-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtest_bays.py
125 lines (102 loc) · 4.01 KB
/
test_bays.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
import numpy as np
from BaysianOptimizer import BaysianMaximization
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
import pytest
def model(a,b,c,d, x = 0, noise = 0.05):
try:
noise_term = +noise*np.random.randn(len(x))
except:
noise_term = +noise*np.random.randn()
return (x**2 * np.sin(5 * np.pi * x)**6.0) + noise_term
def FrankeFunction(a,b,c,d,x= 0,y = 0, noise = 0.1):
try:
noise_term = +noise*np.random.randn(len(x))
except:
noise_term = +noise*np.random.randn()
term1 = 0.75*np.exp(-(0.25*(9*x-2)**2) - 0.25*((9*y-2)**2))
term2 = 0.75*np.exp(-((9*x+1)**2)/49.0 - 0.1*(9*y+1))
term3 = 0.5*np.exp(-(9*x-7)**2/4.0 - 0.25*((9*y-3)**2))
term4 = -0.2*np.exp(-(9*x-4)**2 - (9*y-7)**2)
return term1 + term2 + term3 + term4 + noise_term
def test_optimizer():
opt = BaysianMaximization(model, {},{'x':(0,1)})
fill = np.random.rand(100).reshape((25,4))
opt.SetData(fill, fill)
opt.InitialGuesses(50)
opt.OptimizeHyperPar(cycles=50)
best_opt = opt.best_model_kargs['x']
x = np.linspace(0,1,500)
y = model(0,0,0,0, x=x , noise = 0)
y_gp, sd = opt.gp_estimator.predict(x.reshape(len(x),1), return_std=True)
actual_opt = x[np.argmax(y)]
plt.figure(figsize =(10,10))
plt.subplot(1,2,1)
plt.scatter(opt.hyperpar,opt.model_score, label = 'sample points')
plt.plot(x, y, label='true function')
plt.fill_between(x, y_gp -sd, y_gp + sd, alpha =0.4, label = 'GP', color ='tab:orange')
plt.plot(x, y_gp, color ='tab:orange')
plt.ylabel("f(x)", fontsize=28)
plt.xlabel("x", fontsize = 28)
plt.legend(loc='best', fontsize =24)
plt.tick_params(size = 20, labelsize=22)
plt.subplot(1,2,2)
plt.hist(opt.hyperpar, bins = 20)
plt.xlabel("x", fontsize = 28)
plt.ylabel("count", fontsize=28)
plt.legend(loc='best', fontsize =24)
plt.tick_params(size = 20, labelsize=22)
plt.tight_layout()
plt.savefig('Results/BaysianOpt/test_1D.pdf')
print("found best", best_opt)
print("Actual best", actual_opt)
#assert pytest.approx(abs(best_opt - actual_opt) == 0,abs =1e-3)
def test_optimizer2D():
opt = BaysianMaximization(FrankeFunction, {},{'x':(0,1), 'y':(0,1)})
fill = np.random.rand(100).reshape((25,4))
opt.SetData(fill, fill)
opt.InitialGuesses(200)
opt.OptimizeHyperPar(cycles=200)
best_opt_x = opt.best_model_kargs['x']
best_opt_y = opt.best_model_kargs['y']
x = np.linspace(0,1,500)
y = np.linspace(0,1,500)
X,Y = np.meshgrid(x,y)
Z = FrankeFunction(0,0,0,0, x=X , y= Y, noise = 0)
y_gp = opt.gp_estimator.predict(np.transpose([X.flatten(),Y.flatten()]))
ind = np.unravel_index(np.argmax(Z,axis= None), Z.shape)
plt.subplot(1,2,1)
ax = plt.axes(projection='3d')
ax.set_title("Franke + Sample")
ax.plot_surface(X,Y,Z)
par = opt.hyperpar.T
ax.scatter(par[0], par[1], opt.model_score )
plt.show()
plt.figure(figsize=(10,10))
plt.subplot(2,2,1)
plt.title("Franke Function", fontsize =28)
plt.contourf(X,Y,Z, vmin =0, vmax= 1.1*Z.max(), cmap='coolwarm')
plt.xlabel("x", fontsize = 28)
plt.xlabel("y", fontsize = 28)
plt.subplot(2,2,2)
plt.title("GP", fontsize =28)
c =plt.contourf(X,Y, y_gp.reshape((500,500)), vmin =0, vmax= 1.1*Z.max(), cmap='coolwarm')
plt.xlabel("x", fontsize = 28)
plt.xlabel("y", fontsize = 28)
plt.subplot(2,2,3)
plt.title("Histogram", fontsize =28)
d =plt.hist2d(*par)
dbar=plt.colorbar(d[-1], label='count')
dbar.ax.tick_params(labelsize =22)
dbar.ax.set_title('count',fontsize =28)
plt.plot( X[ind], Y[ind], marker='x', color='r')
plt.subplot(2,2,4)
plt.title("Functions", fontsize = 28)
cbar =plt.colorbar(c, label='f(x,y)')
cbar.ax.tick_params(labelsize =22)
cbar.ax.set_title('f(x,y)',fontsize =28)
plt.tight_layout()
plt.savefig('Results/BaysianOpt/test_2D.pdf')
print("found best", (best_opt_x, best_opt_y))
print("Actual best", X[ind], Y[ind])
test_optimizer2D()