Source code for compas.datastructures.network.matrices


from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

from compas.numerical import adjacency_matrix
from compas.numerical import degree_matrix
from compas.numerical import connectivity_matrix
from compas.numerical import laplacian_matrix


__all__ = [
    'network_adjacency_matrix',
    'network_degree_matrix',
    'network_connectivity_matrix',
    'network_laplacian_matrix',
]


def _return_matrix(M, rtype):
    if rtype == 'list':
        return M.toarray().tolist()
    if rtype == 'array':
        return M.toarray()
    if rtype == 'csr':
        return M.tocsr()
    if rtype == 'csc':
        return M.tocsc()
    if rtype == 'coo':
        return M.tocoo()
    return M


def network_adjacency_matrix(network, rtype='array'):
    """Creates a node adjacency matrix from a Network datastructure.

    Parameters
    ----------
    network : obj
        Network datastructure object to get data from.
    rtype : {'array', 'csc', 'csr', 'coo', 'list'}
        Format of the result.

    Returns
    -------
    array-like
        Constructed adjacency matrix.

    """
    key_index = network.key_index()
    adjacency = [[key_index[nbr] for nbr in network.neighbors(key)] for key in network.nodes()]
    return adjacency_matrix(adjacency, rtype=rtype)


def network_degree_matrix(network, rtype='array'):
    """Creates a node degree matrix from a Network datastructure.

    Parameters
    ----------
    network : obj
        Network datastructure object to get data from.
    rtype : {'array', 'csc', 'csr', 'coo', 'list'}
        Format of the result.

    Returns
    -------
    array-like
        Constructed node degree matrix.

    """
    key_index = network.key_index()
    adjacency = [[key_index[nbr] for nbr in network.neighbors(key)] for key in network.nodes()]
    return degree_matrix(adjacency, rtype=rtype)


def network_connectivity_matrix(network, rtype='array'):
    """Creates a connectivity matrix from a Network datastructure.

    Parameters
    ----------
    network : obj
        Network datastructure object to get data from.
    rtype : {'array', 'csc', 'csr', 'coo', 'list'}
        Format of the result.

    Returns
    -------
    array-like
        Constructed connectivity matrix.

    """
    key_index = network.key_index()
    edges = [(key_index[u], key_index[v]) for u, v in network.edges()]
    return connectivity_matrix(edges, rtype=rtype)


def network_laplacian_matrix(network, normalize=False, rtype='array'):
    r"""Construct a Laplacian matrix from a Network datastructure.

    Parameters
    ----------
    network : obj
        Network datastructure object to get data from.
    normalize : bool
        Normalize the entries such that the value on the diagonal is ``1``.
    rtype : {'array', 'csc', 'csr', 'coo', 'list'}
        Format of the result.

    Returns
    -------
    array-like
        Constructed Laplacian matrix.

    Notes
    -----
    ``d = L.dot(xyz)`` is currently a vector that points from the centroid to the node.
    Therefore ``c = xyz - d``. By changing the signs in the laplacian, the dsiplacement
    vectors could be used in a more natural way ``c = xyz + d``.

    Examples
    --------
    >>>

    """
    key_index = network.key_index()
    edges = [(key_index[u], key_index[v]) for u, v in network.edges()]
    return laplacian_matrix(edges, normalize=normalize, rtype=rtype)