"""Random sparse matrices."""
# pylint: disable=redefined-builtin
import numpy as np
import scipy as sp
def _rand_sparse(m, n, density):
"""Construct base function for sprand, sprandn."""
nnz = max(min(int(m*n*density), m*n), 0)
row = np.random.randint(low=0, high=m-1, size=nnz)
col = np.random.randint(low=0, high=n-1, size=nnz)
data = np.ones(nnz, dtype=float)
# duplicate (i,j) entries will be summed together
return sp.sparse.csr_matrix((data, (row, col)), shape=(m, n))
[docs]
def sprand(m, n, density, format='csr'):
"""Return a random sparse matrix.
Parameters
----------
m, n : int
shape of the result
density : float
target a matrix with nnz(A) = m*n*density, 0<=density<=1
format : string
sparse matrix format to return, e.g. 'csr', 'coo', etc.
Returns
-------
A : sparse matrix
m x n sparse matrix
Examples
--------
>>> from pyamg.gallery import sprand
>>> A = sprand(5,5,3/5.0)
"""
m, n = int(m), int(n)
# get sparsity pattern
A = _rand_sparse(m, n, density)
# replace data with random values
A.data = np.random.rand(A.nnz)
return A.asformat(format)
# currently returns positive semi-definite matrices
# def sprand_spd(m, n, density, a=1.0, b=2.0, format='csr'):
# """Returns a random sparse, symmetric positive definite matrix
#
# Parameters
# ----------
# n : int
# shape of the result
# density : float
# target a matrix with nnz(A) = m*n*density, 0<=density<=1
# a,b : float
# eigenvalues of the result will lie in the range [a,b]
# format : string
# sparse matrix format to return, e.g. "csr", "coo", etc.
#
# Returns
# -------
# A : sparse, SPD matrix
# n x n sparse matrix with eigenvalues in the interval [a,b]
#
# Examples
# --------
#
# See Also
# --------
# pyamg.classical.cr.binormalize
#
# """
# # get sparsity pattern
# A = _rand_sparse(n, n, density, format='csr')
#
# A.data = np.random.rand(A.nnz)
#
# A = sp.sparse.tril(A, -1)
#
# A = A + A.T
#
# d = np.array(A.sum(axis=1)).ravel()
#
# from pyamg.util.utils import symmetric_rescaling, diag_sparse
#
# D = diag_sparse(d)
# A = D - A
#
# # A now has zero row sums
# D_sqrt,D_sqrt_inv,A = symmetric_rescaling(A)
#
# # A now has unit diagonals
#
# return A.asformat(format)
