import unittest
import importlib
import re
import types
import gymnasium as gym
import numpy as np

# Helper function to require a successful test
def _require_success(test_case, condition, failure_message):
  if not condition:
    test_case._testMethodName = failure_message  # Rename the test with a failure description
    test_case.fail()  # Fail the test with a failure message

# Helper function to check if test is successful
def _check_success(test_case, condition, success_message, failure_message):
  if condition:
    test_case._testMethodName = success_message  # Rename the test for better readability in test reports
  else:
    test_case._testMethodName = failure_message  # Rename the test with a failure description
    test_case.fail()  # Fail the test with a failure message

class TestUserCode(unittest.TestCase):
  # Test that gymnasium is imported
  def test_gymnasium_imported(self):
    import user_code
    _require_success(
      self,
      hasattr(user_code, "gym"),
      "`gym` is not declared"
    )
    _check_success(
      self,
      user_code.gym == gym,
      "`gymnasium` is imported",
      "`gym` is declared, but it's not `gymnasium` library"
    )
    
  # Test that environment is created
  def test_env_created(self):
    import user_code
    _require_success(
      self,
      hasattr(user_code, "env"),
      "`env` is not declared"
    )
    _check_success(
      self,
      isinstance(user_code.env, gym.Env),
      "`env` is created",
      "`env` is declared, but it's not a gymnasium environment"
    )

  # Test that environment was reset
  def test_env_reset(self):
    with open("user_code.py", "r") as f:
      user_code_text = f.read()
      _check_success(
        self,
        re.search(r"observation *, *info *= *env *\. *reset *\( *\)", user_code_text),
        "`reset` was called",
        "`reset` was not called"
      )
    
  # Test that step was performed, and values were saved
  def test_step_performed(self):
    import user_code
    _require_success(
      self,
      hasattr(user_code, "observation"),
      "`observation` is not declared"
    )
    _require_success(
      self,
      isinstance(user_code.observation, np.ndarray),
      "`observation` is not saved after step"
    )
    _require_success(
      self,
      hasattr(user_code, "reward"),
      "`reward` is not declared"
    )
    _require_success(
      self,
      isinstance(user_code.reward, float),
      "`reward` is not saved after step"
    )
    _require_success(
      self,
      hasattr(user_code, "terminated"),
      "`terminated` is not declared"
    )
    _require_success(
      self,
      isinstance(user_code.terminated, bool),
      "`terminated` is not saved after step"
    )
    _require_success(
      self,
      hasattr(user_code, "truncated"),
      "`truncated` is not declared"
    )
    _require_success(
      self,
      isinstance(user_code.truncated, bool),
      "`truncated` is not saved after step"
    )
    _require_success(
      self,
      hasattr(user_code, "info"),
      "`info` is not declared"
    )
    _require_success(
      self,
      isinstance(user_code.info, dict),
      "`info` is not saved after step"
    )
    self._testMethodName = "`step` performed successfully"

  # Test that environment was closed
  def test_env_closed(self):
    with open("user_code.py", "r") as f:
      user_code_text = f.read()
      _check_success(
        self,
        re.search(r"env *\. *close *\( *\)", user_code_text),
        "`close` was called",
        "`close` was not called"
      )

# Run the tests when executing the script
if __name__ == '__main__':
  unittest.main()

test_main.py

Reinforcement Learning (RL) is a powerful branch of machine learning focused on training intelligent agents through interaction with their environment. In this course, you'll learn how agents gradually discover effective behaviors through trial and error. Beginning with core concepts like Markov decision processes and multi-armed bandits, you'll work your way through dynamic programming, Monte Carlo methods, and temporal difference learning.

Discover how to train agents to make optimal decisions through trial and error. Explore the essentials of reinforcement learning theory. Get hands-on experience setting up and running a Gymnasium environment.

Master the exploration-exploitation trade-off through the multi-armed bandit problem. Implement action-value estimation, ε-greedy, upper confidence bound, and gradient-bandit methods. Evaluate algorithms' performance on simulated reward-maximization tasks.

Master dynamic programming for model-based RL. Discover how Bellman equations can be used to evaluate and improve policies. Implement policy and value iteration algorithms. Explore generalized policy iteration as the theoretical foundation for model-free methods.

Master Monte Carlo methods for model-free RL. Estimate value functions and derive optimal policies from complete episodes. Implement on-policy and off-policy Monte Carlo control algorithms. Discover exploration strategies to optimize model-free learning.

Master temporal difference learning for model-free RL. Estimate value functions from partial episodes using TD(0) updates. Implement on-policy SARSA and off-policy Q-Learning algorithms. Discover how Monte Carlo methods and TD learning combine in n-step TD and TD(λ).

Challenge: Setting Up an Environment

Solution