test_that("dense vs bigmem parity is tight under deterministic settings", {
  skip_on_cran()
  skip_on_ci() 
  set.seed(123)
  
  n <- 200; p <- 40
  X <- matrix(rnorm(n * p), n, p)
  y <- 2 * X[, 1] - 0.5 * X[, 2] + rnorm(n, sd = 0.1)
  
  fit_dense <- pls_fit(X, y, ncomp = 3, tol = 1e-10,
                       backend = "arma", scores = "r")
  
  skip_if_not_installed("bigmemory")
  bmX <- bigmemory::as.big.matrix(X)
  bmy <- bigmemory::as.big.matrix(matrix(y, n, 1))
  
  fit_big <- pls_fit(bmX, bmy, ncomp = 3, tol = 1e-10,
                     backend = "bigmem", scores = "none")
  
  expect_equal(as.numeric(fit_dense$coefficients),
               as.numeric(fit_big$coefficients), tolerance = 1e-7)
  expect_equal(as.numeric(fit_dense$intercept),
               as.numeric(fit_big$intercept),   tolerance = 1e-7)
})

test_that("file-backed scores sink matches dense scores on small data", {
  skip_on_cran()
  skip_on_ci() 
  skip_if_not_installed("bigmemory")
  set.seed(321)
  
  n <- 120; p <- 25; k <- 2
  X <- matrix(rnorm(n * p), n, p)
  y <- 0.8 * X[, 1] + rnorm(n, sd = 0.2)
  
  fit_dense <- pls_fit(X, y, ncomp = k, backend = "arma", scores = "r")
  
  bmX <- bigmemory::as.big.matrix(X)
  bmy <- bigmemory::as.big.matrix(matrix(y, n, 1))
  
  tmp <- tempdir()
  if(file.exists(paste(tmp,"scores.desc",sep="/"))){unlink(paste(tmp,"scores.desc",sep="/"))}
  if(file.exists(paste(tmp,"scores.bin",sep="/"))){unlink(paste(tmp,"scores.bin",sep="/"))}

  sink_bm <- bigmemory::filebacked.big.matrix(
    nrow = n, ncol = k, type = "double",
    backingfile = "scores.bin", backingpath = tmp,
    descriptorfile = "scores.desc"
  )
  
  fit_big <- pls_fit(bmX, bmy, ncomp = k,
                     backend = "bigmem", scores = "big",
                     scores_target = "existing", scores_bm = sink_bm)
  
  expect_true(inherits(fit_big$scores, "big.matrix"))
  expect_equal(nrow(fit_big$scores), n)
  expect_equal(ncol(fit_big$scores), k)
  
  scores_from_file <- as.matrix(fit_big$scores[])
  expect_equal(scores_from_file, fit_dense$scores, tolerance = 1e-6)
})