q columns, not the original weights columns. Use complete_rotation to keep the named axes verbatim.

Usage

orthogonal_svd(data, weights, named_labels)

Arguments

data

Numeric matrix (n_units x n_dims)

weights

Numeric matrix (n_dims x k); columns are the named axes

named_labels

Character vector of length k

Value

List with rotation, eigenvalues, column_names