ui export

.gitignore

@@ -2,3 +2,5 @@ logs/*
 _pycache_
 *.pyc
 *tfevents*
+db/econ_data/*
+db/metabase_data/*

db/.env

@@ -0,0 +1,2 @@
+MB=123456789123456798
+POSTGRES_PASSWORD=change_me

db/docker-compose.yml

@@ -5,7 +5,7 @@ services:
   postgres:
     image: postgres:10
     volumes:
-      - "./db_data:/var/lib/postgresql/data"
+      - "./metabase_data:/var/lib/postgresql/data"
     restart: always
     environment:
       POSTGRES_DB: metabase
@@ -20,7 +20,6 @@ services:
       - "3000:3000"
     restart: always
     environment:
-      JAVA_TIMEZONE: Europe/Istanbul
       MB_ENCRYPTION_SECRET_KEY: ${MB}
       MB_DB_TYPE: postgres
       MB_DB_DBNAME: metabase
@@ -33,9 +32,11 @@ services:
     image: mongo
     restart: always
     environment:
-      MONGO_INITDB_DATABASE: example
+      MONGO_INITDB_DATABASE: econ
       MONGO_INITDB_ROOT_USERNAME: root
-      MONGO_INITDB_ROOT_PASSWORD: example
+      MONGO_INITDB_ROOT_PASSWORD: econ
+    ports:
+      - '27017:27017'
     volumes:
-      - /home/angelo/mongodata:/data/db
+      - ./econ_data:/data/db
   

db/export_to_mongo.py

@@ -0,0 +1,48 @@
+from pymongo import MongoClient
+cmt="current_market_trades"
+cma="current_market_asks"
+cmb="current_market_bids"
+
+class MongoExporter():
+    
+    def __init__(self,conn_string,db,econ) -> None:
+        
+    # Provide the mongodb atlas url to connect python to mongodb using pymongo
+  
+ 
+    # Create a connection using MongoClient. You can import MongoClient or use pymongo.MongoClient
+        client = MongoClient(conn_string)
+ 
+     # Create the database for our example (we will use the same database throughout the tutorial
+        self.client=client[db]
+        self.econ=econ
+
+    def reset_market(self):
+        """Resets the current market data in the analytics stack"""
+        c=self.client
+        c[cmt].drop()
+        c[cma].drop()
+        c[cmb].drop()
+
+    def submit_full_market(self):
+        c=self.client
+        market=self.econ.get_component("ContinuousDoubleAuction")
+        self.reset_market()
+
+        for i in range(len(market.executed_trades)):
+            step=market.executed_trades[i]
+            if len(step)>0:
+                for transaction in step:
+                   transaction["step"]=i
+                   c[cmt].insert_one(transaction)
+        c[cma].insert_many(market.asks)
+        c[cmb].insert_many(market.bids)
+
+    
+
+
+
+ 
+
+  
+# This is added so that many files can reuse the function get_database()

main.py

@@ -64,7 +64,7 @@ env_config = {
     'allow_observation_scaling': True,
     'dense_log_frequency': 100, 
     'world_dense_log_frequency':1,
-    'energy_cost':0,
+    'energy_cost':0.2,
     'energy_warmup_method': "auto",
     'energy_warmup_constant': 4000,
     
@@ -120,7 +120,7 @@ eval_env_config = {
     'isoelastic_eta':0.001,
     'dense_log_frequency': 1, 
     'world_dense_log_frequency':1, 
-    'energy_cost':0,
+    'energy_cost':0.2,
     'energy_warmup_method': "auto",
     'energy_warmup_constant': 4000,
     
@@ -138,6 +138,7 @@ eval_env_config = {
 }
 
 num_frames=1
+CONNECTION_STRING = "mongodb+srv://root:econ@localhost:/econ"
 
 class TensorboardCallback(BaseCallback):
     """
@@ -292,8 +293,8 @@ runname="run_{}".format(run_number)
 model_db=[None,None] # object for storing model
 
 
-model = MaskablePPO("MlpPolicy",n_steps=int(env_config['episode_length']*2),ent_coef=0.001, vf_coef=0.5 ,gamma=0.99, learning_rate=learning_rate_adj_basic,env=stackenv_basic, seed=445,verbose=1,device="cuda",tensorboard_log="./log")
-model_trade=MaskablePPO("MlpPolicy",n_steps=int(env_config['episode_length']*2),ent_coef=0.001, vf_coef=0.5 ,gamma=0.99, learning_rate=learning_rate_adj_trade,env=stackenv_traid, seed=445,verbose=1,device="cuda",tensorboard_log="./log")
+model = MaskablePPO("MlpPolicy",n_steps=int(env_config['episode_length']*2),ent_coef=0.001, vf_coef=0.5 ,gamma=0.99, learning_rate=4e-4,env=stackenv_basic, seed=500,verbose=1,device="cuda",tensorboard_log="./log")
+model_trade=MaskablePPO("MlpPolicy",n_steps=int(env_config['episode_length']*2),ent_coef=0.001, vf_coef=0.5 ,gamma=0.99, learning_rate=4e-4,env=stackenv_traid, seed=500,verbose=1,device="cuda",tensorboard_log="./log")
 
 n_agents=econ.n_agents
 
@@ -308,9 +309,9 @@ while True:
     #Train
     runname="run_{}_{}".format(run_number,"basic")
 
-    thread_model=Thread(target=train,args=(model,total_required_for_episode_basic*150,econ,True,runname,model_db,0))
+    thread_model=Thread(target=train,args=(model,total_required_for_episode_basic*50,econ,True,runname,model_db,0))
     runname="run_{}_{}".format(run_number,"trader")
-    thread_model_traid=Thread(target=train,args=(model_trade,total_required_for_episode_traid*150,econ,False,runname,model_db,1))
+    thread_model_traid=Thread(target=train,args=(model_trade,total_required_for_episode_traid*50,econ,False,runname,model_db,1))
     
     thread_model.start()
     thread_model_traid.start()

reqirements.txt

@@ -1,5 +1,4 @@
-dash
 pandas
-
+pymongo
 gym
 ray[rllib]

test.py

@@ -25,6 +25,7 @@ from stable_baselines3.common.callbacks import BaseCallback
 import yaml
 import time
 from threading import Thread 
+from db.export_to_mongo import MongoExporter
 
 env_config = {
     # ===== SCENARIO CLASS =====
@@ -52,7 +53,7 @@ env_config = {
     # ===== SCENARIO CLASS ARGUMENTS =====
     # (optional) kwargs that are added by the Scenario class (i.e. not defined in BaseEnvironment)
     
-    'starting_agent_coin': 20,
+    'starting_agent_coin': 50,
     'fixed_four_skill_and_loc': True,
     
     # ===== STANDARD ARGUMENTS ======
@@ -82,63 +83,8 @@ env_config = {
 }
 
 
-eval_env_config = {
-    # ===== SCENARIO CLASS =====
-    # Which Scenario class to use: the class's name in the Scenario Registry (foundation.scenarios).
-    # The environment object will be an instance of the Scenario class.
-    'scenario_name': 'econ',
-    
-    # ===== COMPONENTS =====
-    # Which components to use (specified as list of ("component_name", {component_kwargs}) tuples).
-    #   "component_name" refers to the Component class's name in the Component Registry (foundation.components)
-    #   {component_kwargs} is a dictionary of kwargs passed to the Component class
-    # The order in which components reset, step, and generate obs follows their listed order below.
-    'components': [
-        # (1) Building houses
-        ('Craft', {'skill_dist': "pareto", 'commodities': ["Gem"],'max_skill_amount_benefit':2}),
-        # (2) Trading collectible resources
-        ('ContinuousDoubleAuction', {'max_num_orders': 10}),
-        # (3) Movement and resource collection
-        ('SimpleGather', {}),
-        ('ExternalMarket',{'market_demand':{
-            'Gem': 15
-        }}),
-    ],
-    
-    # ===== SCENARIO CLASS ARGUMENTS =====
-    # (optional) kwargs that are added by the Scenario class (i.e. not defined in BaseEnvironment)
-    
-    'starting_agent_coin': 20,
-    'fixed_four_skill_and_loc': True,
-    
-    # ===== STANDARD ARGUMENTS ======
-    # kwargs that are used by every Scenario class (i.e. defined in BaseEnvironment)
-    'agent_composition': {"BasicMobileAgent": 20,"TradingAgent":5},            # Number of non-planner agents (must be > 1)
-    'world_size': [1, 1], # [Height, Width] of the env world
-    'episode_length': 256, # Number of timesteps per episode
-    'allow_observation_scaling': True,
-    'isoelastic_eta':0.001,
-    'dense_log_frequency': 1, 
-    'world_dense_log_frequency':1, 
-    'energy_cost':0,
-    'energy_warmup_method': "auto",
-    'energy_warmup_constant': 4000,
-    
-    # In multi-action-mode, the policy selects an action for each action subspace (defined in component code).
-    # Otherwise, the policy selects only 1 action.
-    'multi_action_mode_agents': False,
-    'multi_action_mode_planner': False,
-    
-    # When flattening observations, concatenate scalar & vector observations before output.
-    # Otherwise, return observations with minimal processing.
-    'flatten_observations': False,
-    # When Flattening masks, concatenate each action subspace mask into a single array.
-    # Note: flatten_masks = True is required for masking action logits in the code below.
-    'flatten_masks': True,
-}
-
 num_frames=1
-
+CONNECTION_STRING = "mongodb://root:econ@localhost:27017/"
 class TensorboardCallback(BaseCallback):
     """
     Custom callback for plotting additional values in tensorboard.
@@ -238,7 +184,6 @@ def printReplay(econ,agentid):
 
 #Setup Env Objects
 econ=foundation.make_env_instance(**env_config)
-
 market=econ.get_component("ContinuousDoubleAuction")
 action=market.get_n_actions("TradingAgent")
 baseEconWrapper=BaseEconWrapper(econ)
@@ -246,8 +191,8 @@ baseEconWrapper.run()
 time.sleep(0.5)
 mobileRecieverEconWrapper=RecieverEconWrapper(base_econ=baseEconWrapper,agent_classname="BasicMobileAgent")
 tradeRecieverEconWrapper=RecieverEconWrapper(base_econ=baseEconWrapper,agent_classname="TradingAgent")
-sb3_traderConverter=SB3EconConverter(tradeRecieverEconWrapper,econ,"TradingAgent",True)
-sb3Converter=SB3EconConverter(mobileRecieverEconWrapper,econ,"BasicMobileAgent",True)
+sb3_traderConverter=SB3EconConverter(tradeRecieverEconWrapper,econ,"TradingAgent",False)
+sb3Converter=SB3EconConverter(mobileRecieverEconWrapper,econ,"BasicMobileAgent",False)
 # attach sb3 wrappers
 
 monenv=VecMonitor(venv=sb3Converter,info_keywords=["social/productivity","trend/productivity"])
@@ -257,28 +202,8 @@ stackenv_basic=vec_frame_stack.VecFrameStack(venv=monenv,n_stack=num_frames)
 stackenv_traid=vec_frame_stack.VecFrameStack(venv=montraidingenv,n_stack=num_frames)
 # Model setup complete
 
-# Setup Eval Env
-econ_eval=foundation.make_env_instance(**eval_env_config)
-
-
-baseEconWrapper_eval=BaseEconWrapper(econ_eval)
-baseEconWrapper_eval.run()
-time.sleep(0.5)
-mobileRecieverEconWrapper_eval=RecieverEconWrapper(base_econ=baseEconWrapper_eval,agent_classname="BasicMobileAgent")
-tradeRecieverEconWrapper_eval=RecieverEconWrapper(base_econ=baseEconWrapper_eval,agent_classname="TradingAgent")
-sb3_traderConverter_eval=SB3EconConverter(tradeRecieverEconWrapper_eval,econ_eval,"TradingAgent",False)
-sb3Converter_eval=SB3EconConverter(mobileRecieverEconWrapper_eval,econ_eval,"BasicMobileAgent",False)
-# attach sb3 wrappers
-
-monenv_eval=VecMonitor(venv=sb3Converter_eval,info_keywords=["social/productivity","trend/productivity"])
-montraidingenv_eval=VecMonitor(venv=sb3_traderConverter_eval)
-
-stackenv_basic_eval=vec_frame_stack.VecFrameStack(venv=monenv_eval,n_stack=num_frames)
-stackenv_traid_eval=vec_frame_stack.VecFrameStack(venv=montraidingenv_eval,n_stack=num_frames)
-
-
 obs=monenv.reset()
-
+obs=montraidingenv.reset()
 
 # define training functions
 def train(model,timesteps, econ_call,process_bar,name,db,index):
@@ -297,39 +222,41 @@ model_trade=MaskablePPO("MlpPolicy",n_steps=int(env_config['episode_length']*2),
 # Setup complete
 
 # Load models
-model.load("basic.ai")
-model_trade.load("trade.ai")
+model=model.load("basic.ai")
+model_trade=model_trade.load("trade.ai")
+
+exporter=MongoExporter(CONNECTION_STRING,"econ",econ)
 
 while True:
     print("### EVAL ###")
-    obs_basic=stackenv_basic_eval.reset()
-    obs_trade=stackenv_traid_eval.reset()
+    obs_basic=stackenv_basic.reset()
+    obs_trade=stackenv_traid.reset()
     done=False
-    for i in tqdm(range(eval_env_config['episode_length'])):
+    for i in tqdm(range(env_config['episode_length'])):
         #create masks
-        masks_basic=stackenv_basic_eval.action_masks()
-        masks_trade=stackenv_traid_eval.action_masks()
+        masks_basic=stackenv_basic.action_masks()
+        masks_trade=stackenv_traid.action_masks()
         # get actions
         action_basic=model.predict(obs_basic,action_masks=masks_basic)
         action_trade=model_trade.predict(obs_trade,action_masks=masks_trade)
         #submit async directly for non blocking operation
-        sb3Converter_eval.step_async(action_basic[0])
-        sb3_traderConverter_eval.step_async(action_trade[0])
+        sb3Converter.step_async(action_basic[0])
+        sb3_traderConverter.step_async(action_trade[0])
         # retieve full results
-        obs_basic,rew_basic,done_e,info=stackenv_basic_eval.step(action_basic[0])
-        obs_trade,rew_trade,done_e,info=stackenv_traid_eval.step(action_trade[0])
+        obs_basic,rew_basic,done_e,info=stackenv_basic.step(action_basic[0])
+        obs_trade,rew_trade,done_e,info=stackenv_traid.step(action_trade[0])
         done=done_e[0]
 
 
-
-    market=econ_eval.get_component("ContinuousDoubleAuction")
-    craft=econ_eval.get_component("Craft")
+    exporter.submit_full_market()
+    market=econ.get_component("ContinuousDoubleAuction")
+    craft=econ.get_component("Craft")
    # trades=market.get_dense_log()
     build=craft.get_dense_log()
-    met=econ_eval.previous_episode_metrics
-    printReplay(econ_eval,21)
-   # printMarket(trades)
-  #  printBuilds(builds=build)
+    met=econ.previous_episode_metrics
+    printReplay(econ,21)
+   #printMarket(trades)
+  #printBuilds(builds=build)
     print("social/productivity: {}".format(met["social/productivity"]))
     print("labor/weighted_cost: {}".format(met["labor/weighted_cost"]))
     print("labor/warmup_integrator: {}".format(met["labor/warmup_integrator"]))

ui/.env

@@ -1,3 +0,0 @@
-MB=change_me
-POSTGRES_PASSWORD=change_me
-PGADMIN_PASSWORD=change_me