Environments

EnvironmentInfo

class nnabla_rl.environments.environment_info.EnvironmentInfo(observation_space, action_space, max_episode_steps, unwrapped_env, reward_function: Callable[[Any, Any, Dict], int] | None = None)

Environment Information class.

This class contains the basic information of the target training environment.

property action_dim

The dimension of action assuming that the action is flatten.

property action_high

The upper limit of action space.

property action_low

The lower limit of action space.

property action_shape

The shape of action space.

static from_env(env)

Create env_info from environment.

Parameters:

env (gym.Env) – the environment

Returns:

EnvironmentInfo (EnvironmentInfo)

Example

>>> import gym
>>> from nnabla_rl.environments.environment_info import EnvironmentInfo
>>> env = gym.make("CartPole-v0")
>>> env_info = EnvironmentInfo.from_env(env)
>>> env_info.state_shape
(4,)

is_continuous_action_env()

Check whether the action to execute in the environment is continuous or not.

Returns:

True if the action to execute in the environment is continuous. Otherwise False.

Note that if the action is gym.spaces.Tuple and all of the element are continuous, it returns True.

Return type:

bool

is_continuous_state_env()

Check whether the state of the environment is continuous or not.

Returns:

True if the state of the environment is continuous. Otherwise False.

Note that if the state is gym.spaces.Tuple and all of the element are continuous, it returns True.

Return type:

bool

is_discrete_action_env()

Check whether the action to execute in the environment is discrete or not.

Returns:

True if the action to execute in the environment is discrete. Otherwise False.

Note that if the action is gym.spaces.Tuple and all of the element are discrete, it returns True.

Return type:

bool

is_discrete_state_env()

Check whether the state of the environment is discrete or not.

Returns:

True if the state of the environment is discrete. Otherwise False.

Note that if the state is gym.spaces.Tuple and all of the element are discrete, it returns True.

Return type:

bool

is_goal_conditioned_env()

Check whether the environment is gym.GoalEnv or not.

Returns:

True if the environment is GoalEnv. Otherwise False.

Return type:

bool

is_mixed_action_env()

Check whether the action of the environment consists of either continuous or discrete action.

Returns:

True if the action of the environment is either continuous or discrete. Otherwise False.

Note that if the action is not a gym.spaces.Tuple, then returns False.

Return type:

bool

is_mixed_state_env()

Check whether the state of the environment consists of either continuous or discrete state.

Returns:

True if the state of the environment is either continuous or discrete. Otherwise False.

Note that if the state is not a gym.spaces.Tuple, then returns False.

Return type:

bool

is_tuple_action_env()

Check whether the action of the environment is tuple or not.

Returns:

True if the action of the environment is tuple. Otherwise False.

Return type:

bool

is_tuple_state_env()

Check whether the state of the environment is tuple or not.

Returns:

True if the state of the environment is tuple. Otherwise False.

Return type:

bool

property state_dim

The dimension of state assuming that the state is flatten.

property state_high

The upper limit of observation space.

property state_low

The lower limit of observation space.

property state_shape

The shape of observation space.