swarmrl.agents.classical_agent Module API Reference¶
Class for classical agents. These are agents not controlled by machine learning. They should also not be trainable.
ClassicalAgent
¶
Bases: Agent
Class to handle the actor-critic RL Protocol.
Source code in swarmrl/agents/classical_agent.py
15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 |
|
__init__(particle_type, actions, task=None, observable=None)
¶
Constructor for the actor-critic protocol.
Parameters¶
network : Network Shared Actor-Critic Network for the RL protocol. The apply function should return a tuple of (logits, value). particle_type : int Particle ID this RL protocol applies to. observable : Observable Observable for this particle type and network input task : Task Task for this particle type to perform. actions : dict Actions allowed for the particle.
Source code in swarmrl/agents/classical_agent.py
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 |
|
calc_action(colloids)
¶
Copmute the new state for the agent.
Returns the chosen actions to the force function which talks to the espresso engine.
Parameters¶
colloids : List[Colloid] List of colloids in the system.
Source code in swarmrl/agents/classical_agent.py
49 50 51 52 53 54 55 56 57 58 59 60 61 |
|