ochre_gym.spaces.act_spaces

get_action_space_from_ochre

ochre_gym.spaces.act_spaces.get_action_space_from_ochre(equipment_by_end_use: Dict, user_config: Dict, vectorize_actions: bool) -> Tuple[Union[spaces.Dict, spaces.Box], OrderedDict]

Obtain the action space for an OCHRE dwelling simulation.

Parameters:

Name	Type	Description	Default
equipment_by_end_use	Dict	A dictionary of equipment by end use for the OCHRE Dwelling	required
user_config	Dict	A dictionary specifying which control types to use for each device in a Dwelling	required
vectorize_actions	bool	Whether to vectorize the actions	required

Returns:

Name	Type	Description
action_space	Union[spaces.Dict, spaces.Box]	The action space for an OCHRE Dwelling.
action_keys	OrderedDict	The keys of the action dict, keys are equipment category and values are control types.

vector_action_to_ochre_control

ochre_gym.spaces.act_spaces.vector_action_to_ochre_control(action: np.ndarray, action_keys: OrderedDict) -> OrderedDict

Unflatten a numpy array into a dict of actions.

Parameters:

Name	Type	Description	Default
action	np.ndarray	The flattened action array.	required
action_keys	OrderedDict	The keys of the action dict.	required

Returns:

Name	Type	Description
action_dict	OrderedDict	The action dict.

ClipActionComposite

ochre_gym.spaces.act_spaces.ClipActionComposite

Bases: gym.ActionWrapper, gym.utils.RecordConstructorArgs

Clip the spaces.Box action spaces of a composite action space to their bounds.

__init__(env: gym.Env)

A wrapper for clipping continuous actions in a composite action space.

Parameters:

Name	Type	Description	Default
env	Env	The environment to wrap.	required

action(action_: Dict) -> Dict

Recursively clip the flattened, continuous actions in a composite action space.

Parameters:

Name	Type	Description	Default
action_	Dict	The action to clip.	required

Returns:

Name	Type	Description
action_	Dict	The clipped composite action.