DAGMA
dagma.linear.DagmaLinear._h
Initializing search
    kevinsbello/dagma
    • Home
    • Getting Started
    • API Reference
    kevinsbello/dagma
      • Home
      • Getting Started
      • dagma
        • dagma.linear
          • DagmaLinear
            • _score
            • _h
            • _func
            • _adam_update
            • minimize
            • fit
        • dagma.locally_connected
          • LocallyConnected
            • reset_parameters
            • forward
            • extra_repr
        • dagma.nonlinear
          • DagmaMLP
            • forward
            • h_func
            • fc1_l1_reg
            • fc1_to_adj
          • DagmaNonlinear
            • log_mse_loss
            • minimize
            • fit
        • dagma.utils
          • set_random_seed
          • is_dag
          • simulate_dag
          • simulate_parameter
          • simulate_linear_sem
          • simulate_nonlinear_sem
          • count_accuracy

    dagma.linear.DagmaLinear._h¶

    dagma.linear.DagmaLinear._h(W: numpy.ndarray, s: float = 1.0) → tuple[float, numpy.ndarray]¶

    Evaluate value and gradient of the logdet acyclicity constraint.

    Parameters:¶
    W : np.ndarray¶

    \((d,d)\) adjacency matrix

    s : float, optional¶

    Controls the domain of M-matrices. Defaults to 1.0.

    Returns:¶

    h value, and gradient of h

    Return type:¶

    Tuple[float, np.ndarray]


    Last update: Jan 14, 2024
    Back to top
    Previous dagma.linear.DagmaLinear._score
    Next dagma.linear.DagmaLinear._func
    © Copyright 2023, Kevin Bello.
    Created using Sphinx 7.1.2. and Sphinx-Immaterial