tests/testthat/test-compute_functions.R

In abichat/zazou: Z-Scores As Ornstein-Uhlenbeck

context("Compute objective function")

n <- 3
m <- 4

Y <- 10*1:m
X <- matrix(1:(m*(n+m)), nrow = m, ncol = n+m)
lambda <- 5
Delta <- -(1:(n+m))

t(Y - X %*% Delta) %*% (Y - X %*% Delta) - lambda * sum(Delta)


test_that("objective function is correct", {
  obj <- compute_objective_function(Y, X, lambda, type = "lasso")
  value <- obj(Delta)
  expect_is(obj, "function")
  expect_length(value, 1)
  expect_equal(value,
               (1/2 * (t(Y - X %*% Delta) %*% (Y - X %*% Delta)) -
                  lambda * sum(Delta))[1, 1])
})


test_that("gradient function is correct", {
  grad <- compute_gradient_function(Y, X, lambda)
  vect <- grad(Delta)
  expect_is(grad, "function")
  expect_length(vect, length(Delta))
  expect_equal(vect, -t(X) %*% Y + t(X) %*% X %*% Delta - lambda)
})