R/mlnormal_update_V_R.R

In alexanderrobitzsch/LAM: Some Latent Variable Models

## File Name: mlnormal_update_V_R.R
## File Version: 0.28

##############################################
# update matrix V and its inverse
mlnormal_update_V_R <- function( Z_index, G, theta,
        Z_list, use_ginverse, variance_shortcut, freq_id,
        do_compute, rcpp_args){
    dimZ <- dim( Z_index )
    Z2 <- dimZ[2]
    V_list <- as.list(1:G)
    V1_list <- V_list
    dimZ <- dim( Z_index )
    Z2 <- dimZ[2]
    do_computation <- TRUE

    for (gg in 1:G){
        # gg <- 1
        # compute V for group gg
        if ( do_compute[gg] ){
            Z_index_gg <- Z_index[gg,,,drop=FALSE]
            Z_list_gg <- Z_list[[gg]]
            V_gg <- 0*Z_list_gg[[1]]
            for (pp in 1:Z2){
                # pp <- 1
                # theta^q
                a1 <- prod( theta^( Z_index_gg[1,pp,] ) )
                V_gg <- V_gg + a1 * Z_list_gg[[pp]]
            }
            ## use generalized inverse instead of inverse if
            ## solve does not work in case of singularity
            if ( ! use_ginverse ){
                V_gg1 <- solve( V_gg )
            } else {
                V_gg1 <- sirt::ginverse_sym( V_gg )
            }
        }  #     end do computation
        V_list[[gg]] <- V_gg
        V1_list[[gg]] <- V_gg1
    }
    #--- output
    res <- list("V_list"=V_list, "V1_list"=V1_list,
                    "rcpp_args"=rcpp_args )
    return(res)
}
######################################################################