R/pm.R

In Tenenhaus/RGCCA: Regularized and Sparse Generalized Canonical Correlation Analysis for Multiblock Data

#' Product for Matrix (pm) is a generalization of the matrix product %*%
#' for matrices with missing data. Missing data are replaced by 0.
#' @param M1  A matrix with n1 lines and p columns
#' @param M2  A matrix with p lines and n2 columns
#' @param na.rm if TRUE calculates the matrix product only on available data.
#' Else returns NA.
#' @return The resulting matrix with n1 lines and n2 columns
#' @title Product for Matrices with missing data (pm)
#' @noRd
pm <- function(M1, M2, na.rm = TRUE) {
  if (na.rm) {
    M1[is.na(M1)] <- 0
    M2[is.na(M2)] <- 0
  }
  return(M1 %*% M2)
}