Reset and randomize the goal pose Args: env_ids (tensor): The index of the environment that needs to reset goal pose

Restore the model

Restore model parameters.

Reward function when self.reward_spare==False. Returns accumulative hit/shield point damage dealt to the enemy + reward_death_value pe

Sample random actions for warmup. Args: available_actions: (np.ndarray) denotes which actions are available to agent (if None, all

Sample from Gumbel(0, 1)

Save the model

Save model parameters.

Perform a training update for parameter-sharing MAPPO using minibatch GD. Args: actor_buffer: (list[OnPolicyActorBuffer]) buffer c

Tell all the environments to start taking a step with the given actions. Call step_wait() to get the results of the step.

Wait for the step taken with step_async(). Returns (obs, rews, dones, infos): - obs: an array of observations, or a dict of

Tile N images into one big PxQ image (P,Q) are chosen to be as close as possible, and if N is square, then P=Q. input: img_nhwc, list

Train the model

Train the model.

Perform a training update using minibatch GD. Args: actor_buffer: (OnPolicyActorBuffer) buffer containing training data related to

Perform a training update using minibatch GD. Args: critic_buffer: (OnPolicyCriticBufferEP or OnPolicyCriticBufferFP) buffer conta

Turn off the gradient for the critic.

Turn off gradient for critic.

Turn off the gradient for the critic network.

Returns the shooting range for an agent.

Update actor network. Args: sample: (Tuple) contains data batch with which to update networks. Returns: policy

Update actor network. Args: sample: (Tuple) contains data batch with which to update networks. Returns: policy

Update actor networks. Args: sample: (Tuple) contains data batch with which to update networks. Returns: kl: (

Auto-tune the temperature parameter alpha.

Update loaded config with unparsed command-line arguments. Args: unparsed_dict: (dict) Unparsed command-line arguments. *args: (li

Update current end flag for computing n-step return. End flag is True at the steps which are the end of an episode or the latest but unfinishe

Update current end flag for computing n-step return. End flag is True at the steps which are the end of an episode or the latest but unfinishe

Save factor for this actor.

Update units after an environment step. This function assumes that self._obs is up-to-date.

Warmup the replay buffer with random actions

Warm up the replay buffer.

Proportional Guidance Coefficient

Cross-Sectional area, unit m^2

Initialize the OffPolicyBaseRunner class. Args: args: command-line arguments parsed by argparse. Three keys: algo, env, exp_name.

Initialize the OnPolicyBaseRunner class. Args: args: command-line arguments parsed by argparse. Three keys: algo, env, exp_name.

envs: list of gym environments to run in subprocesses