Skip to contents

A series with an sd of zero has no innovation, so its rows of \((I - B)(x - \mu) = \varepsilon\) read zero. Writing \(u\) for its cells (unobs_idx) and \(o\) for the rest (obs_idx), those rows give every one of those cells from the others, $$x_u - \mu_u = -(I - B)_{uu}^{-1}(I - B)_{uo}(x_o - \mu_o),$$ and this returns the matrix in the middle.

Usage

get_dsem_solve_mat(IminusB, dsem_cells)

Arguments

IminusB

Sparse \(I - B\), from get_dsem_matrices.

dsem_cells

Output of get_dsem_cells.

Value

Dense matrix, one row per solved cell and one column per cell that keeps an innovation, or NULL when nothing is solved out.