import abc from traits.api import ABCHasStrictTraits, Instance from .i_mco_factory import IMCOFactory class BaseMCOCommunicator(ABCHasStrictTraits): """Communicator class that defines how the MCO communicates with the evaluator program that does the actual heavylifting of the calculation. The model assumes that the MCO spawns a process to perform the evaluation, and passes data to this process that define the parameters for the evaluation. Once completed, the evaluation will return a set of results, that we interpret as KPIs. These KPIs are encoded in some form, which is again specified by the MCO. """ #: A reference to the factory factory = Instance(IMCOFactory) def __init__(self, factory, *args, **kwargs): self.factory = factory super(BaseMCOCommunicator, self).__init__(*args, **kwargs) @abc.abstractmethod def receive_from_mco(self, model): """ Receives the parameters from the MCO. The conversion is specific to the format of the communication between the MCO and its evaluator program. Must return a list of DataValue objects, containing the data passed by the MCO. Parameters ---------- model: BaseMCOModel The model of the optimizer, instantiated through create_model() Returns ------- List(DataValue) A list of the DataValues with the appropriate information filled in """ @abc.abstractmethod def send_to_mco(self, model, kpi_results): """Send the KPI results from the evaluation to the MCO Must be reimplemented to perform the conversion between the two formats. This is of course dependent on the specifics of the MCO and how it interacts with the external evaluator program. Parameters ---------- model: BaseMCOModel The model of the optimizer, instantiated through create_model() kpi_results: List(DataValue) A list of KPI calculator results, one per each KPI calculator. """