Convolution

degree_matrix

spektral.utils.degree_matrix(A)


Computes the degree matrix of the given adjacency matrix.

Arguments

• A: rank 2 array or sparse matrix.

Return
If A is a dense array, a dense array; if A is sparse, a sparse matrix in DIA format.

degree_power

spektral.utils.degree_power(A, k)


Computes $\D^{k}$ from the given adjacency matrix. Useful for computing normalised Laplacian.

Arguments

• A: rank 2 array or sparse matrix.

• k: exponent to which elevate the degree matrix.

Return
If A is a dense array, a dense array; if A is sparse, a sparse matrix in DIA format.

spektral.utils.normalized_adjacency(A, symmetric=True)


Normalizes the given adjacency matrix using the degree matrix as either $\D^{-1}\A$ or $\D^{-1/2}\A\D^{-1/2}$ (symmetric normalization).

Arguments

• A: rank 2 array or sparse matrix;

• symmetric: boolean, compute symmetric normalization;

Return

laplacian

spektral.utils.laplacian(A)


Computes the Laplacian of the given adjacency matrix as $\D - \A$.

Arguments

• A: rank 2 array or sparse matrix;

Return
The Laplacian.

normalized_laplacian

spektral.utils.normalized_laplacian(A, symmetric=True)


Computes a normalized Laplacian of the given adjacency matrix as $\I - \D^{-1}\A$ or $\I - \D^{-1/2}\A\D^{-1/2}$ (symmetric normalization).

Arguments

• A: rank 2 array or sparse matrix;

• symmetric: boolean, compute symmetric normalization;

Return
The normalized Laplacian.

rescale_laplacian

spektral.utils.rescale_laplacian(L, lmax=None)


Rescales the Laplacian eigenvalues in [-1,1], using lmax as largest eigenvalue.

Arguments

• L: rank 2 array or sparse matrix;

• lmax: if None, compute largest eigenvalue with scipy.linalg.eisgh. If the eigendecomposition fails, lmax is set to 2 automatically. If scalar, use this value as largest eigenvalue when rescaling.

Return

spektral.utils.add_self_loops(a, value=1)


Sets the inner diagonals of a to value.

Arguments

• a: a np.array or scipy.sparse matrix, the innermost two dimensions must be equal.

• value: value to set the diagonals to.

Return
A np.array or scipy.sparse matrix with the same shape as a.

gcn_filter

spektral.utils.gcn_filter(A, symmetric=True)


Computes the graph filter described in Kipf & Welling (2017).

Arguments

• A: array or sparse matrix with rank 2 or 3;

• symmetric: boolean, whether to normalize the matrix as $\D^{-\frac{1}{2}}\A\D^{-\frac{1}{2}}$ or as $\D^{-1}\A$;

Return
Array or sparse matrix with rank 2 or 3, same as A;

chebyshev_polynomial

spektral.utils.chebyshev_polynomial(X, k)


Calculates Chebyshev polynomials of X, up to order k.

Arguments

• X: rank 2 array or sparse matrix;

• k: the order up to which compute the polynomials,

Return
A list of k + 1 arrays or sparse matrices with one element for each degree of the polynomial.

chebyshev_filter

spektral.utils.chebyshev_filter(A, k, symmetric=True)


Computes the Chebyshev filter from the given adjacency matrix, as described in Defferrard et at. (2016).

Arguments

• A: rank 2 array or sparse matrix;

• k: integer, the order of the Chebyshev polynomial;

• symmetric: boolean, whether to normalize the matrix as $\D^{-\frac{1}{2}}\A\D^{-\frac{1}{2}}$ or as $\D^{-1}\A$;

Return
A list of k + 1 arrays or sparse matrices with one element for each degree of the polynomial.