tests/testthat/test-simulate.R

In morphErr: Measurement Error Models for Morphometric Data

# sim.measurements() tests

test_that("sim.measurements handles inputs correctly", {
  # Test 1: basic functionality with equal photos
  mus <- c(315, 150, 100)
  sigmas <- c(25, 15, 10)
  rhos <- c(0.85, 0.80, 0.75)
  psis <- c(10, 6, 4)
  phis <- c(0.5, 0.4, 0.3)

  # 3 animals, 2 photos each:
  data1 <- sim.measurements(n.animals = 3, n.photos = 2, mus = mus,
                            sigmas = sigmas, rhos = rhos, psis = psis,
                            phis = phis)

  # Basic structure checkss
  expect_s3_class(data1, "data.frame")
  expect_named(data1, c("animal.id", "photo.id", "dim", "measurement"))
  expect_equal(nrow(data1), 3 * 2 * 3)  # n_animals * n_photos * n_dims

  # Test 2: dif numbers of photos per animal
  n_photos <- c(2, 3, 1)
  data2 <- sim.measurements(n.animals = 3, n.photos = n_photos, mus = mus,
                            sigmas = sigmas, rhos = rhos, psis = psis,
                            phis = phis)
  expect_equal(nrow(data2), sum(n_photos) * 3)  # total_photos * n_dims

  # Test 3: error handling
  expect_error(
      sim.measurements(n.animals = 2, n.photos = c(2, 2, 2), mus = mus,
                       sigmas = sigmas, rhos = rhos, psis = psis, phis = phis),
    "length of 'n.photos' should be equal to 'n.animals'"
  )
})

# -------------------------------------------------------------------------------------------------------

# sim.morph() tests

test_that("sim.morph runs simulations correctly", {
  # Set up small simulation
  n.sims <- 2
  n.animals <- 5
  n.photos <- rep(3,5)
  mus <- c(315, 150, 100)
  sigmas <- c(25, 15, 10)
  rhos <- c(0.85, 0.80, 0.75)
  psis <- c(10, 6, 4)
  phis <- c(0.5, 0.4, 0.3)

  # Run simulation
  set.seed(123)
  result <- sim.morph(
    n.sims = n.sims,
    n.animals = n.animals,
    n.photos = n.photos,
    mus = mus,
    sigmas = sigmas,
    rhos = rhos,
    psis = psis,
    phis = phis,
    progressbar = FALSE
  )

  # check structure
  expect_s3_class(result, "lme.morph.sim")
  expect_named(result, c("settings", "fits"))
  expect_equal(length(result$fits), n.sims)

  # Check settings are stored correctly
  expect_equal(result$settings$n.sims, n.sims)
  expect_equal(result$settings$mus, mus)

  # Check model fits
  expect_s3_class(result$fits[[1]], "lme.morph")

  # Test error handling
  expect_error(
    sim.morph(n.sims = 2, n.animals = 3, n.photos = 2,
              mus = c(315, 150, 100), sigmas =  c(25, 15), rhos = c(0.85, 0.80, 0.75),
              psis = c(10, 6, 4), phis = c(0.5, 0.4, 0.3)),
    "The 'sigmas' argument must have an element for each dimension"
  )
})

# -------------------------------------------------------------------------------------------------------

# extract.sim.morph() tests

test_that("extract.sim.morph works correctly", {
  # Create a small simulation study
  set.seed(123)
  sim_result <- sim.morph(
    n.sims = 2,
    n.animals = 5,
    n.photos = rep(3,5),
    mus = c(315, 150, 100),
    sigmas = c(25, 15, 10),
    rhos = c(0.85, 0.80, 0.75),
    psis = c(10, 6, 4),
    phis = c(0.5, 0.4, 0.3),
    progressbar = FALSE
  )

  # Test basic extraction
  extracted <- extract.sim.morph(sim_result)
  expect_true(is.array(extracted))
  # Should be at least 2D
  expect_true(length(dim(extracted)) >= 2)

  # Test custom function
  get_fixed <- function(fit) fit$coefficients$fixed
  fixed_effects <- extract.sim.morph(sim_result, FUN = get_fixed)
  expect_true(is.array(fixed_effects))
  # Number of simulations
  expect_equal(dim(fixed_effects)[2], 2)

  # Test eror handling
  expect_error(
    extract.sim.morph(list(a = 1)),
    "'sim.res' must be an object of class 'lme.morph.sim'"
  )
})