odys.optimization.variable_owner¶

class VariableOwner(ABC, Generic[VariablesT]):
    """A part of the model that holds its decision variables in a typed, frozen variables object.

    The variables are created once, in `add_variables`, and belong to that one
    model: a second `add_variables` raises, so a problem is built into one model only.
    Entity formulations and objective terms with variables both use it. List it
    before the formulation base class, whose `add_variables` adds nothing.
    """

    _variables: VariablesT | None = None

    @property
    def variables(self) -> VariablesT:
        """Return the decision variables.

        Raises:
            OdysError: If `add_variables` has not run yet.
        """
        if self._variables is None:
            msg = f"{type(self).__name__}.add_variables must run before its variables are used."
            raise OdysError(msg)
        return self._variables

    def add_variables(self, model: linopy.Model) -> None:
        """Add the decision variables to the model.

        Args:
            model: The linopy model to add the variables to.

        Raises:
            OdysError: If the variables were already added to a model; build a new problem for each model.
        """
        if self._variables is not None:
            msg = f"{type(self).__name__} already has variables; build a new problem for each model."
            raise OdysError(msg)
        self._variables = self._create_variables(model)

    @abstractmethod
    def _create_variables(self, model: linopy.Model) -> VariablesT:
        """Add the decision variables to the model and return them as the typed variables object."""