def __init__(self):
# Define
obs_high = np.array([
10000, #demand
1000, #available MW
])
obs_low = np.array([
0, #demand
0, #available MW
])
# self.observation_space = spaces.Box(obs_low, obs_high, dtype=np.float32)
self.observation_space = spaces.Box(obs_low, obs_high )
self.action_space = spaces.Discrete(10)
self.seed()
self.viewer = None
self.state = None
self._state_dict = None
self.steps_beyond_done = None
# Need a way to assign or find id.
self.id = 3
self.io = AsyncClient(self.id)
self.label = 'Bayswater'
self.total_steps = 0
self.epoch_reward = 0
self.last_action = 0
def connect(self, participant_name, id_no):
print("Connecting as ", participant_name, id_no)
self.id = id_no
self.label = participant_name
self.io = AsyncClient(self.id)
class MultiBidMarket(gym.Env):
"""
"""
metadata = {
'render.modes': ['human'],
}
def __init__(self):
obs_high = [
market_config['MAX_DEMAND'], #demand
TOTAL_CAPACITY, #Amount Dispatched
market_config['MAX_PRICE'], #Last Price
]
obs_low = [
0, #demand
0, #Amount Dispatched
0, #Last Price
]
# Create a spot in the observation space for each participant other than this one.
if market_config['REVEAL_PREVIOUS_BIDS']:
for i in range(len(market_config['PARTICIPANTS']) - 1):
for band in range(market_config['NUM_BANDS']):
obs_high.append(market_config['MAX_PRICE'])
obs_low.append(market_config['MIN_PRICE'])
# Create a spot in the observation space for each participant other than this one
# These spots are used to deliver the way the other participants last bid in this context.
if market_config['PROVIDE_HISTORICAL_CONTEXT']:
for i in range(len(market_config['PARTICIPANTS']) - 1):
self._saved_history = {}
for band in range(market_config['NUM_BANDS']):
obs_high.append(market_config['MAX_PRICE'])
obs_low.append(market_config['MIN_PRICE'])
# Define
obs_high = np.array(obs_high)
obs_low = np.array(obs_low)
print('Obs High:', obs_high)
print('Obs Low:', obs_low)
# self.observation_space = spaces.Box(obs_low, obs_high, dtype=np.float32)
self.observation_space = spaces.Box(obs_low, obs_high)
# self.action_space = spaces.Discrete(10)
try:
self.action_space = spaces.MultiDiscrete([
market_config['MAX_PRICE']
for b in range(market_config['NUM_BANDS'])
]) #5 bands of $0 - $9 bids.
except: # It seems that in some versions of openai gym, the MultiDiscrete constructor needs an array of high/lows.
self.action_space = spaces.MultiDiscrete([
[0, market_config['MAX_PRICE']]
for b in range(market_config['NUM_BANDS'])
]) #5 bands of $0 - $9 bids.
self.seed()
self.viewer = None
self.state = np.array(obs_low)
self._state_dict = None
self.steps_beyond_done = None
# Need a way to assign or find id.
self.total_steps = 0
self.epoch_reward = 0
self.last_action = 0
def connect(self, participant_name, id_no):
print("Connecting as ", participant_name, id_no)
self.id = id_no
self.label = participant_name
self.io = AsyncClient(self.id)
def seed(self, seed=None):
self.np_random, seed = seeding.np_random(seed)
return [seed]
def step(self, action):
# print("ACTION: ", action)
# print(self.action_space)
# assert self.action_space.contains(action), "%r (%s) invalid"%(action, type(action))
self.total_steps += 1
# self.state = (x,x_dot,theta,theta_dot)
# print("Action:", action)
self.last_action = action
data = {
'id':
self.id,
'label':
self.label,
'bids': [[int(a), TOTAL_CAPACITY / market_config['NUM_BANDS']]
for a in action],
}
reply = self.io.send(data)
self._state_dict = reply
amount_dispatched = 0 if self.label not in reply[
'dispatch'] else float(reply['dispatch'][self.label])
next_state = [
int(reply['next_demand'])
if market_config['SHOW_NEXT_DEMAND'] else 0, #Next Demand
int(amount_dispatched), #Amount Dispatched
int(reply['price']), # Last Price,
]
# Add each participant's previous bids to the observation space.
# Loop through each participant in the sorted participant list
if market_config['REVEAL_PREVIOUS_BIDS']:
for p in market_config['PARTICIPANTS']:
if p != self.label:
prices = [0] * market_config['NUM_BANDS']
for bid in reply['all_bids'][p]:
prices[bid['band']] = int(bid['price'])
# Add the bid info to the observation
next_state += prices
# Add each participant's previous bids in a similar demand situation.
if market_config['PROVIDE_HISTORICAL_CONTEXT']:
for p in market_config['PARTICIPANTS']:
if p != self.label:
prices = [0] * market_config['NUM_BANDS']
if reply['next_demand'] in self._saved_history:
for bid in reply['all_bids'][p]:
prices[bid['band']] = int(bid['price'])
# Add the bid info to the observation
next_state += prices
self._saved_history[reply['next_demand']] = reply['all_bids']
# Reward is product of dispatched and
reward = amount_dispatched * float(reply['price'])
self.epoch_reward += reward
# Every day, start a new epoch.
done = False
if self.total_steps % 48 == 0:
done = True
# the next next_state, the reward for the last action, the current next_state, a boolean representing whether the current episode of our model is done and some additional info on our problem
return np.array(next_state), reward, done, {}
def reset(self):
# We used to reset the state each time - dont see how this is sensible as its a continuation in simulation backend.
# Still worthwhile having regular epochs as we want it to think about maximising daily bidding (or some other period)
# But resetting state is a little incompehensible.
# Edge case here is first instance, whereby state is set to obs_low.
# self.state = self.np_random.uniform(low=0, high=0.05, size=(8,))
# print(str({"metric": "epoch_reward", "value": self.epoch_reward, "step": self.total_steps}))
print('{"metric": "epoch_reward", "value": ' + str(self.epoch_reward) +
', "step":' + str(self.total_steps) + '}')
print('{"metric": "unique_bids", "value": ' +
str(logbook().get_num_unique_bids(previous_steps=50)) +
', "step":' + str(self.total_steps) + '}')
logbook().record_epoch_reward(self.epoch_reward)
self.epoch_reward = 0
# Every X steps, write results to file in case of dramatic failure.
if self.total_steps % 20000 == 0:
# if self.total_steps % 100 == 0:
logbook().save_json(label=str(self.label))
return np.array(self.state)
def render(self, mode='human'):
# Log bid/value in Floydhub
# print('{"metric": "bid", "value": '+str(self.last_action)+', "step":'+str(self.total_steps)+'}')
# print("Rendering")
# Log in logbook suite
logbook().record_price(self._state_dict['price'], self.total_steps)
logbook().record_demand(self._state_dict['demand'], self.total_steps)
# Log bidstack in logbook suite.
for label in self._state_dict['all_bids']:
for bid in self._state_dict['all_bids'][label]:
logbook().record_bid(bid['label'], bid['price'],
bid['quantity'], self.total_steps)
return None
def close(self):
pass
class SimpleMarket(gym.Env):
"""
"""
metadata = {
'render.modes': ['human'],
}
def __init__(self):
# Define
obs_high = np.array([
10000, #demand
1000, #available MW
])
obs_low = np.array([
0, #demand
0, #available MW
])
# self.observation_space = spaces.Box(obs_low, obs_high, dtype=np.float32)
self.observation_space = spaces.Box(obs_low, obs_high )
self.action_space = spaces.Discrete(10)
self.seed()
self.viewer = None
self.state = None
self._state_dict = None
self.steps_beyond_done = None
# Need a way to assign or find id.
self.id = 3
self.io = AsyncClient(self.id)
self.label = 'Bayswater'
self.total_steps = 0
self.epoch_reward = 0
self.last_action = 0
def seed(self, seed=None):
self.np_random, seed = seeding.np_random(seed)
return [seed]
def step(self, action):
assert self.action_space.contains(action), "%r (%s) invalid"%(action, type(action))
self.total_steps += 1
# self.state = (x,x_dot,theta,theta_dot)
# print("Action:", action)
self.last_action = action
data = {
'id': self.id,
'label':self.label,
'bids' : [
[int(action),1000],
],
}
reply = self.io.send(data)
self._state_dict = reply
# state should be a tuple of vals.
next_state = (reply['next_demand'],10)
# Reward is product of dispatched and
reward = 0 if self.label not in reply['dispatch'] else float(reply['dispatch'][self.label]) * float(reply['price'])
self.epoch_reward += reward
# Every day, start a new epoch.
done = False
if self.total_steps % 48 == 0:
done = True
# the next next_state, the reward for the last action, the current next_state, a boolean representing whether the current episode of our model is done and some additional info on our problem
return np.array(next_state), reward, done, {}
def reset(self):
self.state = self.np_random.uniform(low=-0.05, high=0.05, size=(2,))
# print(str({"metric": "epoch_reward", "value": self.epoch_reward, "step": self.total_steps}))
print('{"metric": "epoch_reward", "value": '+str(self.epoch_reward)+', "step":'+str(self.total_steps)+'}')
logbook().record_epoch_reward(self.epoch_reward)
self.epoch_reward = 0
return np.array(self.state)
def render(self, mode='human'):
# Log bid/value in Floydhub
# print('{"metric": "bid", "value": '+str(self.last_action)+', "step":'+str(self.total_steps)+'}')
# Log in logbook suite
logbook().record_price(self._state_dict['price'], self.total_steps)
logbook().record_demand(self._state_dict['demand'], self.total_steps)
# Log bidstack in logbook suite.
for bid in self._state_dict['all_bids']:
logbook().record_bid(bid['label'], bid['price'], bid['quantity'], self.total_steps)
return None
def close(self):
pass