bsparse.BCOO

class BCOO(rows, cols, data, bshape=None, dtype=None, sizes=None, symmetry=None)

A sparse matrix container in COOrdinate format.

The BCOO class represents a sparse matrix using three arrays:

• rows: contains the row coordinates of the non-zero elements.

• cols: contains the column coordinates of the non-zero elements.

• data: contains the values of the non-zero elements.

Upon creation, the matrix is sorted (lexicographically) by rows and columns.

Duplicate elements are not allowed.

Parameters:

rowsarray_like

The row coordinates of the non-zero elements.

colsarray_like

The column coordinates of the non-zero elements.

dataarray_like

The values of the non-zero constituents. This is an array of dtype object. Each element of the array can be either a 2D dense array, a scipy sparse matrix, or any bsparse matrix.

bshapetuple, optional

The shape of the matrix container. If not given, it is inferred from rows and cols.

dtypedtype, optional

The data type of the matrix elements. If not given, it is inferred from data.

sizestuple[np.ndarray, np.ndarray], optional

The sizes of the blocks. If not given, they are inferred from data.

symmetrystr, optional

The symmetry of the matrix. If not given, no symmetry is assumed. This is only applicable for square matrices, where possible values are 'symmetric' and 'hermitian'. Note that when setting a symmetry, the lower triangular part of the matrix is discarded.

Examples

>>> import numpy as np
>>> rows = [0, 1, 2, 1, 2, 2]
>>> cols = [0, 1, 2, 2, 0, 1]
>>> data = [i * np.ones((2, 2)) for i in range(1, 7)]
>>> bcoo = BCOO(rows, cols, data)
>>> bcoo
BCOO(bshape=(3, 3), bnnz=6 | shape=(6, 6), nnz=24)
>>> bcoo.toarray()
array([[1., 1., 0., 0., 0., 0.],
       [1., 1., 0., 0., 0., 0.],
       [0., 0., 2., 2., 4., 4.],
       [0., 0., 2., 2., 4., 4.],
       [5., 5., 6., 6., 3., 3.],
       [5., 5., 6., 6., 3., 3.]])

Attributes:

rowsndarray

The row coordinates of the non-zero elements.

colsndarray

The column coordinates of the non-zero elements.

datalist of ndarray, scipy.sparse, sparse.Sparse, or BSparse

The values of the non-zero elements.

shapetuple[int, int]

The shape of the equivalent dense matrix.

bshapetuple[int, int]

The block shape of the matrix.

row_sizesndarray

The sizes of the row elements.

col_sizesndarray

The sizes of the column elements.

dtypedtype

The data type of the matrix elements.

symmetrystr

The symmetry of the matrix.

nnzint

The number of non-zero elements in the matrix.

bnnzint

The number of non-zero elements in the matrix.

TBCOO

The transpose of the matrix.

HBCOO

The conjugate transpose of the matrix.

Methods

asbformat(format)	Converts the underlying matrix blocks to a given format.
astype(dtype)	Returns a copy of the matrix with a different data type.
bapply(func[, copy])	Applies a function to each matrix block.
conj()	The complex conjugate of the matrix.
conjugate()	The complex conjugate of the matrix.
copy()	Returns a copy of the matrix.
diagonal([offset])	Returns the block diagonal of the matrix.
from_array(arr, sizes[, symmetry])	Creates a BCOO matrix from a dense numpy.ndarray.
from_sparray(mat, sizes[, symmetry])	Creates a BCOO matrix from a scipy.sparse.sparray.
save_npz(filename)	Saves the matrix to a numpy.npz file.
toarray()	Converts the matrix to a dense numpy.ndarray.
tocoo()	Converts the matrix to BCOO format.
tocsr()	Converts the matrix to BCSR format.
todia()	Converts the matrix to BDIA format.