graph2mat.core.data.sparse

Conversion between different sparse representations.

Different sparse representations of a matrix are required during the different steps of a typical workflow using graph2mat.

Functions

block_dict_to_coo(block_dict, first_orb[, ...])	Converts a block dictionary into a coo array.
csr_to_block_dict(spmat, atoms, nsc[, ...])	Creates a BasisMatrix object from a SparseCSR matrix
csr_to_sisl_sparse_orbital(csr, geometry[, ...])	Converts a scipy CSR array to a sisl sparse orbital matrix.
nodes_and_edges_to_coo(node_vals, edge_vals, ...)	Converts an orbital matrix from node and edges array to scipy coo.
nodes_and_edges_to_sparse_orbital(node_vals, ...)
register_all_sisl(source, target, converter, ...)

graph2mat.core.data.sparse.block_dict_to_coo(block_dict: Dict[Tuple[int, int, int], ndarray], first_orb: ndarray, n_supercells: int = 1, threshold: float = 1e-08) → coo_array

Converts a block dictionary into a coo array.

Conversions to any other sparse structure can be done once we’ve got the coo array.

graph2mat.core.data.sparse.csr_to_block_dict(spmat: SparseCSR, atoms: Atoms, nsc: ndarray, geometry_atoms: Atoms | None = None, matrix_cls: Type[BasisMatrix] = graph2mat.core.data.matrices.physics.orbital_matrix.OrbitalMatrix) → BasisMatrix

Creates a BasisMatrix object from a SparseCSR matrix

Parameters:

• spmat – The sparse matrix to convert to a block dictionary.

• atoms – The atoms object for the matrix, containing orbital information.

• nsc – The auxiliary supercell size.

• matrix_cls – Matrix class to initialize.

• geometry_atoms – The atoms object for the full geometry. This allows the matrix to contain atoms without any orbital. Geometry atoms should contain the matrix atoms first and then the orbital-less atoms.

graph2mat.core.data.sparse.csr_to_sisl_sparse_orbital(csr: csr_array, geometry: Geometry, sp_class: Type[SparseOrbital] = sisl.SparseOrbital) → SparseOrbital

Converts a scipy CSR array to a sisl sparse orbital matrix.

graph2mat.core.data.sparse.nodes_and_edges_to_coo(node_vals: ndarray, edge_vals: ndarray, edge_index: ndarray, orbitals: ndarray, n_supercells: int = 1, edge_neigh_isc: ndarray | None = None, threshold: float | None = None, symmetrize_edges: bool = False) → coo_array

Converts an orbital matrix from node and edges array to scipy coo.

Conversions to any other sparse structure can be done once we’ve got the coo array.

Parameters:

• node_vals – Flat array containing the values of the node blocks. The order of the values is first by node index, then row then column.

• edge_vals – Flat array containing the values of the edge blocks. The order of the values is first by edge index, then row then column.

• edge_index – Array of shape (2, n_edges) containing the indices of the atoms that participate in each edge.

• orbitals – Array of shape (n_nodes, ) containing the number of orbitals for each atom.

• n_supercells – Number of auxiliary supercells.

• edge_neigh_isc – Array of shape (n_edges, ) containing the supercell index of the second atom in each edge with respect to the first atom. If not provided, all interactions are assumed to be in the unit cell.

• threshold – Matrix elements with a value below this number are set to 0.

• symmetrize_edges – whether for each edge only one direction is provided. The edge block for the opposite direction is then created as the transpose.

graph2mat.core.data.sparse.nodes_and_edges_to_sparse_orbital(node_vals: ndarray, edge_vals: ndarray, edge_index: ndarray, geometry: Geometry, sp_class: Type[SparseOrbital] = sisl.SparseOrbital, edge_neigh_isc: ndarray | None = None, threshold: float = 1e-08, symmetrize_edges: bool = False) → SparseOrbital

graph2mat.core.data.sparse.register_all_sisl(source, target, converter, manager)