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
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
|
import os
import time
from colorama import init, Fore, Style
# import math
# import random
import numpy as np
# from numpy.linalg import cholesky, eig
# import matplotlib.pyplot as plt
from read_nc_data import read_nc_data
from read_db_edata import read_db_edata
from c_matrix import c_matrix
from lj_matrix import lj_matrix
# from frob_norm import frob_norm
from gauss_kernel import gauss_kernel
from cholesky_solve import cholesky_solve
# Initialization.
init_time = time.perf_counter()
init()
# Move to data folder.
init_path = os.getcwd()
os.chdir('../../data')
data_path = os.getcwd()
print(Fore.CYAN
+ 'Data reading started.'
+ Style.RESET_ALL)
tic = time.perf_counter()
# Read nuclear charge data.
zi_data = read_nc_data(data_path)
# Read molecule data.
molecules, nuclear_charge, energy_pbe0, energy_delta = \
read_db_edata(zi_data, data_path)
toc = time.perf_counter()
print(Fore.GREEN
+ '\tData reading took {:.4f} seconds.'.format(toc-tic)
+ Style.RESET_ALL)
# Go back to main folder.
os.chdir(init_path)
print(Fore.CYAN
+ 'Coulomb Matrices calculation started.'
+ Style.RESET_ALL)
tic = time.perf_counter()
cm_data = np.array([c_matrix(mol, nc, as_eig=True)
for mol, nc in zip(molecules, nuclear_charge)])
toc = time.perf_counter()
print(Fore.GREEN
+ '\tCoulomb matrices calculation took {:.4f} seconds.'.format(toc-tic)
+ Style.RESET_ALL)
print(Fore.CYAN
+ 'L-J Matrices calculation started.'
+ Style.RESET_ALL)
tic = time.perf_counter()
ljm_data = np.array([lj_matrix(mol, nc, as_eig=True)
for mol, nc in zip(molecules, nuclear_charge)])
toc = time.perf_counter()
print(Fore.GREEN
+ '\tL-J matrices calculation took {:.4f} seconds.'.format(toc-tic)
+ Style.RESET_ALL)
#
# Problem solving with Coulomb Matrix.
#
print(Fore.CYAN
+ 'CM ML started.'
+ Style.RESET_ALL)
tic = time.perf_counter()
sigma = 1000.0
Xcm_training = cm_data[:6000]
Ycm_training = energy_pbe0[:6000]
Kcm_training = gauss_kernel(Xcm_training, Xcm_training, sigma)
alpha_cm = cholesky_solve(Kcm_training, Ycm_training)
Xcm_test = cm_data[-1000:]
Ycm_test = energy_pbe0[-1000:]
Kcm_test = gauss_kernel(Xcm_test, Xcm_training, sigma)
Ycm_predicted = np.dot(Kcm_test, alpha_cm)
print('\tMean absolute error for CM: {}'.format(np.mean(np.abs(Ycm_predicted
- Ycm_test))))
toc = time.perf_counter()
print(Fore.GREEN
+ '\tCM ML took {:.4f} seconds.'.format(toc-tic)
+ Style.RESET_ALL)
#
# Problem solving with L-J Matrix.
#
print(Fore.CYAN
+ 'L-JM ML started.'
+ Style.RESET_ALL)
tic = time.perf_counter()
sigma = 1000.0
Xljm_training = ljm_data[:6000]
Yljm_training = energy_pbe0[:6000]
Kljm_training = gauss_kernel(Xljm_training, Xljm_training, sigma)
alpha_ljm = cholesky_solve(Kljm_training, Yljm_training)
Xljm_test = ljm_data[-1000:]
Yljm_test = energy_pbe0[-1000:]
Kljm_test = gauss_kernel(Xljm_test, Xljm_training, sigma)
Yljm_predicted = np.dot(Kljm_test, alpha_ljm)
print('\tMean absolute error for LJM: {}'.format(np.mean(np.abs(Yljm_predicted
- Yljm_test))))
toc = time.perf_counter()
print(Fore.GREEN
+ '\tL-JM ML took {:.4f} seconds.'.format(toc-tic)
+ Style.RESET_ALL)
# End of program
end_time = time.perf_counter()
print(Fore.CYAN
+ 'The program took {:.4f} seconds of runtime.'.format(end_time
- init_time)
+ Style.RESET_ALL)
|
