Class VarConstraint

maraboupy/MarabouPythonic.py:202–217

Define constraints on neurons of a neural network. Assuming neuron x, constants a b, and an affine transformation a * x + b, then example constraints are a * x + b <= or >= or == a' * x' + b'. Attributes: combination: an instance of the VarAffine class, i.e., affine transfo

@dataclass()
class VarConstraint:
    """
    Define constraints on neurons of a neural network.
    Assuming neuron x, constants a b, and an affine transformation a * x + b,
    then example constraints are a * x + b <= or >= or == a' * x' + b&#x27;.

    Attributes:
        combination: an instance of the VarAffine class, i.e., affine transformation of neuron(s)
        isEquality: whether it is an equality or inequality, i.e., True if ==; False if >= or <=.
        lowerBound: whether it is a lower bound, e.g., True if x + b >= b&#x27;.
        upperBound: whether it is an upper bound, e.g., True if x + b <= b&#x27;.
    """
    combination: VarAffine
    isEquality: bool
    lowerBound: Optional[float]
    upperBound: Optional[float]


def Var(index):