agentlib_mpc.modules.dmpc package

Subpackages

• agentlib_mpc.modules.dmpc.admm package
  • Submodules
  • agentlib_mpc.modules.dmpc.admm.admm module
    • ADMM
      • ADMM.admm_step()
      • ADMM.all_coupling_statuses()
      • ADMM.assert_mpc_variables_are_in_model()
      • ADMM.collect_couplings_for_optimization()
      • ADMM.collect_variables_for_optimization()
      • ADMM.cons_and_exchange
      • ADMM.deregister_all_participants()
      • ADMM.get_participants_values()
      • ADMM.get_results()
      • ADMM.participant_callback()
      • ADMM.penalty_factor_var
      • ADMM.process()
      • ADMM.receive_participant()
      • ADMM.registered_participants
      • ADMM.reset_participants_ready()
      • ADMM.send_coupling_values()
      • ADMM.send_coupling_variable()
      • ADMM.update_lambda()
      • ADMM.var_ref
    • ADMMConfig
      • ADMMConfig.couplings
      • ADMMConfig.exchange
      • ADMMConfig.iteration_timeout
      • ADMMConfig.max_iterations
      • ADMMConfig.penalty_factor
      • ADMMConfig.registration_period
      • ADMMConfig.check_prefixes_of_variables
      • ADMMConfig.model_post_init()
    • ADMMParticipation
      • ADMMParticipation.de_register()
      • ADMMParticipation.empty_memory()
    • LocalADMM
      • LocalADMM.process()
      • LocalADMM.registration_delay
      • LocalADMM.sync_delay
    • LocalADMMConfig
      • LocalADMMConfig.registration_delay
      • LocalADMMConfig.sync_delay
      • LocalADMMConfig.model_post_init()
    • ModuleStatus
      • ModuleStatus.at_registration
      • ModuleStatus.not_started
      • ModuleStatus.optimizing
      • ModuleStatus.sleeping
      • ModuleStatus.syncing
      • ModuleStatus.updating
      • ModuleStatus.waiting_for_other_agents
    • ParticipantStatus
      • ParticipantStatus.available
      • ParticipantStatus.confirmed
      • ParticipantStatus.not_available
      • ParticipantStatus.not_participating
  • agentlib_mpc.modules.dmpc.admm.admm_coordinated module
    • CoordinatedADMM
      • CoordinatedADMM.get_new_measurement()
      • CoordinatedADMM.init_iteration_callback()
      • CoordinatedADMM.optimize()
      • CoordinatedADMM.process()
      • CoordinatedADMM.registration_callback()
    • CoordinatedADMMConfig
      • CoordinatedADMMConfig.shared_variable_fields
      • CoordinatedADMMConfig.couplings_should_have_values
      • CoordinatedADMMConfig.model_post_init()
    • coupInput
  • agentlib_mpc.modules.dmpc.admm.admm_coordinator module
    • ADMMCoordinator
      • ADMMCoordinator.cleanup_results()
      • ADMMCoordinator.get_results()
      • ADMMCoordinator.optim_results_callback()
      • ADMMCoordinator.register_agent()
      • ADMMCoordinator.registration_callback()
      • ADMMCoordinator.trigger_optimizations()
    • ADMMCoordinatorConfig
      • ADMMCoordinatorConfig.abs_tol
      • ADMMCoordinatorConfig.admm_iter_max
      • ADMMCoordinatorConfig.dual_tol
      • ADMMCoordinatorConfig.penalty_change_factor
      • ADMMCoordinatorConfig.penalty_change_threshold
      • ADMMCoordinatorConfig.penalty_factor
      • ADMMCoordinatorConfig.prediction_horizon
      • ADMMCoordinatorConfig.primal_tol
      • ADMMCoordinatorConfig.registration_period
      • ADMMCoordinatorConfig.rel_tol
      • ADMMCoordinatorConfig.sampling_time
      • ADMMCoordinatorConfig.save_iter_interval
      • ADMMCoordinatorConfig.save_solve_stats
      • ADMMCoordinatorConfig.solve_stats_file
      • ADMMCoordinatorConfig.time_step
      • ADMMCoordinatorConfig.use_relative_tolerances
      • ADMMCoordinatorConfig.wait_time_on_start_iters
      • ADMMCoordinatorConfig.default_sampling_time
      • ADMMCoordinatorConfig.model_post_init()
      • ADMMCoordinatorConfig.solve_stats_file_is_csv
  • Module contents

Submodules

agentlib_mpc.modules.dmpc.coordinator module

class agentlib_mpc.modules.dmpc.coordinator.Coordinator(*, config: dict, agent: Agent)

Bases: BaseModule

Class implementing the base coordination for distributed MPC

property all_finished

Returns: True, if there are no busy agents, else False

init_iteration_callback(variable: AgentVariable)

Processes and Agents InitIteration confirmation. :param variable:

Returns:

optim_results_callback(variable: AgentVariable)

Saves the results of a local optimization. :param variable:

Returns:

process()

This abstract method must be implemented in order to sync the module with the other processes of the agent and the whole MAS.

register_callbacks()

registration_callback(variable: AgentVariable)

trigger_optimizations()

Triggers the optimization for all agents with status ready. Returns:

pydantic model agentlib_mpc.modules.dmpc.coordinator.CoordinatorConfig

Bases: BaseModuleConfig

Config:

• arbitrary_types_allowed: bool = True

• validate_assignment: bool = True

• extra: str = forbid

• frozen: bool = True

Fields:

• maxIter (int)

• messages_in (List[agentlib.core.datamodels.AgentVariable])

• messages_out (List[agentlib.core.datamodels.AgentVariable])

• shared_variable_fields (list[str])

• time_out_non_responders (float)

Validators:

• check_valid_fields » shared_variable_fields

field maxIter: int = 10

Maximum number of iterations

field messages_in: List[AgentVariable] = [AgentVariable(name='registration_agent_to_coordinator', type=None, timestamp=None, unit='Not defined', description='Not defined', ub=inf, lb=-inf, clip=False, allowed_values=[], value=None, alias='registration_agent_to_coordinator', source=Source(agent_id=None, module_id=None), shared=None, rdf_class=None), AgentVariable(name='startIteration_agent_to_coordinator', type=None, timestamp=None, unit='Not defined', description='Not defined', ub=inf, lb=-inf, clip=False, allowed_values=[], value=None, alias='startIteration_agent_to_coordinator', source=Source(agent_id=None, module_id=None), shared=None, rdf_class=None), AgentVariable(name='optimization_agent_to_coordinator', type=None, timestamp=None, unit='Not defined', description='Not defined', ub=inf, lb=-inf, clip=False, allowed_values=[], value=None, alias='optimization_agent_to_coordinator', source=Source(agent_id=None, module_id=None), shared=None, rdf_class=None)]

field messages_out: List[AgentVariable] = [AgentVariable(name='registration_coordinator_to_agent', type=None, timestamp=None, unit='Not defined', description='Not defined', ub=inf, lb=-inf, clip=False, allowed_values=[], value=None, alias='registration_coordinator_to_agent', source=Source(agent_id=None, module_id=None), shared=None, rdf_class=None), AgentVariable(name='startIteration_coordinator_to_agent', type=None, timestamp=None, unit='Not defined', description='Not defined', ub=inf, lb=-inf, clip=False, allowed_values=[], value=None, alias='startIteration_coordinator_to_agent', source=Source(agent_id=None, module_id=None), shared=None, rdf_class=None), AgentVariable(name='optimization_coordinator_to_agent', type=None, timestamp=None, unit='Not defined', description='Not defined', ub=inf, lb=-inf, clip=False, allowed_values=[], value=None, alias='optimization_coordinator_to_agent', source=Source(agent_id=None, module_id=None), shared=None, rdf_class=None)]

field shared_variable_fields: list[str] = ['messages_out']

Validated by:

• check_valid_fields

field time_out_non_responders: float = 1

Maximum wait time for subsystems in seconds

model_post_init(context: Any, /) → None

We need to both initialize private attributes and call the user-defined model_post_init method.

agentlib_mpc.modules.dmpc.employee module

class agentlib_mpc.modules.dmpc.employee.MiniEmployee(*, config: dict, agent: Agent)

Bases: BaseModule

get_new_measurement()

Retrieve new measurement from relevant sensors Returns:

init_iteration_callback(variable: AgentVariable)

Callback that processes the coordinators ‘startIteration’ flag. :param variable:

abstract optimize(variable: AgentVariable)

Performs the optimization given the information from the coordinator. Replies with local information. Returns:

pre_computation_hook()

This method is called in every computation step before the optimization starts. Overwrite this method in a derived subclass if you want to take some actions each time before the optimal control problem is solved.

process()

This abstract method must be implemented in order to sync the module with the other processes of the agent and the whole MAS.

register_callbacks()

abstract registration_callback(variable: AgentVariable)

callback for registration

shift_trajectories()

Shifts algorithm specific trajectories. Returns:

pydantic model agentlib_mpc.modules.dmpc.employee.MiniEmployeeConfig

Bases: BaseModuleConfig

Config:

• arbitrary_types_allowed: bool = True

• validate_assignment: bool = True

• extra: str = forbid

• frozen: bool = True

Fields:

• coordinator (agentlib.core.datamodels.Source)

• messages_in (List[agentlib.core.datamodels.AgentVariable])

• messages_out (List[agentlib.core.datamodels.AgentVariable])

• registration_interval (float)

• request_frequency (float)

• shared_variable_fields (list[str])

Validators:

• check_valid_fields » shared_variable_fields

field coordinator: Source [Required]

Define the agents coordinator

field messages_in: List[AgentVariable] = [AgentVariable(name='registration_coordinator_to_agent', type=None, timestamp=None, unit='Not defined', description='Not defined', ub=inf, lb=-inf, clip=False, allowed_values=[], value=None, alias='registration_coordinator_to_agent', source=Source(agent_id=None, module_id=None), shared=None, rdf_class=None), AgentVariable(name='startIteration_coordinator_to_agent', type=None, timestamp=None, unit='Not defined', description='Not defined', ub=inf, lb=-inf, clip=False, allowed_values=[], value=None, alias='startIteration_coordinator_to_agent', source=Source(agent_id=None, module_id=None), shared=None, rdf_class=None), AgentVariable(name='optimization_coordinator_to_agent', type=None, timestamp=None, unit='Not defined', description='Not defined', ub=inf, lb=-inf, clip=False, allowed_values=[], value=None, alias='optimization_coordinator_to_agent', source=Source(agent_id=None, module_id=None), shared=None, rdf_class=None)]

field messages_out: List[AgentVariable] = [AgentVariable(name='registration_agent_to_coordinator', type=None, timestamp=None, unit='Not defined', description='Not defined', ub=inf, lb=-inf, clip=False, allowed_values=[], value=None, alias='registration_agent_to_coordinator', source=Source(agent_id=None, module_id=None), shared=None, rdf_class=None), AgentVariable(name='startIteration_agent_to_coordinator', type=None, timestamp=None, unit='Not defined', description='Not defined', ub=inf, lb=-inf, clip=False, allowed_values=[], value=None, alias='startIteration_agent_to_coordinator', source=Source(agent_id=None, module_id=None), shared=None, rdf_class=None), AgentVariable(name='optimization_agent_to_coordinator', type=None, timestamp=None, unit='Not defined', description='Not defined', ub=inf, lb=-inf, clip=False, allowed_values=[], value=None, alias='optimization_agent_to_coordinator', source=Source(agent_id=None, module_id=None), shared=None, rdf_class=None)]

field registration_interval: float = 10

Interval in seconds after which a registration attempt is made.

Constraints:

• ge = 0

field request_frequency: float = 1

Wait time between signup_requests

field shared_variable_fields: list[str] = ['messages_out']

Validated by:

• check_valid_fields

model_post_init(context: Any, /) → None

We need to both initialize private attributes and call the user-defined model_post_init method.

Module contents

class agentlib_mpc.modules.dmpc.DistributedMPC(config: dict, agent: Agent)

Bases: MPC

Base class which defines common interfaces among all distributed mpc approaches (either optimization based, game theory based or some other).

pydantic model agentlib_mpc.modules.dmpc.DistributedMPCConfig

Bases: MPCConfig

Base config class with common configurations

Config:

• arbitrary_types_allowed: bool = True

• validate_assignment: bool = True

• extra: str = forbid

• frozen: bool = True

Fields:

Validators:

model_post_init(context: Any, /) → None

We need to both initialize private attributes and call the user-defined model_post_init method.