pyDONEc/test.py at master · csi-dcsc/pyDONEc · GitHub

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
import numpy as np
import pyDONEc


#test function to minimize
def paraboloid(x):
    centers=np.linspace(-10.0,10.0,degrees_of_freedom)
    parab= np.sum((x-centers)**2)
    noise=np.random.normal(0.0,1.0)

    return parab+noise

#hyperparameters, see readme
degrees_of_freedom=10
initial_guess=np.ones(degrees_of_freedom)*0.3
lower_bounds=np.ones(degrees_of_freedom)*(-20.0)
upper_bounds=np.ones(degrees_of_freedom)*(20.0)
cosine_number=1000
regularization_factor=1.0
cosine_sigma=0.05
expl_factors=np.ones(degrees_of_freedom)*0.5
memory_size=100

#initial guess
min_pos=np.linspace(-10.0,10.0,degrees_of_freedom)


#use with direct function call

opt=pyDONEc.optimizer(initial_guess,degrees_of_freedom,lower_bounds,upper_bounds,cosine_number,regularization_factor,cosine_sigma,expl_factors,memory_size)

for i in range(200):
    opt.step(paraboloid)
    print "iteration: "+str(i)+" rms error= "+str(np.sqrt(np.mean((opt.getmin()-min_pos)**2/degrees_of_freedom)))
print "minimum location: "+str(min_pos)
print "estimated minimum: "+str(opt.getmin())
print "final rms error: "+str(np.sqrt(np.mean((opt.getmin()-min_pos)**2/degrees_of_freedom)))

#use with separate metric measurement

opt=pyDONEc.optimizer(initial_guess,degrees_of_freedom,lower_bounds,upper_bounds,cosine_number,regularization_factor,cosine_sigma,expl_factors,memory_size)

a=np.ones(degrees_of_freedom)*0.3
for i in range(200):
    met=paraboloid(a)
    opt.nullstep(a,met)
    a=opt.getlaststep()[0]
    print "iteration: "+str(i)+" rms error= "+str(np.sqrt(np.mean((opt.getmin()-min_pos)**2/degrees_of_freedom)))
print "minimum location: "+str(min_pos)
print "estimated minimum: "+str(opt.getmin())
print "final rms error: "+str(np.sqrt(np.mean((opt.getmin()-min_pos)**2/degrees_of_freedom)))