判断sparse matrix是否是对称矩阵

参考：
https://stackoverflow.com/questions/48798893/error-in-checking-symmetric-sparse-matrix

import scipy.sparse as sp
import numpy as np
np.random.seed(1)

a = sp.random(5, 5, density=0.5)
1
2
3
4
5

a
1

结果如下

sym_err = a - a.T
sym_check_res = np.all(np.abs(sym_err.data) < 1e-10)  # tune this value
print(sym_check_res)
1
2
3

adj_dense =np.array([[0, 0, 0, 0, 0, 0, 0, 0, 1, 0],
       [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 1, 0, 0, 0, 0],
       [0, 0, 0, 0, 1, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 0, 0, 0, 1, 0],
       [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
       [1, 0, 0, 0, 0, 0, 1, 0, 0, 0],
       [0, 0, 0, 0, 0, 0, 0, 0, 0, 0]])
print(adj_dense)

adj = sp.csr_matrix(adj_dense)
adj
1
2
3
4
5
6
7
8
9
10
11
12
13
14

sym_err = adj - adj.T
sym_check_res=np.all(np.abs(sym_err.data) < 1e-10) 
print(sym_check_res)
1
2
3