tests/testthat/test_sugar.R

In mlr3mbo: Flexible Bayesian Optimization

test_that("SurrogateLearner sugar", {
  surrogate = srlrn(REGR_FEATURELESS, catch_errors = FALSE)
  expect_r6(surrogate, classes = "SurrogateLearner")
  expect_false(surrogate$param_set$values$catch_errors)
})

test_that("SurrogateLearner sugar with a single learner in a list", {
  surrogate = srlrn(list(REGR_FEATURELESS), catch_errors = FALSE)
  expect_r6(surrogate, classes = "SurrogateLearner")
  expect_r6(surrogate$learner, classes = "Learner")
  expect_false(surrogate$param_set$values$catch_errors)
})

test_that("SurrogateLearnerCollection sugar", {
  surrogate = srlrn(list(REGR_FEATURELESS, REGR_FEATURELESS$clone(deep = TRUE)), catch_errors = FALSE)
  expect_r6(surrogate, classes = "SurrogateLearnerCollection")
  expect_false(surrogate$param_set$values$catch_errors)
})

test_that("AcqFunction sugar", {
  acqfunction = acqf("cb", lambda = 3)
  expect_acqfunction(acqfunction)
  expect_equal(acqfunction$constants$values$lambda, 3)
})

test_that("AcqOptimizer sugar", {
  acqoptimizer = acqo(opt("random_search"), trm("evals"), catch_errors = FALSE)
  expect_r6(acqoptimizer, "AcqOptimizer")
  expect_false(acqoptimizer$param_set$values$catch_errors)
})

test_that("ResultAssigner sugar", {
  resultassigner = ras("surrogate")
  expect_r6(resultassigner, "ResultAssigner")
})

test_that("InputTrafo sugar", {
  inputtrafo = it("unitcube")
  expect_r6(inputtrafo, "InputTrafo")
})

test_that("OutputTrafo sugar", {
  outputtrafo = ot("log")
  expect_r6(outputtrafo, "OutputTrafo")
})

test_that("acqo errors when terminator or callbacks are combined with a dictionary key", {
  expect_error(acqo("random_search", terminator = trm("evals")), "must not be given")
  expect_error(acqo("random_search", callbacks = list()), "must not be given")
})

test_that("srlrn replicates a single learner for multiple targets", {
  surrogate = srlrn(lrn("regr.featureless"), cols_y = c("y1", "y2"))
  expect_r6(surrogate, "SurrogateLearnerCollection")
  expect_length(surrogate$learner, 2L)
  expect_false(address(surrogate$learner[[1L]]) == address(surrogate$learner[[2L]]))

  inst = MAKE_INST(OBJ_1D_2, search_space = PS_1D, terminator = trm("evals", n_evals = 5L))
  surrogate = srlrn(lrn("regr.featureless"), archive = inst$archive)
  expect_r6(surrogate, "SurrogateLearnerCollection")
  expect_length(surrogate$learner, 2L)
})