gxm.Trajectory#
- class Trajectory(obs, true_obs, action, reward, terminated, truncated, info)#
Bases:
objectClass representing a trajectory \((S_0, A_0, R_0, S_1, ..., S_n)\) in an environment.
- __init__(obs, true_obs, action, reward, terminated, truncated, info)#
Methods
__init__(obs, true_obs, action, reward, ...)Attributes
Return whether the episode has ended (either terminated or truncated).
The observations \((S_0, S_1, ..., S_n)\) in the trajectory.
The true observations \((\hat{S}_0, \hat{S}_1, ..., \hat{S}_n)\) in the trajectory.
The actions \((A_0, A_1, ..., A_{n-1})\) taken in the trajectory.
The rewards \((R_0, R_1, ..., R_{n-1})\) received in the trajectory.
Whether the episode terminated at each timestep in the trajectory.
Whether the episode was truncated at each timestep in the trajectory.
Additional information about the trajectory.
-
action:
Any# The actions \((A_0, A_1, ..., A_{n-1})\) taken in the trajectory.
- property done: Array#
Return whether the episode has ended (either terminated or truncated).
-
info:
dict[str,Any]# Additional information about the trajectory.
-
obs:
Any# The observations \((S_0, S_1, ..., S_n)\) in the trajectory.
-
reward:
Array# The rewards \((R_0, R_1, ..., R_{n-1})\) received in the trajectory.
-
terminated:
Array# Whether the episode terminated at each timestep in the trajectory.
-
true_obs:
Any# The true observations \((\hat{S}_0, \hat{S}_1, ..., \hat{S}_n)\) in the trajectory. These may differ from
obsin environments that allow truncation.
-
truncated:
Array# Whether the episode was truncated at each timestep in the trajectory.