!pip install aicrowd-cli > /dev/null

import os

AICROWD_DATASET_PATH = os.getenv("DATASET_PATH", os.getcwd()+"/a5562c7d-55f0-4d06-841c-110655bb04ec_a2_gridworld_inputs.zip")
AICROWD_RESULTS_DIR = os.getenv("OUTPUTS_DIR", "results")

!unzip -q $AICROWD_DATASET_PATH

DATASET_DIR = 'inputs/'

import numpy as np

class GridEnv_HW2:
    def __init__(self, 
                 goal_location, 
                 action_stochasticity,
                 non_terminal_reward,
                 terminal_reward,
                 grey_in,
                 brown_in,
                 grey_out,
                 brown_out
                ):

        # Do not edit this section 
        self.action_stochasticity = action_stochasticity
        self.non_terminal_reward = non_terminal_reward
        self.terminal_reward = terminal_reward
        self.grid_size = [10, 10]

        # Index of the actions 
        self.actions = {'N': (1, 0), 
                        'E': (0,1),
                        'S': (-1,0), 
                        'W': (0,-1)}
        
        self.perpendicular_order = ['N', 'E', 'S', 'W']
        
        l = ['normal' for _ in range(self.grid_size[0]) ]
        self.grid = np.array([l for _ in range(self.grid_size[1]) ], dtype=object)

        self.grid[goal_location[0], goal_location[1]] = 'goal'
        self.goal_location = goal_location

        for gi in grey_in:
            self.grid[gi[0],gi[1]] = 'grey_in'
        for bi in brown_in:    
            self.grid[bi[0], bi[1]] = 'brown_in'

        for go in grey_out:    
            self.grid[go[0], go[1]] = 'grey_out'
        for bo in brown_out:    
            self.grid[bo[0], bo[1]] = 'brown_out'

        self.grey_outs = grey_out
        self.brown_outs = brown_out

    def _out_of_grid(self, state):
        if state[0] < 0 or state[1] < 0:
            return True
        elif state[0] > self.grid_size[0] - 1:
            return True
        elif state[1] > self.grid_size[1] - 1:
            return True
        else:
            return False

    def _grid_state(self, state):
        return self.grid[state[0], state[1]]        
        
    def get_transition_probabilites_and_reward(self, state, action):
        """ 
        Returns the probabiltity of all possible transitions for the given action in the form:
        A list of tuples of (next_state, probability, reward)
        Note that based on number of state and action there can be many different next states
        Unless the state is All the probabilities of next states should add up to 1
        """

        grid_state = self._grid_state(state)
        
        if grid_state == 'goal':
            return [(self.goal_location, 1.0, 0.0)]
        elif grid_state == 'grey_in':
            npr = []
            for go in self.grey_outs:
                npr.append((go, 1/len(self.grey_outs), 
                            self.non_terminal_reward))
            return npr
        elif grid_state == 'brown_in':
            npr = []
            for bo in self.brown_outs:
                npr.append((bo, 1/len(self.brown_outs), 
                            self.non_terminal_reward))
            return npr
        
        direction = self.actions.get(action, None)
        if direction is None:
            raise ValueError("Invalid action %s , please select among" % action, list(self.actions.keys()))

        dir_index = self.perpendicular_order.index(action)
        wrap_acts = self.perpendicular_order[dir_index:] + self.perpendicular_order[:dir_index]
        next_state_probs = {}
        for prob, a in zip(self.action_stochasticity, wrap_acts):
            d = self.actions[a]
            next_state = (state[0] + d[0]), (state[1] + d[1])
            if self._out_of_grid(next_state):
                next_state = state
            next_state_probs.setdefault(next_state, 0.0)
            next_state_probs[next_state] += prob

        npr = []
        for ns, prob in next_state_probs.items():
            next_grid_state = self._grid_state(ns)
            reward = self.terminal_reward if next_grid_state == 'goal' else self.non_terminal_reward
            npr.append((ns, prob, reward))

        return npr

    def step(self, state, action):
        npr = self.get_transition_probabilites_and_reward(state, action)
        probs = [t[1] for t in npr]
        sampled_idx = np.random.choice(range(len(npr)), p=probs)
        sampled_npr = npr[sampled_idx]
        next_state = sampled_npr[0]
        reward = sampled_npr[2]
        is_terminal = next_state == tuple(self.goal_location)
        return next_state, reward, is_terminal

def get_base_kwargs():
    goal_location = (9,9)
    action_stochasticity = [0.8, 0.2/3, 0.2/3, 0.2/3]
    grey_out = [(3,2), (4,2), (5,2), (6,2)]
    brown_in = [(9,7)]
    grey_in = [(0,0)]
    brown_out = [(1,7)]
    non_terminal_reward = 0
    terminal_reward = 10

    base_kwargs =  {"goal_location": goal_location, 
            "action_stochasticity": action_stochasticity,
            "brown_in": brown_in, 
            "grey_in": grey_in, 
            "brown_out": brown_out,
            "non_terminal_reward": non_terminal_reward,
            "terminal_reward": terminal_reward,
            "grey_out": grey_out,}
    
    return base_kwargs

base_kwargs = get_base_kwargs()

def value_iteration(env, gamma):
    # Initial Values
    values = np.zeros((10, 10))

    # Initial policy
    policy = np.empty((10, 10), object)
    policy[:] = 'N' # Make all the policy values as 'N'

    # Begin code here
    value_history = [values.copy()]

    states = [(i,j) for i in range(env.grid_size[0]) for j in range(env.grid_size[1])]

    epsilon = 1e-8
    delta = epsilon
    while delta >= epsilon:
      delta = 0
      H = np.zeros((10, 10)) - float('inf')
      for s in states:
        for action in env.perpendicular_order:
          npr = env.get_transition_probabilites_and_reward(s, action)
          c = 0
          for next_state, probability, reward in npr:
            c += probability * (reward + gamma*values[next_state])
          H[s] = max(H[s], c)
        delta = max(delta, abs(values[s] - H[s]))

      values = H.copy()
      value_history.append(values.copy())

    for s in states:
      C = -float('inf')
      A = env.perpendicular_order[0]
      for action in env.perpendicular_order:
        npr = env.get_transition_probabilites_and_reward(s, action)
        c = 0
        for next_state, probability, reward in npr:
          c += probability * (reward + gamma*values[next_state])
          
        if c > C:
          C = c
          A = action
      policy[s] = A
    
    # Put your extra information needed for plots etc in this dictionary
    extra_info = {"All Values:": value_history}

    # End code

    # Do not change the number of output values
    return {"Values": values, "Policy": policy}, extra_info

env = GridEnv_HW2(**base_kwargs)
res, extra_info = value_iteration(env, 0.7)

 # The rounding off is just for making print statement cleaner
print(np.flipud(np.round(res['Values'], decimals=2)))
print(np.flipud(res['Policy']))

def policy_iteration(env, gamma):
    # Initial Values
    values = np.zeros((10, 10))

    # Initial policy
    policy = np.empty((10, 10), object)
    policy[:] = 'N' # Make all the policy values as 'N'

    # Begin code here  

    reduced_value_history = [values.copy()]
    value_history = [values.copy()]
    states = [(i,j) for i in range(env.grid_size[0]) for j in range(env.grid_size[1])] 

    epsilon = 1e-8
    done = False
    while not done:
      delta = epsilon
      while delta >= epsilon:
        delta = 0
        for s in states:
          j = values[s]
          npr = env.get_transition_probabilites_and_reward(s, policy[s])
          c = 0
          for next_state, probability, reward in npr:
            c += probability * (reward + gamma*values[next_state])
          values[s] = c
          delta = max(delta, abs(j - values[s]))
        value_history.append(values.copy())
      reduced_value_history.append(values.copy())

      done = True

      for s in states:
        b = policy[s]
        
        C = -float('inf')
        A = env.perpendicular_order[0]
        for action in env.perpendicular_order:
          npr = env.get_transition_probabilites_and_reward(s, action)
          c = 0
          for next_state, probability, reward in npr:
            c += probability * (reward + gamma*values[next_state])
            
          if c > C:
            C = c
            A = action
        policy[s] = A
        
        if b != policy[s]:
          done = False
    
    # Put your extra information needed for plots etc in this dictionary
    extra_info = {"All Values:": value_history, "Reduced Values:": reduced_value_history}

    # End code

    # Do not change the number of output values
    return {"Values": values, "Policy": policy}, extra_info

env = GridEnv_HW2(**base_kwargs)
res, extra_info = policy_iteration(env, 0.7)

 # The rounding off is just for making print statement cleaner
print(np.flipud(np.round(res['Values'], decimals=2)))
print(np.flipud(res['Policy']))

# The policy mentioned in the pdf to be used for TD lambda, do not modify this
def heuristic_policy(env, state):
    goal = env.goal_location
    dx = goal[0] - state[0]
    dy = goal[1] - state[1]
    if abs(dx) >= abs(dy):
        direction = (np.sign(dx), 0)
    else:
        direction = (0, np.sign(dy))
    for action, dir_val in env.actions.items():
        if dir_val == direction:
            target_action = action
            break
    return target_action

def td_lambda(env, lamda, seeds):
    alpha = 0.5
    gamma = 0.7
    N = len(seeds)
    # Usage of input_policy
    # heuristic_policy(env, state) -> action
    example_action = heuristic_policy(env, (1,2)) # Returns 'N' if goal is (9,9)

    # Example of env.step
    # env.step(state, action) -> Returns next_state, reward, is_terminal

    # Initial values
    values = np.zeros((10, 10))
    es = np.zeros((10,10))
    value_history = [values.copy()]

    for episode_idx in range(N):
         # Do not change this else the results will not match due to environment stochas
        np.random.seed(seeds[episode_idx])
        grey_in_loc = np.where(env.grid == 'grey_in')
        state = grey_in_loc[0][0], grey_in_loc[1][0]
        done = False
        while not done:
            action = heuristic_policy(env, state)
            ns, rew, is_terminal = env.step(state, action) 
            # env.step is already taken inside the loop for you, 
            # Don't use env.step anywhere else in your code

            # Begin code here
            delta = rew + gamma*values[ns] - values[state]
            es[state] = es[state] + 1
            
            for i in range(env.grid_size[0]):
              for j in range(env.grid_size[1]):
                values[i,j] = values[i,j] + alpha * delta * es[i,j]
                es[i,j] = gamma * lamda * es[i,j]
            
            state = ns
            value_history.append(values.copy())
            if state == env.goal_location:
              done=True
            
    # Put your extra information needed for plots etc in this dictionary
    extra_info = {"All Values:": value_history}

    # End code

    # Do not change the number of output values
    return {"Values": values}, extra_info

env = GridEnv_HW2(**base_kwargs)
res, extra_info = td_lambda(env, lamda=0.5, seeds=np.arange(1000))

 # The rounding off is just for making print statement cleaner
print(np.flipud(np.round(res['Values'], decimals=2)))

# This cell is only for your subjective evaluation results, display the results as asked in the pdf
# You can change it as you require, this code should run TD lamdba by default for different values of lambda

lamda_values = np.arange(0, 100+5, 5)/100
td_lamda_results = {}
extra_info = {}
for lamda in lamda_values:
    env = GridEnv_HW2(**base_kwargs)
    td_lamda_results[lamda], extra_info[lamda] = td_lambda(env, lamda,
                                                           seeds=np.arange(1000))

def get_results(kwargs):

    gridenv = GridEnv_HW2(**kwargs)

    policy_iteration_results = policy_iteration(gridenv, 0.7)[0]
    value_iteration_results = value_iteration(gridenv, 0.7)[0]
    td_lambda_results = td_lambda(env, 0.5, np.arange(1000))[0]

    final_results = {}
    final_results["policy_iteration"] = policy_iteration_results
    final_results["value_iteration"] = value_iteration_results
    final_results["td_lambda"] = td_lambda_results

    return final_results

# Do not edit this cell, generate results with it as is
if not os.path.exists(AICROWD_RESULTS_DIR):
    os.mkdir(AICROWD_RESULTS_DIR)

for params_file in os.listdir(DATASET_DIR):
  kwargs = np.load(os.path.join(DATASET_DIR, params_file), allow_pickle=True).item()
  results = get_results(kwargs)
  idx = params_file.split('_')[-1][:-4]
  np.save(os.path.join(AICROWD_RESULTS_DIR, 'results_' + idx), results)

# Check your score on the given test cases (There are more private test cases not provided)
target_folder = 'targets'
result_folder = AICROWD_RESULTS_DIR

def check_algo_match(results, targets):
    if 'Policy' in results:
        policy_match = results['Policy'] == targets['Policy']
    else:
        policy_match = True
    # Reference https://numpy.org/doc/stable/reference/generated/numpy.allclose.html
    rewards_match = np.allclose(results['Values'], targets['Values'], rtol=3)
    equal = rewards_match and policy_match
    return equal

def check_score(target_folder, result_folder):
    match = []
    for out_file in os.listdir(result_folder):
        res_file = os.path.join(result_folder, out_file)
        results = np.load(res_file, allow_pickle=True).item()
        idx = out_file.split('_')[-1][:-4]  # Extract the file number
        target_file = os.path.join(target_folder, f"targets_{idx}.npy")
        targets = np.load(target_file, allow_pickle=True).item()
        algo_match = []
        for k in targets:
            algo_results = results[k]
            algo_targets = targets[k]
            algo_match.append(check_algo_match(algo_results, algo_targets))
        match.append(np.mean(algo_match))
    return np.mean(match)

if os.path.exists(target_folder):
    print("Shared data Score (normalized to 1):", check_score(target_folder, result_folder))

import matplotlib.pyplot as plt

env = GridEnv_HW2(**base_kwargs)
val_res, val_extra_info = value_iteration(env, 0.7)
pol_res, pol_extra_info = policy_iteration(env, 0.7)

bin = [(i,j) for i in range(env.grid.shape[0]) for j in range(env.grid.shape[1]) if env.grid[i,j] == 'brown_in']
bout = [(i,j) for i in range(env.grid.shape[0]) for j in range(env.grid.shape[1]) if env.grid[i,j] == 'brown_out']
gin = [(i,j) for i in range(env.grid.shape[0]) for j in range(env.grid.shape[1]) if env.grid[i,j] == 'grey_in']
gout = [(i,j) for i in range(env.grid.shape[0]) for j in range(env.grid.shape[1]) if env.grid[i,j] == 'grey_out']

#print(bin, "\n", bout, "\n", gin, "\n", gout)

val_bin = [vals[bi] for vals in val_extra_info["All Values:"] for bi in bin]
val_bout = [vals[bi] for vals in val_extra_info["All Values:"] for bi in bout]
val_gin = [vals[bi] for vals in val_extra_info["All Values:"] for bi in gin]
val_gout = [[vals[bi] for vals in val_extra_info["All Values:"]] for bi in gout]

pol_bin = [vals[bi] for vals in pol_extra_info["All Values:"] for bi in bin]
pol_bout = [vals[bi] for vals in pol_extra_info["All Values:"] for bi in bout]
pol_gin = [vals[bi] for vals in pol_extra_info["All Values:"] for bi in gin]
pol_gout = [[vals[bi] for vals in pol_extra_info["All Values:"]] for bi in gout]

pol_bin2 = [vals[bi] for vals in pol_extra_info["Reduced Values:"] for bi in bin]
pol_bout2 = [vals[bi] for vals in pol_extra_info["Reduced Values:"] for bi in bout]
pol_gin2 = [vals[bi] for vals in pol_extra_info["Reduced Values:"] for bi in gin]
pol_gout2 = [[vals[bi] for vals in pol_extra_info["Reduced Values:"]] for bi in gout]


######################### bin ##############################
plt.figure(figsize=(12, 8), dpi=80)

plt.title("J('brown_in') vs number of iterations")
plt.xlabel("number of iterations")
plt.ylabel("J('brown_in')")

plt.plot(val_bin, label="value iteration")
plt.plot(pol_bin2, label="policy iteration")
plt.scatter(range(len(val_bin)), val_bin)
plt.scatter(range(len(pol_bin2)), pol_bin2)
plt.legend()

######################### bout ##############################
plt.figure(figsize=(12, 8), dpi=80)

plt.title("J('brown_out') vs number of iterations")
plt.xlabel("number of iterations")
plt.ylabel("J('brown_out')")

plt.plot(val_bout, label="value iteration")
plt.plot(pol_bout2, label="policy iteration")
plt.scatter(range(len(val_bout)), val_bout)
plt.scatter(range(len(pol_bout2)), pol_bout2)
plt.legend()

######################### gin ##############################
plt.figure(figsize=(12, 8), dpi=80)

plt.title("J('grey_in') vs number of iterations")
plt.xlabel("number of iterations")
plt.ylabel("J('grey_in')")

plt.plot(val_gin, label="value iteration")
plt.plot(pol_gin2, label="policy iteration")
plt.scatter(range(len(val_gin)), val_gin)
plt.scatter(range(len(pol_gin2)), pol_gin2)
plt.legend()

######################### gout ##############################
for i in range(len(gout)):
  plt.figure(figsize=(12, 8), dpi=80)

  plt.title("J('grey_out %d') vs number of iterations" % i)
  plt.xlabel("number of iterations")
  plt.ylabel("J('grey_out %d')" % i)

  plt.plot(val_gout[i], label="value iteration")
  plt.plot(pol_gout2[i], label="policy iteration")
  plt.scatter(range(len(val_gout[i])), val_gout[i])
  plt.scatter(range(len(pol_gout2[i])), pol_gout2[i])
  plt.legend()

td_bin = {l: td_lamda_results[l]["Values"][bi] for bi in bin for l in lamda_values}
td_bout = {l: td_lamda_results[l]["Values"][bi] for bi in bout for l in lamda_values}
td_gin = {l: td_lamda_results[l]["Values"][bi] for bi in gin for l in lamda_values}
td_gout1 = {l: td_lamda_results[l]["Values"][bi] for bi in [gout[0]] for l in lamda_values}

vbin = pol_res['Values'][bin[0]]
vbout = pol_res['Values'][bout[0]]
vgin = pol_res['Values'][gin[0]]
vgout = pol_res['Values'][gout[0]]

plt.figure(figsize=(12, 8), dpi=80)

plt.title("J vs lambda in TD(lambda)")
plt.xlabel("lamdba")
plt.ylabel("J")

plt.plot(list(td_bin.keys()), list(td_bin.values()), label="J('brown_in')")
plt.scatter(list(td_bin.keys()), list(td_bin.values()))
plt.plot(list(td_bout.keys()), list(td_bout.values()), label="J('brown_out')")
plt.scatter(list(td_bout.keys()), list(td_bout.values()))
plt.plot(list(td_gin.keys()), list(td_gin.values()), label="J('grey_in')")
plt.scatter(list(td_gin.keys()), list(td_gin.values()))
plt.plot(list(td_gout1.keys()), list(td_gout1.values()), label="J('grey_out 0')")
plt.scatter(list(td_gout1.keys()), list(td_gout1.values()))


#plt.plot([0,1], [vbin, vbin], linestyle = 'dashed', label="J*('brown_in')", color="blue", linewidth=1)
#plt.plot([0,1], [vbout, vbout], linestyle = 'dashed', label="J*('brown_out')", color="orange", linewidth=1)
#plt.plot([0,1], [vgin, vgin], linestyle = 'dashed', label="J*('grey_in')", color="green", linewidth=1)
#plt.plot([0,1], [vgout, vgout], linestyle = 'dashed', label="J*('grey_out')", color="red", linewidth=1)
#plt.plot([[0], keys[-1]], [vi_results[0.9]['Expected Reward']['B'], vi_results[0.9]['Expected Reward']['B']], linestyle = 'dashed', label="J*(B)", color="orange")
#plt.plot([keys[0], keys[-1]], [vi_results[0.9]['Expected Reward']['C'], vi_results[0.9]['Expected Reward']['C']], linestyle = 'dashed', label="J*(C)", color="green")
plt.legend()

#print(pol_res['Values'])
#print(pol_extra_info["Reduced Values:"])
pol_errors = [np.sum(np.square(pol_extra_info['Reduced Values:'][i] - val_res['Values'])/(env.grid_size[0] * env.grid_size[1]))**0.5 for i in range(len(pol_extra_info["Reduced Values:"]))]

plt.figure(figsize=(12, 8), dpi=80)
plt.title("Error in J_i vs number of iterations for policy iteration")
plt.xlabel("number of iterations i")
plt.ylabel("J* - J_i")

plt.plot(pol_errors, label="error")
plt.scatter(range(len(pol_errors)), pol_errors)
plt.legend()

lamda_values = [0, 0.25, 0.5, 0.75, 1]
td_errors = [np.sum(np.square(td_lamda_results[l]['Values'] - val_res['Values'])/(env.grid_size[0] * env.grid_size[1]))**0.5 for l in lamda_values]

plt.figure(figsize=(12, 8), dpi=80)

plt.title("Error in J vs lambda for TD(lambda)")
plt.xlabel("lambda")
plt.ylabel("error")

plt.plot(lamda_values, td_errors, label="error")
plt.scatter(lamda_values, td_errors)
plt.legend()

!DATASET_PATH=$AICROWD_DATASET_PATH aicrowd notebook submit --no-verify -c iit-m-rl-assignment-2-gridworld -a assets