YATSOp/py/calldfomidfuns.py at main · POptUS/YATSOp · 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
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
import numpy as np
import time
import sys
import pickle
from calfun_sample import calfun_sample
from ecnoise import ecnoise
from dfoxsnew import dfoxsnew

# Define the variable matrix (Var)
Var = np.array([
    [1, 100, 200, 1e-7],
    [2, 100, 200, 5e-8],
    [3, 100, 200, 5e-8],
    [201, 100, 100, 4e-9],
    [202, 100, 198, 1e-8],
    [19, 100, 192, 9e-9],
    [128, 100, 100, 1e-6],
    [203, 100, 100, 4e-10],
    [204, 100, 100, 5e-10],
    [205, 125, 125, 1e-9],
    [16, 100, 100, 4e-8],
    [130, 100, 100, 4e-9],
    [131, 100, 100, 4e-9],
    [206, 98, 98, 4e-10],
    [207, 100, 100, 6e-9],
    [15, 100, 100, 4e-10],
    [216, 100, 100, 5e-9],
    [120, 100, 100, 4e-9],
    [219, 110, 110, 2e-8],
    [208, 110, 110, 6e-9],
    [220, 110, 110, 1e-8],
    [4, 100, 100, 1e-8],
    [217, 100, 198, 2e-8],
    [129, 100, 100, 1e-9],
    [21, 100, 100, 2e-7],
    [228, 100, 100, 1e-11],
    [210, 100, 100, 2e-8],
    [211, 100, 100, 2e-8],
    [112, 100, 100, 9e-9],
    [124, 100, 200, 5e-9],
    [123, 100, 101, 8e-7],
    [218, 100, 100, 3e-8],
    [113, 100, 100, 2e-8],
    [5, 100, 198, 1e-8],
    [213, 100, 164, 3e-8],
    [121, 100, 100, 4e-9],
    [125, 100, 102, 1e-8],
    [126, 100, 100, 3e-7],
    [214, 99, 99, 9e-8],
    [215, 99, 99, 9e-8],
    [212, 100, 100, 0],
    [126, 100, 100, 0],
    [209, 100, 198, 0],
    [228, 100, 100, 0]
])

noiseflag = 0

nrows = Var.shape[0]
probtype = 'smooth'
probspecs = {
    'trunc': 10**16
}


namestr = []

if noiseflag:
    sys.path.append('~/repos/randprojections21/src/testfuncs/')
    print('num     prob     n     m            f0     noise     hopt    time')
else:
    print('Problem     n    m            f0     h')


for i in range(40):      #26
    probspecs['nprob'] = int(Var[i, 0])
    probspecs['n'] = int(Var[i, 1])
    probspecs['m'] = int(Var[i, 2])

    factor = 10**0  # revisit!
    X0, prob = dfoxsnew(probspecs['m'], probspecs['n'], probspecs['nprob'])
    # X0 = X0 + 100 * np.ones(X0.shape)
    namestr.append(prob['name'])

    X0 = factor * X0

    start_time = time.time()
    y, fvec = calfun_sample(X0, probspecs, probtype)
    ti = time.time() - start_time

    if noiseflag:
        h = 1e-11
        nf = 23
        np.random.seed(1)  # To replicate the random state

        p = np.random.rand(probspecs['n'])  # Direction to compute derivative
        fval = np.zeros(nf)
        for j in range(nf):
            fval[j] = calfun_sample(X0 + h * (j - 7) * p, probspecs, probtype)

        # Compute noise estimate
        fnoise, level, inform = ecnoise(nf, fval)
        fder2, s2n = calfun_sample(X0, probspecs, probtype)
        hopt = 1.68 * np.sqrt(fnoise / np.abs(fder2))

        p = np.random.rand(probspecs['n'])  # Compute derivative again
        fval = np.zeros(nf)
        for j in range(nf):
            fval[j] = calfun_sample(X0 + h * (j - 7) * p, probspecs, probtype)

        # Compute noise estimate
        fnoise2, level, inform = ecnoise(nf, fval)
        fder2, s2n = calfun_sample(X0, probspecs, probtype)
        hopt2 = 1.68 * np.sqrt(fnoise2 / np.abs(fder2))

        p = np.random.rand(probspecs['n'])  # Compute derivative again
        fval = np.zeros(nf)
        for j in range(nf):
            fval[j] = calfun_sample(X0 + h * (j - 7) * p, probspecs, probtype)

        # Compute noise estimate
        fnoise3, level, inform = ecnoise(nf, fval)
        fder2, s2n = calfun_sample(X0, probspecs, probtype)
        hopt3 = 1.68 * np.sqrt(fnoise3 / np.abs(fder2))

        fnoise0 = (fnoise + fnoise2 + fnoise3) / 3
        hopt0 = (hopt + hopt2 + hopt3) / 3

        print(f"{namestr[i]} {probspecs['n']} {probspecs['m']} {y} {fnoise0[i]} {hopt0[i]}")
    else:
        print(f"{i} {namestr[i]} {probspecs['n']} {probspecs['m']} {y} {prob['h']}")
if noiseflag:
    ratio_hopt = np.min(np.min(np.array([hopt, hopt2, hopt3]).T, axis=1) / np.max(np.array([hopt, hopt2, hopt3]).T, axis=1))
    ratio_fnoise = np.min(np.min(np.array([fnoise, fnoise2, fnoise3]).T, axis=1) / np.max(np.array([fnoise, fnoise2, fnoise3]).T, axis=1))
    hopt = np.round(hopt, decimals=-int(np.floor(np.log10(abs(hopt))) - 1))
    with open('hvals_mid.pkl', 'wb') as file:
        pickle.dump((hopt, namestr), file)