如何使稀疏 csc 矩阵的多列中的所有元素为 0答案

【问题标题】：How to make all elements in many columns of a sparse csc matrix 0如何使稀疏 csc 矩阵的多列中的所有元素为 0
【发布时间】：2018-07-18 10:09:10
【问题描述】：

my_csr_matrix

my_csr_matrix[:,736225:1783504] = 0

Traceback (most recent call last):
  File "C:\Users\abhatia\AppData\Local\Continuum\anaconda3\lib\site-packages\IPython\core\interactiveshell.py", line 2862, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-135-b0e125d5d27e>", line 1, in <module>
    my_csr_matrix[:,736225:1783504] = 0
  File "C:\Users\abhatia\AppData\Local\Continuum\anaconda3\lib\site-packages\scipy\sparse\compressed.py", line 695, in __setitem__
    i, j = self._swap((i.ravel(), j.ravel()))
MemoryError

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "C:\Users\abhatia\AppData\Local\Continuum\anaconda3\lib\site-packages\IPython\core\interactiveshell.py", line 2802, in run_ast_nodes
    if self.run_code(code, result):
  File "C:\Users\abhatia\AppData\Local\Continuum\anaconda3\lib\site-packages\IPython\core\interactiveshell.py", line 2879, in run_code
    self.showtraceback(running_compiled_code=True)
TypeError: showtraceback() got an unexpected keyword argument 'running_compiled_code'

【问题讨论】：

标签： python scipy sparse-matrix

【解决方案1】：

In [5]: M = sparse.random(10,10,.2,format='csr')
In [6]: M
Out[6]: 
<10x10 sparse matrix of type '<class 'numpy.float64'>'
    with 20 stored elements in Compressed Sparse Row format>

In [8]: M[:,5:]
Out[8]: 
<10x5 sparse matrix of type '<class 'numpy.float64'>'
    with 12 stored elements in Compressed Sparse Row format>
In [9]: M[:,5:] = 0
/usr/local/lib/python3.5/dist-packages/scipy/sparse/compressed.py:742: SparseEfficiencyWarning: Changing the sparsity structure of a csr_matrix is expensive. lil_matrix is more efficient.
  SparseEfficiencyWarning)
In [10]: M
Out[10]: 
<10x10 sparse matrix of type '<class 'numpy.float64'>'
    with 58 stored elements in Compressed Sparse Row format>

设置这些元素会添加一堆non-zero 元素（即使值为0）。此分配不区分设置为 0 和设置为其他值。

我们通过单独的步骤删除 0；

In [11]: M.eliminate_zeros?
Signature: M.eliminate_zeros()
Docstring:
Remove zero entries from the matrix

This is an *in place* operation
File:      /usr/local/lib/python3.5/dist-packages/scipy/sparse/compressed.py
Type:      method
In [12]: M.eliminate_zeros()
In [13]: M
Out[13]: 
<10x10 sparse matrix of type '<class 'numpy.float64'>'
    with 8 stored elements in Compressed Sparse Row format>

乘以 0 不会添加新值；它只是改变了现有的：

In [16]: M[:,5:] *= 0
In [17]: M
Out[17]: 
<10x10 sparse matrix of type '<class 'numpy.float64'>'
    with 20 stored elements in Compressed Sparse Row format>
In [18]: M.data
Out[18]: 
array([0.89042028, 0.        , 0.        , 0.93756551, 0.34072221,
       0.3883514 , 0.        , 0.18581085, 0.        , 0.        ,
       0.        , 0.76948544, 0.        , 0.        , 0.        ,
       0.        , 0.90694047, 0.00354749, 0.        , 0.        ])
In [19]: M.eliminate_zeros()
In [20]: M
Out[20]: 
<10x10 sparse matrix of type '<class 'numpy.float64'>'
    with 8 stored elements in Compressed Sparse Row format>

这应该可以避免大型矩阵中的内存错误。

这些列上的矩阵乘积可能会更快 - 尽管它会生成一个新矩阵。

【讨论】：