-
Notifications
You must be signed in to change notification settings - Fork 0
/
sy_lqcao_k.py
178 lines (173 loc) · 5.6 KB
/
sy_lqcao_k.py
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
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
import numpy as np
import math
import pandas as pd
from keras.optimizers import Adam, RMSprop
from ase.db import connect
from ase import Atom
import tensorflow as tf
import keras
from keras.models import Sequential
from keras.layers import Dense,Dropout,Activation, Flatten
from keras.optimizers import RMSprop
#import matplotlib.pyplot as plt
#import tensorflow.compat.v1 as tf
#tf.compat.v1.disable_eager_execution()
#tf.compat.v1.Session()
#tf.compat.v1.Variable()
####dataset
a=[]
x=[]
y=[]
z=[]
E_reference=[]
F_reference=[]
path_to_db='/home/lqcao/work/symmetry-function/iso17/reference.db'
with connect(path_to_db) as conn:
for row in conn.select(limit=10):
for i in row.toatoms():
atoms=i
x.append(atoms.position[0])
y.append(atoms.position[1])
z.append(atoms.position[2])
a.append(atoms.symbol)
E_reference.append(row['total_energy'])
x=np.reshape(z,(-1,19))
y=np.reshape(y,(-1,19))
z=np.reshape(z,(-1,19))
a=np.reshape(a,(-1,19))
print(E_reference)
####symmetry functions
elta=[0.01,0.06,0.2]
Rs=[2,3]
zeta=[1,2]
def G1(rc):
g11=[]
for ii in range(len(a)):
for i in range(len(a[ii])):
fc=[]
for j in range(len(a[ii])):
if i!= j:
Rij=((x[ii][i]-x[ii][j])**2+(y[ii][i]-y[ii][j])**2+(z[ii][i]-z[ii][j])**2)**0.5
if Rij <= float(rc):
fc.append(0.5*(math.cos(math.pi*Rij/float(rc))+1))
else:
fc.append(float(0.0))
f1=sum(fc)
g11.append(f1)
g11=np.reshape(g11,(-1,19))
return g11
#print((G1(6.0)))
def G2(rc,elta,Rs):
g22=[]
for ii in range(len(a)):
for i in range(len(a[ii])):
g2=[]
for j in range(len(a[ii])):
if i!= j:
Rij=((x[ii][i]-x[ii][j])**2+(y[ii][i]-y[ii][j])**2+(z[ii][i]-z[ii][j])**2)**0.5
if Rij <= float(rc):
fc=(0.5*(math.cos(math.pi*Rij/float(rc))+1))
g2.append(math.exp(-elta*((i-Rs)**2))*fc)
else:
fc=float(0.0)
g2.append(math.exp(-elta*((i-Rs)**2))*fc)
g2_value=sum(g2)
g22.append(g2_value)
g22=np.reshape(g22,(-1,19))
return g22
#print(G2(6.0,1,3))
def G4(rc,elta,lam,zeta):
g44=[]
the=[]
for ii in range(len(a)):
for i in range(len(a[ii])):
g4=[]
for j in range(len(a[ii])):
for k in range(len(a[ii])):
if i != j and i != k and j != k:
Rij=((x[ii][i]-x[ii][j])**2+(y[ii][i]-y[ii][j])**2+(z[ii][i]-z[ii][j])**2)**0.5
Rik=((x[ii][i]-x[ii][k])**2+(y[ii][i]-y[ii][k])**2+(z[ii][i]-z[ii][k])**2)**0.5
Rjk=((x[ii][j]-x[ii][k])**2+(y[ii][j]-y[ii][k])**2+(z[ii][j]-z[ii][k])**2)**0.5
if Rij > float(rc) or Rik > float(rc) or Rjk > float(rc):
fc=float(0.0)
g4.append(fc)
else:
fc1=0.5*(math.cos(math.pi*Rij/float(rc))+1)
fc2=0.5*(math.cos(math.pi*Rik/float(rc))+1)
fc3=0.5*(math.cos(math.pi*Rjk/float(rc))+1)
d=((x[ii][j]-x[ii][i])*(x[ii][k]-x[ii][i])+(y[ii][j]-y[ii][i])*(y[ii][k]-y[ii][i])+(z[ii][j]-z[ii][i])*(z[ii][k]-z[ii][i]))
#print(d)
theta=(math.acos(d/(Rij*Rik)))/math.pi*180
the.append(theta)
g4.append(((1+lam*(d/(Rij*Rik)))**zeta)*(math.exp(-elta*(Rij**2+Rik**2+Rjk**2)))*fc1*fc2*fc3)
#print(g4)
g4_value=(2**(1-zeta))*sum(g4)
g44.append(g4_value)
g44=np.reshape(g44,(-1,19))
return g44
#print(G4(4.0,1,1,1))
G=[]
for i in elta:
for j in Rs:
G.append(G2(9.0,i,j))
for i in elta:
for j in zeta:
G.append(G4(9.0,i,1,j))
#print((G))
GG=[]
for k in range(19):
g=[]
for i in range(len(G)):
for j in range(len(G[i])):
g.append(G[i][j][k])
g=np.reshape(g,(len(G),-1))
g_mat=np.matrix(g)
g_mat = np.transpose(g_mat)
g_mat = g_mat.tolist()
GG.append(g_mat)
#print(len(GG))
data_x=[]
for j in range(len(GG[0])):
for i in range(len(GG)):
data_x.append(GG[i][j])
#print(data_x)
#data_a=np.reshape(data_x,(len(GG[0]),19,len(G)))
##normalization
#for e in E_reference:
E_reference=np.array(E_reference)
mine=min(E_reference)
maxe=max(E_reference)
#print(mine,maxe)
data_y=[]
for e in E_reference:
e=(e-mine)/(maxe-mine)
data_y.append(e)
data_y=np.array(data_y)
data_x=np.array(data_x)
data_x=np.reshape(data_x,(1,-1))
data_normal=[]
for i in data_x:
ming=min(i)
maxg=max(i)
data_g=(i-ming)/(maxg-ming)
data_normal.append(data_g)
data_g_normal=np.reshape(data_normal,(len(GG[0]),19,len(G)))
#print(data_g_normal)
##train
adam=Adam(lr=0.001, amsgrad=True, epsilon=1e-5, decay=0.0)
y_data=E_reference
dim=100
model =Sequential()
model.add(Dense(dim,activation='relu',input_shape=(19,len(G))))
model.add(Dropout(0.2))
model.add(Dense(dim,activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(dim,activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(1,activation='sigmoid'))
model.summary()
model.compile(loss='mse',optimizer='adam',metrics=['mse'])
history=model.fit(data_g_normal,data_y,
batch_size=1,
epochs=40000,
verbose=1)