Sni*_*rey 6 python numpy linear-algebra scipy sparse-matrix
我必须求解 x 的大量“Ax=B”类型的线性矩阵方程,其中 A 是一个稀疏矩阵,主要填充主对角线,B 是一个向量。
\n\n我的第一种方法是通过 numpy.linalg.solve 使用密集的 numpy 数组来实现此目的,并且它可以很好地处理 (N,n,n) 维数组,其中 N 是线性矩阵方程的数量,n 是方阵维度。我首先将它与迭代所有方程的 for 循环一起使用,这实际上相当慢。但后来意识到,您也可以将 (N,n,n) 维矩阵直接传递给 numpy.linalg.solve ,而无需任何 for 循环(顺便说一下,我在阅读的文档中没有找到)。这已经大大提高了计算速度(详细信息见下文)。
\n\n但是,因为我有稀疏矩阵,所以我还查看了 scipy.sparse.linalg.spsolve 函数,它与相应的 numpy 函数执行类似的操作。使用 for 循环迭代所有方程比 numpy 解决方案快得多,但似乎不可能将 (N,n,n) 维数组直接传递给 scipy\xc2\xb4s spsolve。
\n\n这是我到目前为止所尝试的:
\n\n首先,我计算一些虚构的 A 矩阵和带有随机数的 B 向量以用于测试目的,包括稀疏和密集:
\n\nimport numpy as np\nfrom scipy import sparse\nfrom scipy.sparse.linalg import spsolve\n\nnumber_of_systems = 100 #corresponds to N in the text\nnumber_of_data_points = 1000 #corresponds to n in the text\n\n#calculation of sample matrices (dense and sparse)\nA_sparse = np.empty(number_of_systems,dtype=object)\nA_dense = np.empty((number_of_systems,number_of_data_points,number_of_data_points))\n\nfor ii in np.arange(number_of_systems):\n A_sparse[ii] = sparse.spdiags(np.random.random(number_of_data_points),0,number_of_data_points,number_of_data_points)\n A_dense[ii] = A_sparse[ii].todense()\n\n#calculation of sample vectors\nB = np.random.random((number_of_systems,number_of_data_points))\nRun Code Online (Sandbox Code Playgroud)\n\n1)第一种方法:带有for循环的numpy.linalg.solve:
\n\ndef solve_dense_3D(A,B):\n results = np.empty((A.shape[0],A.shape[1]))\n for ii in np.arange(A.shape[0]):\n results[ii] = np.linalg.solve(A[ii],B[ii])\n return results\n\nresult_dense_for = solve_dense_3D(A_dense,B)\nRun Code Online (Sandbox Code Playgroud)\n\n定时:
\n\ntimeit(solve_dense_3D(A_dense,B))\n1.25 s \xc2\xb1 27.6 ms per loop (mean \xc2\xb1 std. dev. of 7 runs, 1 loop each)\nRun Code Online (Sandbox Code Playgroud)\n\n2)第二种方法:将(N,n,n)维矩阵直接传递给numpy.linalg.solve:
\n\nresult_dense = np.linalg.solve(A_dense,B)\nRun Code Online (Sandbox Code Playgroud)\n\n定时:
\n\ntimeit(np.linalg.solve(A_dense,B))\n769 ms \xc2\xb1 9.68 ms per loop (mean \xc2\xb1 std. dev. of 7 runs, 1 loop each)\nRun Code Online (Sandbox Code Playgroud)\n\n3)第三种方法:使用 scipy.sparse.linalg.spsolve 和 for 循环:
\n\ndef solve_sparse_3D(A,B):\n results = np.empty((A.shape[0],A[0].shape[0]))\n for ii in np.arange(A.shape[0]):\n results[ii] = spsolve(A[ii],B[ii])\n return results\n\nresult_sparse_for = solve_sparse_3D(A_sparse,B)\nRun Code Online (Sandbox Code Playgroud)\n\n定时:
\n\ntimeit(solve_sparse_3D(A_sparse,B))\n30.9 ms \xc2\xb1 132 \xc2\xb5s per loop (mean \xc2\xb1 std. dev. of 7 runs, 10 loops each)\nRun Code Online (Sandbox Code Playgroud)\n\n很明显,能够省略方法 1 和 2 中的 for 循环是有好处的。正如可能预期的那样,迄今为止最快的替代方案是使用稀疏矩阵的方法 3。
\n\n因为这个例子中的方程数量对我来说仍然相当少,而且因为我必须经常做这样的事情,所以如果有一个使用 scipy\xc2\xb4s 稀疏矩阵而不使用 for 循环的解决方案,我会很高兴。有人知道实现这一目标的方法吗?或者也许有另一种方法以甚至不同的方式解决问题?我很乐意提供建议。
\n一个小演示概述了我上面评论中的想法:
""" YOUR CODE (only imports changed + deterministic randomness) """
import numpy as np
from scipy import sparse
from scipy.sparse.linalg import spsolve
from scipy.sparse import block_diag
from time import perf_counter as pc
np.random.seed(0)
number_of_systems = 100 #corresponds to N in the text
number_of_data_points = 1000 #corresponds to n in the text
#calculation of sample matrices (dense and sparse)
A_sparse = np.empty(number_of_systems,dtype=object)
A_dense = np.empty((number_of_systems,number_of_data_points,number_of_data_points))
for ii in np.arange(number_of_systems):
A_sparse[ii] = sparse.spdiags(np.random.random(number_of_data_points),0,number_of_data_points,number_of_data_points)
A_dense[ii] = A_sparse[ii].todense()
#calculation of sample vectors
B = np.random.random((number_of_systems,number_of_data_points))
def solve_sparse_3D(A,B):
results = np.empty((A.shape[0],A[0].shape[0]))
for ii in np.arange(A.shape[0]):
results[ii] = spsolve(A[ii],B[ii])
return results
start = pc()
result_sparse_for = solve_sparse_3D(A_sparse,B)
end = pc()
print(result_sparse_for)
print(end - start)
""" ALTERNATIVE APPROACH """
def solve_sparse_3D_blockdiag(A,B):
oldN = B.shape[0]
A_ = block_diag(A) # huge sparse block-matrix of independent problems
B_ = B.ravel() # flattened vector
results = spsolve(A_, B_)
return results.reshape(oldN, -1) # unflatten results
start = pc()
result_sparse_for = solve_sparse_3D_blockdiag(A_sparse,B)
end = pc()
print(result_sparse_for)
print(end - start)
Run Code Online (Sandbox Code Playgroud)
哪个输出
[[ 0.97529866 1.26406276 0.83348888 ... 0.99310639 3.90781207
0.16724226]
[ 1.23398934 28.82088739 1.6955886 ... 1.85011686 0.23386882
1.17208753]
[ 0.92864777 0.22248781 0.09445412 ... 2.5080376 0.91701228
0.97266564]
...
[ 0.33087093 0.89034736 1.7523883 ... 0.2171746 4.89236164
0.31546549]
[ 1.2163625 3.0100941 0.87216264 ... 1.62105596 0.33211353
2.07929302]
[ 5.35677404 1.23830776 0.16073721 ... 0.26492506 0.53676822
3.73192617]]
0.08764066299999995
###
[[ 0.97529866 1.26406276 0.83348888 ... 0.99310639 3.90781207
0.16724226]
[ 1.23398934 28.82088739 1.6955886 ... 1.85011686 0.23386882
1.17208753]
[ 0.92864777 0.22248781 0.09445412 ... 2.5080376 0.91701228
0.97266564]
...
[ 0.33087093 0.89034736 1.7523883 ... 0.2171746 4.89236164
0.31546549]
[ 1.2163625 3.0100941 0.87216264 ... 1.62105596 0.33211353
2.07929302]
[ 5.35677404 1.23830776 0.16073721 ... 0.26492506 0.53676822
3.73192617]]
0.07241856000000013
Run Code Online (Sandbox Code Playgroud)
有一些事情要做:
permc_spec| 归档时间: |
|
| 查看次数: |
2428 次 |
| 最近记录: |