R/sparse_set_operations.R

In neuroimaginador/fcaR: Formal Concept Analysis

.difference <- function(A, B) {

  if (is.numeric(A)) A <- Matrix::Matrix(A, sparse = TRUE)
  if (is.numeric(B)) B <- Matrix::Matrix(B, sparse = TRUE)

  applicable <- (ncol(A) == ncol(B)) || (ncol(B) == 1) || (ncol(A) == 1)
  stopifnot(applicable)

  if (ncol(A) == ncol(B)) {

    A[B >= A] <- 0

    return(A)

  }

  if (ncol(B) == 1) {

    n <- ncol(A)

    newB <- .replicate_col(B, n)

    A[newB >= A] <- 0

    return(A)

  }

  if (ncol(A) == 1) {

    n <- ncol(B)

    newA <- .replicate_col(A, n)

    newA[B >= newA] <- 0

    return(newA)

  }

}

.union <- function(A, B) {

  if (is.numeric(A)) A <- Matrix::Matrix(A, sparse = TRUE)
  if (is.numeric(B)) B <- Matrix::Matrix(B, sparse = TRUE)

  applicable <- (ncol(A) == ncol(B)) || (ncol(B) == 1) || (ncol(A) == 1)
  stopifnot(applicable)

  if (ncol(A) == ncol(B)) {

    idx <- Matrix::which(B > A)
    A[idx] <- B[idx]

    return(A)

  }

  if (ncol(B) == 1) {

    n <- ncol(A)

    newB <- .replicate_col(B, n)

    idx <- Matrix::which(newB > A)
    A[idx] <- newB[idx]

    return(A)

  }

  if (ncol(A) == 1) {

    n <- ncol(B)

    newA <- .replicate_col(A, n)

    idx <- Matrix::which(B > newA)
    newA[idx] <- B[idx]

    return(newA)

  }

}

.multiunion <- function(M) {

  v <- flatten_sparse_C(M@p, M@i, M@x, M@Dim)

  return(Matrix::Matrix(v, ncol = 1, sparse = TRUE))

}

.replicate_col <- function(A, n) {

  new_i <- rep(A@i, n)

  stopifnot("x" %in% methods::slotNames(A))

  new_p <- c(0, A@p[2] * seq(n))

  newA <- Matrix::sparseMatrix(i = new_i + 1,
                               p = new_p,
                               x = A@x,
                               dims = c(nrow(A), n))

  return(newA)

}