citylearn.reward_function module

class citylearn.reward_function.BuildingDynamicsReward(env: citylearn.citylearn.CityLearnEnv)

Bases: citylearn.reward_function.RewardFunction

calculate() → List[float]

Returned reward assumes that the building-agents act independently of each other, without sharing information through the reward.

Recommended for use with the SAC controllers.

Notes

Reward value is calculated as \([\textrm{min}(-e_0^3, 0), \dots, \textrm{min}(-e_n^3, 0)]\) where \(e\) is electricity_consumption and \(n\) is the number of agents.

calculate_comfort_reward() → List[float]

calculate_peak_reward() → List[float]

calculate_storage_reward() → List[float]

class citylearn.reward_function.IndependentSACReward(env: citylearn.citylearn.CityLearnEnv)

Bases: citylearn.reward_function.RewardFunction

calculate() → List[float]

Returned reward assumes that the building-agents act independently of each other, without sharing information through the reward.

Recommended for use with the SAC controllers.

Notes

Reward value is calculated as \([\textrm{min}(-e_0^3, 0), \dots, \textrm{min}(-e_n^3, 0)]\) where \(e\) is electricity_consumption and \(n\) is the number of agents.

class citylearn.reward_function.MARL(env: citylearn.citylearn.CityLearnEnv)

Bases: citylearn.reward_function.RewardFunction

calculate() → List[float]

Calculates MARL reward.

Notes

Reward value is calculated as \(\textrm{sign}(-e) \times 0.01(e^2) \times \textrm{max}(0, E)\) where \(e\) is the building electricity_consumption and \(E\) is the district electricity_consumption.

class citylearn.reward_function.RewardFunction(env: citylearn.citylearn.CityLearnEnv, **kwargs)

Bases: object

calculate() → List[float]

Calculates default reward.

Notes

Reward value is calculated as \([\textrm{min}(-e_0, 0), \dots, \textrm{min}(-e_n, 0)]\) where \(e\) is electricity_consumption and \(n\) is the number of agents.

property env: citylearn.citylearn.CityLearnEnv

Simulation environment.