Merge pull request #35 from rradules/mo-games

Merge Mo-Games into main

momaland/envs/beach_domain/beach_domain.py

@@ -62,7 +62,7 @@ class MOBeachDomain(MOParallelEnv):
     These attributes should not be changed after initialization.
     """
 
-    metadata = {"render_modes": ["human"], "name": "mobeach_v0"}
+    metadata = {"render_modes": ["human"], "name": "mobeach_v0", "is_parallelizable": True}
 
     # TODO does this environment require max_cycle?
     def __init__(

momaland/envs/breakthrough/__init__.py

@@ -0,0 +1 @@
+"""Adaptation of the Breakthrough game to multi-objective settings."""

momaland/envs/breakthrough/breakthrough_check.py

@@ -0,0 +1,43 @@
+"""MO Breakthrough check."""
+
+import numpy as np
+from breakthrough import MOBreakthrough
+
+
+def check_env():
+    """Check the environment."""
+    environment = MOBreakthrough(board_width=6, board_height=6, num_objectives=4, render_mode="ansi")
+    environment.reset(seed=42)
+
+    for agent in environment.agent_iter():
+        observation, reward, termination, truncation, info = environment.last()
+
+        print("rewards from the last timestep", reward)
+        print("cumulative rewards before action", environment._cumulative_rewards)
+        if termination or truncation:
+            action = None
+        else:
+            if observation:
+                print("environment before action:")
+                environment.render()
+                # print("observation:")
+                # print(observation)
+                # this is where you would insert your policy
+                action = np.where(observation["action_mask"] != 0)[0][0]
+                print("action: ", action)
+
+        environment.step(action)
+        print("environment after action:")
+        environment.render()
+        print("cumulative rewards after action", environment._cumulative_rewards)
+
+    # print("observation: ", observation)
+    # print("reward: ", reward)
+    print("game end: rewards", environment.rewards)
+    print("game end: cumulative rewards", environment._cumulative_rewards)
+    # name = input("Press key to end\n")
+    environment.close()
+
+
+if __name__ == "__main__":
+    (check_env())

momaland/envs/breakthrough/mobreakthrough_v0.py

@@ -0,0 +1,5 @@
+"""Breakthrough environment with multiple objectives."""
+from momaland.envs.breakthrough.breakthrough import env, raw_env
+
+
+__all__ = ["env", "raw_env"]

momaland/envs/congestion_game/congestion_game.py

@@ -132,7 +132,7 @@ def __init__(
         self.cost_function = dict()
         self._create_latency_and_cost_function(nx.get_edge_attributes(self.graph, "latency_function"), num_agents)
 
-    metadata = {"render_modes": ["human"], "name": "mocongestion_v0"}
+    metadata = {"render_modes": ["human"], "name": "mocongestion_v0", "is_parallelizable": True}
 
     # this cache ensures that same space object is returned for the same agent
     # allows action space seeding to work as expected

momaland/envs/connect4/__init__.py

@@ -0,0 +1 @@
+"""Adaptation of the Connect4 game to multi-objective settings."""