if (require(RUnit)) {
  library(Matrix)
  library(FeatureHashing)
  contr <- list(
    Plant = contrasts(CO2$Plant, contrasts = FALSE),
    Type = contrasts(CO2$Type, contrasts = FALSE),
    Treatment = contrasts(CO2$Treatment, contrasts = FALSE)
    )
  m1 <- sparse.model.matrix(~ ., CO2, contr, transpose = TRUE)
  
  v <- numeric(dim(m1)[2])
  r1 <- as.vector(m1 %*% v)
  r2 <- FeatureHashing:::Xv(m1, v, numeric(dim(m1)[1]))
  checkTrue(isTRUE(all.equal(r1, r2)),
            "The matrix-vector multiplication is incorrect")
  
  for(i in 1:ncol(m1)) {
    v <- numeric(dim(m1)[2])
    v[i] <- 1
    r1 <- as.vector(m1 %*% v)
    r2 <- numeric(dim(m1)[1])
    FeatureHashing:::Xv(m1, v, r2)
    checkTrue(isTRUE(all.equal(r1, r2)),
              "The result of C++ Xv is incorrect")
    r <- as.vector(m1[,i])
  }
  
  
  v <- numeric(dim(m1)[1])
  r1 <- as.vector(v %*% m1)
  r2 <- FeatureHashing:::vX(v, m1, numeric(dim(m1)[2]))
  checkTrue(isTRUE(all.equal(r1, r2)),
            "The result of C++ vX is incorrect")
  
  for(i in 1:nrow(m1)) {
    v <- numeric(dim(m1)[1])
    v[i] <- 1
    r1 <- as.vector(v %*% m1)
    r2 <- numeric(dim(m1)[2])
    FeatureHashing:::vX(v, m1, r2)
    checkTrue(isTRUE(all.equal(r1, r2)),
              "The result of matrix-vector multiplication is incorrect")
  }
}