markusdumke/reinforcelearn

Test for Sarsa fails

markusdumke opened this issue · 0 comments

test_that("SARSA(0) converges to correct policy for gridworld", {
  Q = sarsa(grid, n.episodes = 1000)
  policy = make_greedy_policy(Q)
  expect_equal(policy[test_states, ], optimal.policy[test_states, ])
})