sparse.linalg.expm是一个复杂的函数，使用，如docs的Pade approximation。从探索多年前，我记得它涉及到一个矩阵求逆。
让我们用一个小矩阵来测试它：

In [60]: A = sparse.csr_matrix(np.eye(3))

In [61]: A.A
Out[61]: 
array([[1., 0., 0.],
       [0., 1., 0.],
       [0., 0., 1.]])

使用csr：

In [63]: linalg.expm(A)
C:\Users\paul\anaconda3\lib\site-packages\scipy\sparse\linalg\dsolve\linsolve.py:318: SparseEfficiencyWarning: splu requires CSC matrix format
  warn('splu requires CSC matrix format', SparseEfficiencyWarning)
C:\Users\paul\anaconda3\lib\site-packages\scipy\sparse\linalg\dsolve\linsolve.py:215: SparseEfficiencyWarning: spsolve is more efficient when sparse b is in the CSC matrix format
  warn('spsolve is more efficient when sparse b '
Out[63]: 
<3x3 sparse matrix of type '<class 'numpy.float64'>'
    with 3 stored elements in Compressed Sparse Row format>

请注意，警告来自spsolve和splu步骤。
我注意到了警告，却没有得到警告：

In [64]: linalg.expm(A.tocsc())
Out[64]: 
<3x3 sparse matrix of type '<class 'numpy.float64'>'
    with 3 stored elements in Compressed Sparse Column format>

In [65]: _.A
Out[65]: 
array([[2.71828183, 0.        , 0.        ],
       [0.        , 2.71828183, 0.        ],
       [0.        , 0.        , 2.71828183]])

它也适用于密集阵列：

In [66]: linalg.expm(A.A)
Out[66]: 
array([[2.71828183, 0.        , 0.        ],
       [0.        , 2.71828183, 0.        ],
       [0.        , 0.        , 2.71828183]])

如果我尝试更改csc矩阵的元素，则会收到警告：

In [86]: M = sparse.random(100,100,.2,'csc')

In [87]: M[50,50] = 12
C:\Users\paul\anaconda3\lib\site-packages\scipy\sparse\_index.py:82: SparseEfficiencyWarning: Changing the sparsity structure of a csc_matrix is expensive. lil_matrix is more efficient.
  self._set_intXint(row, col, x.flat[0])