pymasep.common.environment

Classes

Environment

Represents the environment of the game. This object keeps all dynamic value of the game.

Module Contents

class pymasep.common.environment.Environment(id_env=None, cfg=None)

Represents the environment of the game. This object keeps all dynamic value of the game.

Parameters:

  • id_env (str) – the id of the environment. If None, a new uuid will be set

  • cfg (omegaconf.DictConfig)

Parm cfg:

the configuration of the environment

id: str

Environment id.

config

configuration of the environment.

end_episode: bool = False

the current episode is finished

current_step: int = 0

current step of the environment

current_episode: int = 0

current episode of the environment

current_state: pymasep.common.state.State | None = None

current state of the environment

current_additional_info: dict

current additional information of the environment, related to the current state. Sort of meta information on the current state

agents: Set[pymasep.common.agent.Agent]

set of agents in the current state

controllers: Dict[str, pymasep.common.controller.Controller]

set of controllers that can be used in the environment

_game: pymasep.common.game.Game | None = None
next_action: Dict[str, pymasep.common.action.Action]

next action to be executed in the current state

_level_logger
logger

logger for the environment tasks

__del__()
property game: pymasep.common.game.Game

Game of the environment.

Return type:

pymasep.common.game.Game

get_current_coord_method(state=None)

Return the current coordination method according to the current game phase in the state

Parameters:

state (Optional[pymasep.common.state.State]) – State to check the coordination method. If None, take the current state of the environment.

Returns:

The current coordinate method

Return type:

str

get_observation_for_agent(agent)

Get the current observation. Current state at the moment

Parameters:

agent (pymasep.common.agent.Agent) – The agent to get the observation for

Returns:

The current observation

Return type:

pymasep.common.state.State

get_additional_info_for_agent()

Get the current additional information. Could be filtered in the future (by agent)

Returns:

The current additional information

Return type:

dict

get_reward_for_agent(agent)

Get the current reward for an agent.

Parameters:

agent (pymasep.common.agent.Agent) – The agent to get the reward

Returns:

The current reward of the agent

Return type:

dict

get_controller(agent_name, class_controller_)

Get the controller of an agent. If the controller is not created, create it. (Maybe I should change the name of this method…)

Parameters:

  • agent_name (str) – The agent’s name

  • class_controller – the class of the controller to create if it is not created

  • class_controller_ (Type[pymasep.common.controller.Controller])

Returns:

the instance of the agent’s controller

Return type:

pymasep.common.controller.Controller

calculate_end_episode()

Calculate the end of the episode. Modify self.end_episode if the conditions of the episode’s end are met.

Return type:

None

calculate_next_state()

Calculate the next state of the environment. Change the current state. Increase the step

Return type:

None

calculate_reward()

Calculate the reward from the current_state and add the result to additional info with key ‘reward’. For MDP point of view, support only (at the moment) \(R(s)\) and neither \(R(s,a)\) nor \(R(s,a,s')\)

Return type:

None

create_base_object(name=None, template=None, xml_node=None, parent=None)

Create a base object from a string represent the type

Parameters:

  • name (Optional[str]) – The identification name of the BaseObject.

  • template (Optional[pymasep.common.template.Template]) – Template of the created object. If None and no XML is specified, will be the default template for this bo_type

  • xml_node (Optional[xml.etree.ElementTree.Element]) – XML node used for creating the base object

  • parent (Optional[pymasep.common.base_object.BaseObject]) – the parent of the current object. May be None.

Returns:

The BaseObject instance

get_type_controller(xml_node=None, template=None, control=None)

Get the type of controller according to a template or (exclusive) an XML node.

Parameters:

  • xml_node (Optional[xml.etree.ElementTree.Element]) – XML node used to get the controller type. Use the attribute “controller”.

  • template (Optional[pymasep.common.template.Template]) – The template used to get the controller type.

  • control (Optional[str]) – ‘Interface’ if the agent is controlled by interface (only inside engine, and if template is used)

Returns:

the class of the controller to use

Return type:

Type[pymasep.common.controller.Controller]

create_object(name=None, template=None, xml_node=None, parent=None, control=None)

Create an object. Add it into the object dictionary (if not already exists) and agent set if the object is Agent

Parameters:

  • name (Optional[str]) – The identification name of the object. Not used if xml_node is used

  • template (Optional[pymasep.common.template.Template]) – the template of the class to create the object. Not used if xml_node is used

  • xml_node (Optional[xml.etree.ElementTree.Element]) – XML node describing the object to create

  • parent (Optional[pymasep.common.base_object.BaseObject]) – the parent of the current object. Maybe None

  • control (Optional[str]) – ‘interface’ if the agent is controlled by interface (only inside engine and if template is used)

Returns:

the object instance

create_action(action_type=None, xml_node=None)

Create an action

Parameters:

  • action_type (Optional[int]) – The type of the action (Constants in Action class)

  • xml_node (xml.etree.ElementTree.Element) – XML to create Action

Returns:

The Action created from XML or with just a type

Return type:

pymasep.common.action.Action

choose_action()

Tell all agents to choose their action

Return type:

None

clear_all_state_elements()

Clear all in the environment

Return type:

None