Normalize a Kruskal Tensor — normalize • tensory

Normalizes the columns of every factor matrix, absorbing the magnitudes into the weight vector lambda. Mirrors the MATLAB Tensor Toolbox normalize(K, N) semantics.

Usage

normalize(x, ...)

# S3 method for class 'KTensor'
normalize(x, mode = NULL, sort = FALSE, normtype = 2, ...)

Arguments

x: A KTensor.
...: Additional arguments passed to methods.
mode: NULL (default) keeps all weight in lambda; a mode number absorbs the weights into that factor matrix; 0 distributes the weights evenly across all modes (each factor absorbs lambda^(1/N)).
sort: Logical; if TRUE, sort the components by weight, descending. Only allowed when the weights remain in lambda (i.e. mode is NULL).
normtype: Norm to use for the factor columns (passed as the p in the vector p-norm; default 2).

Value

A normalized KTensor.