第十一周作业

Numpy

这节课老师讲了 Numpy ，装个 Numpy 都能装半个钟 TAT ……查一查就知道有镜像源可以用啊 QAQ

9.1 Matrix operations

Calculate $A+A, AA^T, A^TA$ and $AB$. Write a function that computes $A(B-\lambda I)$ for any $\lambda$.

考虑到 $n, m$ 太大，不好调试，也不好把运行结果放在博客里。我把题目中的 $n$ 改为 6， $m$ 改为 4。

所有的函数基本上靠 Google 查文档……查到函数就没什么问题啦！

import numpy
from scipy.linalg import toeplitz

n = 6
m = 4

A = numpy.random.normal(size = (n, m))
B = toeplitz(list(numpy.random.rand(m)))
print("A = ")
print(A)
print("")
print("B = ")
print(B)
print("")
print("A + A =")
print(A + A)
print("")
print("A * A^T =")
print(numpy.dot(A, numpy.transpose(A)))
print("")
print("A^T * A =")
print(numpy.dot(numpy.transpose(A), A))
print("")
print("A * B =")
print(numpy.dot(A, B))
print("")
print("A * (B - xI) = ")
f = lambda x: numpy.dot(A, B - x * numpy.eye(m))
print(f(0.5))

运行结果：

A = 
[[-0.11281631  0.41783085  0.09815799 -1.37959734]
 [ 0.86089005  0.79009271 -0.37353857 -1.41071209]
 [ 0.1595781   1.02578769  0.28194433  1.06656449]
 [-0.72462512 -0.40801415 -0.37193955  1.25537165]
 [-0.24226527  0.39698618 -0.84594276  0.25382869]
 [-1.49292275  0.31526557  0.60724678  1.04393588]]

B = 
[[0.61888271 0.80871496 0.83575727 0.28346094]
 [0.80871496 0.61888271 0.80871496 0.83575727]
 [0.83575727 0.80871496 0.61888271 0.80871496]
 [0.28346094 0.83575727 0.80871496 0.61888271]]

A + A =
[[-0.22563263  0.83566169  0.19631597 -2.75919467]
 [ 1.72178011  1.58018542 -0.74707714 -2.82142418]
 [ 0.3191562   2.05157537  0.56388866  2.13312898]
 [-1.44925024 -0.8160283  -0.7438791   2.5107433 ]
 [-0.48453054  0.79397235 -1.69188551  0.50765738]
 [-2.98584551  0.63053115  1.21449356  2.08787177]]

A * A^T =
[[ 2.10023394  2.14255151 -1.03315172 -1.85714759 -0.24001288 -1.08045132]
 [ 2.14255151  3.49501784 -0.66208593 -2.57822576  0.06300517 -2.73567638]
 [-1.03315172 -0.66208593  2.29475797  0.69989839  0.40077921  1.3697925 ]
 [-1.85714759 -2.57822576  0.69989839  2.40585412  0.64686443  2.03784493]
 [-0.24001288  0.06300517  0.40077921  0.64686443  0.99633864  0.23812427]
 [-1.08045132 -2.73567638  1.3697925   2.03784493  0.23812427  3.78676151]]

A^T * A =
[[ 3.59191675  0.52555226 -0.71977057 -3.41830993]
 [ 0.52555226  2.27453544  0.04247137 -0.67928881]
 [-0.71977057  0.04247137  1.45136548  0.64452848]
 [-3.41830993 -0.67928881  0.64452848  7.76114634]]

A * B =
[[-0.04093971 -0.90627463 -0.81133372 -0.45720095]
 [ 0.45968042 -0.29590974 -0.01358561 -0.27079712]
 [ 1.46629637  1.88329708  1.99997554  1.79063459]
 [-0.73342746 -0.09013517 -0.15052698 -0.07026898]
 [-0.46393742 -0.42222306 -0.19969058 -0.26392503]
 [ 0.13444187  0.35133001  0.22730003  0.97746364]]

A * (B - xI) = 
[[ 0.01546845 -1.11519005 -0.86041272  0.23259772]
 [ 0.02923539 -0.69095609  0.17318367  0.43455893]
 [ 1.38650732  1.37040324  1.85900337  1.25735235]
 [-0.3711149   0.11387191  0.0354428  -0.6979548 ]
 [-0.34280478 -0.62071614  0.22328079 -0.39083938]
 [ 0.88090324  0.19369722 -0.07632336  0.45549569]]

9.2 Solving a linear system

Generate a vector $b$ with $m$ entries and solve $Bx = b$

考虑求 $B$ 的逆，然后与 $b$ 相乘即可。

import numpy
from scipy.linalg import toeplitz

n = 6
m = 4

A = numpy.random.normal(size = (n, m))
B = toeplitz(list(numpy.random.rand(m)))
b = numpy.array(list(numpy.random.rand(m)))
b.shape = (1, m)
b = numpy.transpose(b)
print("B = ")
print(B)
print("")
print("b = ")
print(b)
print("")
C = numpy.linalg.inv(B)
x = numpy.dot(C, b)
print("x = ")
print(x)
print("")

print("Validation: Bx = ")
print(numpy.dot(B, x))
print("")

运行结果：

B = 
[[0.04230256 0.65233112 0.22936666 0.48550987]
 [0.65233112 0.04230256 0.65233112 0.22936666]
 [0.22936666 0.65233112 0.04230256 0.65233112]
 [0.48550987 0.22936666 0.65233112 0.04230256]]

b = 
[[0.90444191]
 [0.27756339]
 [0.12886221]
 [0.14356352]]

x = 
[[-3.33095582]
 [-1.25577034]
 [ 2.98309714]
 [ 2.43106269]]

Validation: Bx = 
[[0.90444191]
 [0.27756339]
 [0.12886221]
 [0.14356352]]

9.3 Norms

好吧，又在 Google 上把两个函数找到

import numpy
from scipy.linalg import toeplitz

n = 6
m = 4

A = numpy.random.normal(size = (n, m))
B = toeplitz(list(numpy.random.rand(m)))
print("A =")
print(A)
print("")

AF = numpy.linalg.norm(A)
print("AF =")
print(AF)
print("")

print("B =")
print(B)
print("")
BF = numpy.linalg.norm(B, numpy.inf)
print("BF =")
print(BF)
print("")

运行结果：

A =
[[-1.85502387  1.45307504 -0.7298224   0.19064925]
 [ 0.42006308 -1.06515485 -0.56644324 -0.86077161]
 [ 0.13983461 -0.26931018 -0.40479553 -1.02185437]
 [-0.05903845  1.20521775  0.03547713 -0.341537  ]
 [ 1.37176362 -0.58666238  1.34213027 -1.47081987]
 [ 0.03120845  0.03860088  0.11806082  0.09013899]]

AF =
4.193262620191245

B =
[[0.57248942 0.37242274 0.31374627 0.67335778]
 [0.37242274 0.57248942 0.37242274 0.31374627]
 [0.31374627 0.37242274 0.57248942 0.37242274]
 [0.67335778 0.31374627 0.37242274 0.57248942]]

BF =
1.9320162030663386

9.4 Power iteration

继续搜索 eigenvalue/eigenvalue 的函数……

import numpy
from scipy.linalg import toeplitz

n = 6

Z = numpy.random.normal(size = (n, n))
print("Z =")
print(Z)
print("")
a = numpy.linalg.eig(Z)
print("eig = ")
print(a)

运行结果

Z =
[[ 1.17077846 -0.49404214 -0.19753952 -0.23422968  0.75231007 -0.42429438]
 [-0.52852624  0.37942899 -0.72593476 -0.32278719  0.41597234 -1.5134676 ]
 [ 0.21585695  0.18076521  0.10678318 -0.7991506  -0.34175139  0.43724075]
 [-0.12561833  1.10554731 -1.92405506 -0.35950713 -2.3248662  -0.63795924]
 [ 0.583325    2.45267851 -0.07688235 -0.5721484   0.92778711  0.10313715]
 [-0.7880677  -1.10875937 -0.75130974 -0.95750364  0.51726749 -1.63898488]]

eig = 
(array([-2.35840503+0.9828976j , -2.35840503-0.9828976j ,
        1.22147278+0.j        ,  1.94255349+0.26928215j,
        1.94255349-0.26928215j,  0.19651602+0.j        ]), array([[ 0.14777223+0.03637904j,  0.14777223-0.03637904j,
         0.89696767+0.j        ,  0.46630275-0.02977809j,
         0.46630275+0.02977809j, -0.24901103+0.j        ],
       [ 0.33707481+0.02375161j,  0.33707481-0.02375161j,
        -0.16268442+0.j        , -0.16453443+0.120872j  ,
        -0.16453443-0.120872j  , -0.08025294+0.j        ],
       [-0.11334332-0.26100894j, -0.11334332+0.26100894j,
        -0.09770529+0.j        ,  0.3044162 -0.07231334j,
         0.3044162 +0.07231334j, -0.71040722+0.j        ],
       [-0.12459487-0.56258514j, -0.12459487+0.56258514j,
         0.25583785+0.j        , -0.7026318 +0.j        ,
        -0.7026318 -0.j        ,  0.10038354+0.j        ],
       [-0.25894077-0.20569013j, -0.25894077+0.20569013j,
        -0.14479392+0.j        ,  0.30822043+0.19914949j,
         0.30822043-0.19914949j,  0.40020334+0.j        ],
       [ 0.5833518 +0.j        ,  0.5833518 -0.j        ,
        -0.27021875+0.j        ,  0.11715391+0.00425655j,
         0.11715391-0.00425655j,  0.50659153+0.j        ]]))