Repository navigation
Expand file tree
/
Copy pathpower_eig.py
More file actions
111 lines (84 loc) · 3.02 KB
/
Copy pathpower_eig.py
File metadata and controls
111 lines (84 loc) · 3.02 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
# power_eig.py: べき乗法
import numpy as np
import scipy as sc
import scipy.linalg as sclinalg
# べき乗法
def power_eig(init_max_ev, mat_a, max_times, rtol=1.0e-10, atol=1.0e-50):
old_ev = init_max_ev
tmp_ev = old_ev
# 絶対値最大成分を取得
index = np.abs(old_ev).argmax()
old_max_eig = old_ev[index]
old_ev /= old_max_eig
# メインループ
for times in range(max_times):
tmp_ev = mat_a.dot(old_ev)
new_max_eig = tmp_ev[index]
if np.abs(new_max_eig - old_max_eig) <= np.abs(old_max_eig) * rtol + atol:
break
index = np.abs(tmp_ev).argmax()
old_max_eig = new_max_eig
new_max_eig = tmp_ev[index]
old_ev = tmp_ev / new_max_eig
print(f'{times:5d}, {new_max_eig:25.17e}')
max_ev = tmp_ev / np.linalg.norm(tmp_ev)
return times, new_max_eig, max_ev
# 逆べき乗法
def inv_power_eig(init_max_ev, mat_a, max_times, rtol=1.0e-10, atol=1.0e-50):
old_ev = init_max_ev
tmp_ev = old_ev
# 絶対値最大成分を取得
index = np.abs(old_ev).argmax()
old_max_eig = old_ev[index]
old_ev /= old_max_eig
# LU分解
mat_lu, pivot = sclinalg.lu_factor(mat_a)
# メインループ
for times in range(max_times):
tmp_ev = sclinalg.lu_solve((mat_lu, pivot), old_ev)
new_max_eig = tmp_ev[index]
if np.abs(new_max_eig - old_max_eig) <= np.abs(old_max_eig) * rtol + atol:
break
index = np.abs(tmp_ev).argmax()
old_max_eig = new_max_eig
new_max_eig = tmp_ev[index]
old_ev = tmp_ev / new_max_eig
print(f'{times:5d}, {1.0 / new_max_eig:25.17e}')
max_ev = tmp_ev / np.linalg.norm(tmp_ev)
return times, 1.0 / new_max_eig, max_ev
# 行列サイズ
str_dim = input('正方行列サイズ dim = ')
dim = int(str_dim) # 文字列→整数
# 乱数行列をAとして与える
np.random.seed(20190515)
mat_a = np.random.rand(dim, dim)
# 対称行列化
for i in range(dim):
for j in range(i):
mat_a[i, j] = mat_a[j, i]
# 行列確認
print('A = ', mat_a)
# 固有値と固有ベクトル
eigval, ev = sclinalg.eig(mat_a)
ev = ev.T
print('Eigenvalues = ', eigval)
print('Eigenvectors = ', ev)
# A * v = lambda * v ?
for i in range(0, eigval.size):
print('|| A * v - lambda[', i, '] * v||_2 = ', sclinalg.norm(mat_a @ ev[i].T - eigval[i] * ev[i]))
# べき乗法実行
max_ev = np.random.rand(dim)
itimes, max_eig, max_ev = power_eig(max_ev, mat_a, dim, 1.0e-13)
print('iterative times = ', itimes)
print('max_eig = ', max_eig)
print('max_ev = ', max_ev)
# 逆べき乗法実行
min_ev = np.random.rand(dim)
itimes, min_eig, min_ev = inv_power_eig(min_ev, mat_a, dim, 1.0e-13)
print('iterative times = ', itimes)
print('min_eig = ', min_eig)
print('min_ev = ', min_ev)
# -------------------------------------
# Copyright (c) 2021 Tomonori Kouya
# All rights reserved.
# -------------------------------------