This function is an implementation of the ρ-operator found in Currie et al 2006. It forms the basis of the GLAM arithmetic.
a n \times p_1 matrix.
a 3d array of size p_1 \times p_2 \times p_3.
For details see Currie et al 2006. Note that this particular implementation is not used in the optimization routines underlying the glamlasso procedure.
A 3d array of size p_2 \times p_3 \times n.
Currie, I. D., M. Durban, and P. H. C. Eilers (2006). Generalized linear array models with applications to multidimensional smoothing. Journal of the Royal Statistical Society. Series B. 68, 259-280.