Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Update Value-Iteration and Q-Learning to be in line with latest version of the script. #13

Merged
merged 3 commits into from
Jan 11, 2024
Merged

Update Value-Iteration and Q-Learning to be in line with latest version of the script. #13

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
1f96d4d
adapted value iteration and q learning to latest version of Russel an…
Dec 22, 2023
f69cc57
ran black format
Jan 10, 2024
ebd6e8b
added missing changes to tests/test_mdp_lecture_tex.py
Jan 10, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 6 additions & 6 deletions src/behavior_generation_lecture_python/mdp/mdp.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -276,7 +276,7 @@ def expected_utility_of_action(
Expected utility
"""
return sum(
p * utility_of_states[next_state]
p * (mdp.get_reward(next_state) + utility_of_states[next_state])
for (p, next_state) in mdp.get_transitions_with_probabilities(
state=state, action=action
)
Expand Down Expand Up @@ -327,12 +327,12 @@ def value_iteration(
history of utility estimates as list, if return_history is true.
"""
utility = {state: 0 for state in mdp.get_states()}
utility_history = []
utility_history = [utility.copy()]
for _ in range(max_iterations):
utility_old = utility.copy()
max_delta = 0
for state in mdp.get_states():
utility[state] = mdp.get_reward(state) + max(
utility[state] = max(
expected_utility_of_action(
mdp, state=state, action=action, utility_of_states=utility_old
)
Expand Down Expand Up @@ -426,8 +426,8 @@ def q_learning(
q_table = {}
for state in mdp.get_states():
for action in mdp.get_actions(state):
q_table[(state, action)] = mdp.get_reward(state)
q_table_history = []
q_table[(state, action)] = 0.0
q_table_history = [q_table.copy()]
state = mdp.initial_state

np.random.seed(1337)
Expand Down Expand Up @@ -455,7 +455,7 @@ def q_learning(
)
q_table[(state, chosen_action)] = (1 - alpha) * q_table[
(state, chosen_action)
] + alpha * (mdp.get_reward(state) + greedy_value_estimate_next_state)
] + alpha * (mdp.get_reward(next_state) + greedy_value_estimate_next_state)

if return_history:
q_table_history.append(q_table.copy())
Expand Down
62 changes: 39 additions & 23 deletions tests/test_mdp.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,7 +31,7 @@ def test_init_grid_mdp():

def test_expected_utility():
mdp = MDP(**SIMPLE_MDP_DICT)
assert 0.8 * 1 + 0.2 * 0.01 == expected_utility_of_action(
assert 0.8 * (-0.5 + 1) + 0.2 * (-0.1 + 0.01) == expected_utility_of_action(
mdp=mdp, state=1, action="A", utility_of_states={1: 0.01, 2: 1}
)

Expand All @@ -44,19 +44,19 @@ def test_derive_policy():

def test_value_iteration():
grid_mdp = GridMDP(**GRID_MDP_DICT)
epsilon = 0.001
epsilon = 0.0005
true_utility = {
(0, 0): 0.705,
(0, 1): 0.762,
(0, 2): 0.812,
(1, 0): 0.655,
(1, 2): 0.868,
(2, 0): 0.611,
(2, 1): 0.660,
(2, 2): 0.918,
(3, 0): 0.388,
(3, 1): -1.0,
(3, 2): 1.0,
(0, 0): 0.745,
(0, 1): 0.802,
(0, 2): 0.852,
(1, 0): 0.695,
(1, 2): 0.908,
(2, 0): 0.651,
(2, 1): 0.700,
(2, 2): 0.958,
(3, 0): 0.428,
(3, 1): 0,
(3, 2): 0,
}

computed_utility = value_iteration(mdp=grid_mdp, epsilon=epsilon, max_iterations=30)
Expand All @@ -66,32 +66,45 @@ def test_value_iteration():

def test_value_iteration_history():
grid_mdp = GridMDP(**GRID_MDP_DICT)
epsilon = 0.001
epsilon = 0.0005
true_utility_0 = {
(0, 0): 0,
(0, 1): 0,
(0, 2): 0,
(1, 0): 0,
(1, 2): 0,
(2, 0): 0,
(2, 1): 0,
(2, 2): 0,
(3, 0): 0,
(3, 1): 0,
(3, 2): 0,
}
true_utility_1 = {
(0, 0): -0.04,
(0, 1): -0.04,
(0, 2): -0.04,
(1, 0): -0.04,
(1, 2): -0.04,
(2, 0): -0.04,
(2, 1): -0.04,
(2, 2): -0.04,
(2, 2): 0.792,
(3, 0): -0.04,
(3, 1): -1.0,
(3, 2): 1.0,
(3, 1): 0,
(3, 2): 0,
}
true_utility_1 = {
true_utility_2 = {
(0, 0): -0.08,
(0, 1): -0.08,
(0, 2): -0.08,
(1, 0): -0.08,
(1, 2): -0.08,
(1, 2): 0.586,
(2, 0): -0.08,
(2, 1): -0.08,
(2, 2): 0.752,
(2, 1): 0.494,
(2, 2): 0.867,
(3, 0): -0.08,
(3, 1): -1.0,
(3, 2): 1.0,
(3, 1): 0,
(3, 2): 0,
}
computed_utility_history = value_iteration(
mdp=grid_mdp, epsilon=epsilon, max_iterations=30, return_history=True
Expand All @@ -102,6 +115,9 @@ def test_value_iteration_history():
for state in true_utility_1.keys():
assert abs(true_utility_1[state] - computed_utility_history[1][state]) < epsilon

for state in true_utility_2.keys():
assert abs(true_utility_2[state] - computed_utility_history[2][state]) < epsilon


def test_best_action_from_q_table():
q_table = {("A", 1): 0.5, ("A", 2): 0.6, ("B", 1): 0.7, ("B", 2): 0.8}
Expand Down
132 changes: 61 additions & 71 deletions tests/test_mdp_lecture_tex.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -93,17 +93,15 @@ def best_policy_to_tex_arrows(mdp: MDP, utility: dict):
\node at (3.5, 1.5) {$-1$};
\node at (3.5, 2.5) {$1$};"""

GRID_MDP_TRUE_UTILITY_TEX = r"""\node at (0.5, 0.5) {$0.705$};
\node at (0.5, 1.5) {$0.762$};
\node at (0.5, 2.5) {$0.812$};
\node at (1.5, 0.5) {$0.655$};
\node at (1.5, 2.5) {$0.868$};
\node at (2.5, 0.5) {$0.611$};
\node at (2.5, 1.5) {$0.66$};
\node at (2.5, 2.5) {$0.918$};
\node at (3.5, 0.5) {$0.388$};
\node at (3.5, 1.5) {$-1$};
\node at (3.5, 2.5) {$1$};"""
GRID_MDP_TRUE_UTILITY_TEX = r"""\node at (0.5, 0.5) {$0.745$};
\node at (0.5, 1.5) {$0.802$};
\node at (0.5, 2.5) {$0.852$};
\node at (1.5, 0.5) {$0.695$};
\node at (1.5, 2.5) {$0.908$};
\node at (2.5, 0.5) {$0.651$};
\node at (2.5, 1.5) {$0.7$};
\node at (2.5, 2.5) {$0.958$};
\node at (3.5, 0.5) {$0.428$};"""


def test_latex_value_over_time():
Expand Down Expand Up @@ -228,36 +226,32 @@ def test_latex_policy_as_arrows():
HIGHWAY_MDP_UTILITY_TEX = r"""\node at (0.5, 1.5) {$-18.6$};
\node at (0.5, 2.5) {$-16.6$};
\node at (0.5, 3.5) {$-15.8$};
\node at (1.5, 1.5) {$-19.8$};
\node at (1.5, 2.5) {$-17.4$};
\node at (1.5, 3.5) {$-15.8$};
\node at (2.5, 1.5) {$-17.7$};
\node at (2.5, 2.5) {$-16$};
\node at (2.5, 3.5) {$-14.8$};
\node at (3.5, 1.5) {$-15.3$};
\node at (3.5, 2.5) {$-14.1$};
\node at (3.5, 3.5) {$-13.8$};
\node at (4.5, 1.5) {$-12.5$};
\node at (4.5, 2.5) {$-12.1$};
\node at (4.5, 3.5) {$-15$};
\node at (5.5, 1.5) {$-9.52$};
\node at (5.5, 2.5) {$-11.8$};
\node at (5.5, 3.5) {$-20.7$};
\node at (6.5, 0.5) {$-6$};
\node at (6.5, 1.5) {$-8.09$};
\node at (6.5, 2.5) {$-14.9$};
\node at (6.5, 3.5) {$-34$};
\node at (7.5, 0.5) {$-4$};
\node at (7.5, 1.5) {$-8.38$};
\node at (7.5, 2.5) {$-26.6$};
\node at (7.5, 3.5) {$-52$};
\node at (8.5, 0.5) {$-2$};
\node at (8.5, 1.5) {$-15.5$};
\node at (8.5, 2.5) {$-52$};
\node at (8.5, 3.5) {$-51$};
\node at (9.5, 1.5) {$-50$};
\node at (9.5, 2.5) {$-50$};
\node at (9.5, 3.5) {$-50$};"""
\node at (1.5, 1.5) {$-16.8$};
\node at (1.5, 2.5) {$-15.4$};
\node at (1.5, 3.5) {$-14.8$};
\node at (2.5, 1.5) {$-14.7$};
\node at (2.5, 2.5) {$-14$};
\node at (2.5, 3.5) {$-13.8$};
\node at (3.5, 1.5) {$-12.3$};
\node at (3.5, 2.5) {$-12.1$};
\node at (3.5, 3.5) {$-12.8$};
\node at (4.5, 1.5) {$-9.52$};
\node at (4.5, 2.5) {$-10.1$};
\node at (4.5, 3.5) {$-14$};
\node at (5.5, 1.5) {$-6.52$};
\node at (5.5, 2.5) {$-9.8$};
\node at (5.5, 3.5) {$-19.7$};
\node at (6.5, 0.5) {$-4$};
\node at (6.5, 1.5) {$-5.09$};
\node at (6.5, 2.5) {$-12.9$};
\node at (6.5, 3.5) {$-33$};
\node at (7.5, 0.5) {$-2$};
\node at (7.5, 1.5) {$-5.38$};
\node at (7.5, 2.5) {$-24.6$};
\node at (7.5, 3.5) {$-51$};
\node at (8.5, 1.5) {$-12.5$};
\node at (8.5, 2.5) {$-50$};
\node at (8.5, 3.5) {$-50$};"""

HIGHWAY_MDP_OPTIMAL_POLICY_ARROWS_TEX = r"""\draw[->,color=blue] (0.3, 1.3) -- (0.7, 1.7);
\draw[->,color=blue] (0.3, 2.3) -- (0.7, 2.7);
Expand Down Expand Up @@ -293,36 +287,32 @@ def test_latex_policy_as_arrows():
HIGHWAY_MDP_LC_R_0DOT4_UTILITY_TEX = r"""\node at (0.5, 1.5) {$-28.6$};
\node at (0.5, 2.5) {$-28.6$};
\node at (0.5, 3.5) {$-31$};
\node at (1.5, 1.5) {$-28.7$};
\node at (1.5, 2.5) {$-28.6$};
\node at (1.5, 3.5) {$-32.6$};
\node at (2.5, 1.5) {$-25.7$};
\node at (2.5, 2.5) {$-27.2$};
\node at (2.5, 3.5) {$-34.5$};
\node at (3.5, 1.5) {$-22.7$};
\node at (3.5, 2.5) {$-27$};
\node at (3.5, 3.5) {$-37.9$};
\node at (4.5, 1.5) {$-19.7$};
\node at (4.5, 2.5) {$-28.5$};
\node at (4.5, 3.5) {$-42.5$};
\node at (5.5, 1.5) {$-16.7$};
\node at (5.5, 2.5) {$-33.1$};
\node at (5.5, 3.5) {$-47.2$};
\node at (6.5, 0.5) {$-6$};
\node at (6.5, 1.5) {$-18.8$};
\node at (6.5, 2.5) {$-39.3$};
\node at (6.5, 3.5) {$-50.8$};
\node at (7.5, 0.5) {$-4$};
\node at (7.5, 1.5) {$-23.6$};
\node at (7.5, 2.5) {$-46.4$};
\node at (7.5, 3.5) {$-52$};
\node at (8.5, 0.5) {$-2$};
\node at (8.5, 1.5) {$-33$};
\node at (8.5, 2.5) {$-52$};
\node at (8.5, 3.5) {$-51$};
\node at (9.5, 1.5) {$-50$};
\node at (9.5, 2.5) {$-50$};
\node at (9.5, 3.5) {$-50$};"""
\node at (1.5, 1.5) {$-25.7$};
\node at (1.5, 2.5) {$-26.6$};
\node at (1.5, 3.5) {$-31.6$};
\node at (2.5, 1.5) {$-22.7$};
\node at (2.5, 2.5) {$-25.2$};
\node at (2.5, 3.5) {$-33.5$};
\node at (3.5, 1.5) {$-19.7$};
\node at (3.5, 2.5) {$-25$};
\node at (3.5, 3.5) {$-36.9$};
\node at (4.5, 1.5) {$-16.7$};
\node at (4.5, 2.5) {$-26.5$};
\node at (4.5, 3.5) {$-41.5$};
\node at (5.5, 1.5) {$-13.7$};
\node at (5.5, 2.5) {$-31.1$};
\node at (5.5, 3.5) {$-46.2$};
\node at (6.5, 0.5) {$-4$};
\node at (6.5, 1.5) {$-15.8$};
\node at (6.5, 2.5) {$-37.3$};
\node at (6.5, 3.5) {$-49.8$};
\node at (7.5, 0.5) {$-2$};
\node at (7.5, 1.5) {$-20.6$};
\node at (7.5, 2.5) {$-44.4$};
\node at (7.5, 3.5) {$-51$};
\node at (8.5, 1.5) {$-30$};
\node at (8.5, 2.5) {$-50$};
\node at (8.5, 3.5) {$-50$};"""

HIGHWAY_MDP_LC_R_0DOT4_OPTIMAL_POLICY_ARROWS_TEX = r"""\draw[->,color=blue] (0.3, 1.3) -- (0.7, 1.7);
\draw[->,color=blue] (0.3, 2.5) -- (0.7, 2.5);
Expand Down
Loading