Question

I have the following problem using numpy 1.3.0 and MATLAB 7.9.0: the python code

import numpy as np    
Lu = [[1.01250000000000,-0.00250000000000000,0,0,-0.00250000000000000,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],[-0.00250000000000000,1.01250000000000,-0.00250000000000000,0,0,-0.00250000000000000,0,0,0,0,0,0,0,0,0,0,0,0,0,0],[0,-0.00250000000000000,1.01250000000000,-0.00250000000000000,0,0,-0.00250000000000000,0,0,0,0,0,0,0,0,0,0,0,0,0],[0,0,-0.00250000000000000,1.01250000000000,0,0,0,-0.00250000000000000,0,0,0,0,0,0,0,0,0,0,0,0],[-0.00250000000000000,0,0,0,1.01000000000000,-0.00250000000000000,0,0,-0.00250000000000000,0,0,0,0,0,0,0,0,0,0,0],[0,-0.00250000000000000,0,0,-0.00250000000000000,1.01000000000000,-0.00250000000000000,0,0,-0.00250000000000000,0,0,0,0,0,0,0,0,0,0],[0,0,-0.00250000000000000,0,0,-0.00250000000000000,1.01000000000000,-0.00250000000000000,0,0,-0.00250000000000000,0,0,0,0,0,0,0,0,0],[0,0,0,-0.00250000000000000,0,0,-0.00250000000000000,1.01000000000000,0,0,0,-0.00250000000000000,0,0,0,0,0,0,0,0],[0,0,0,0,-0.00250000000000000,0,0,0,1.01000000000000,-0.00250000000000000,0,0,-0.00250000000000000,0,0,0,0,0,0,0],[0,0,0,0,0,-0.00250000000000000,0,0,-0.00250000000000000,1.01000000000000,-0.00250000000000000,0,0,-0.00250000000000000,0,0,0,0,0,0],[0,0,0,0,0,0,-0.00250000000000000,0,0,-0.00250000000000000,1.01000000000000,-0.00250000000000000,0,0,-0.00250000000000000,0,0,0,0,0],[0,0,0,0,0,0,0,-0.00250000000000000,0,0,-0.00250000000000000,1.01000000000000,0,0,0,-0.00250000000000000,0,0,0,0],[0,0,0,0,0,0,0,0,-0.00250000000000000,0,0,0,1.01000000000000,-0.00250000000000000,0,0,-0.00250000000000000,0,0,0],[0,0,0,0,0,0,0,0,0,-0.00250000000000000,0,0,-0.00250000000000000,1.01000000000000,-0.00250000000000000,0,0,-0.00250000000000000,0,0],[0,0,0,0,0,0,0,0,0,0,-0.00250000000000000,0,0,-0.00250000000000000,1.01000000000000,-0.00250000000000000,0,0,-0.00250000000000000,0],[0,0,0,0,0,0,0,0,0,0,0,-0.00250000000000000,0,0,-0.00250000000000000,1.01000000000000,0,0,0,-0.00250000000000000],[0,0,0,0,0,0,0,0,0,0,0,0,-0.00250000000000000,0,0,0,1.01250000000000,-0.00250000000000000,0,0],[0,0,0,0,0,0,0,0,0,0,0,0,0,-0.00250000000000000,0,0,-0.00250000000000000,1.01250000000000,-0.00250000000000000,0],[0,0,0,0,0,0,0,0,0,0,0,0,0,0,-0.00250000000000000,0,0,-0.00250000000000000,1.01250000000000,-0.00250000000000000],[0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,-0.00250000000000000,0,0,-0.00250000000000000,1.01250000000000]]
rhs = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0,  0, 0,0, 0,  0, 0, 0.0050, 0.0050, 0.0050, 0.0050]

Lu = np.array(Lu)
rhs = np.array(rhs)
ans = np.linalg.solve(Lu,rhs)
print ans

产出

[  1.87241716e-13   1.89545264e-13   1.89545264e-13   1.87241716e-13
   7.56433496e-11   7.63890449e-11   7.63890449e-11   7.56433496e-11
   3.04833369e-08   3.07089522e-08   3.07089522e-08   3.04833369e-08
   1.22844835e-05   1.23451480e-05   1.23451480e-05   1.22844835e-05
   4.95055571e-03   4.96277946e-03   4.96277946e-03   4.95055571e-03]

whereas using backslash in MATLAB 产出

 0.0000    0.0000    0.0000    0.0000    0.0000    0.0000    0.0000    0.0000    0.0000    0.0000    0.0000    0.0000    0.0000    0.0000    0.0000    0.0000    0.0050    0.0050    0.0050    0.0050

我没有发现任何其他线性薄膜等系统,这种系统产生不同的解决办法。目前,我已经离开城镇,因此无法检查,不同版本的 n(不同计算机)是否会产生正确的结果。否。是否解决利用这一系统的正确职能(系统矩阵L正在稀少)? 难道这是我版本的 n吗? 我的法典中是否有问题?

感谢!

Answer 1

Actually, those are likely the same solution. MATLAB is rounding to 0.0000 (when printing at least), while python is giving you much more detailed numbers (some problems might be from floating point rounding errors). The only numbers that are showing up as 0.0050 are the ones that are e-03. All of the other numbers are smaller than 0.0005, so it might get rounded to 0.0000.

友情链接