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
|
"""
Go through each model, vary one parameter and fit it back to the
default value of that model.
"""
import numpy as np
import os
import pycorrfit
from pycorrfit.correlation import Correlation
from pycorrfit.fit import Fit
import pytest
# GLOBAL PARAMETERS FOR THIS TEST:
TAUMIN = 1e-3
TAUMAX = 1e6
TAULEN = 100
FITALG = "Lev-Mar"
def fit_single_parameter(modelid, fullparms, parmid, parmval, noise=False):
"""
Use the full parameter set `fullparms` and leave a single parameter
`parmid` variable during the fit.
Returns the fitted value of the parameter with index `parmid`
"""
corr = Correlation(fit_model=modelid, fit_algorithm=FITALG, verbose=0)
tau = np.exp(np.linspace(np.log(TAUMIN), np.log(TAUMAX), TAULEN))
# Create artificial data by using the current fit_model
data = corr.fit_model(fullparms, tau)
if noise:
if noise is True:
deltanoise = (np.max(data)-np.min(data))/20
else:
deltanoise = (np.max(data)-np.min(data))*noise
anoise = (np.random.random(data.shape[0])-.5)*deltanoise
data += anoise
# Add artificial data to data set
corr.correlation = np.dstack((tau, data))[0]
# Set variable parameters
fit_bool = np.zeros(fullparms.shape[0])
fit_bool[parmid] = True
corr.fit_parameters_variable = fit_bool
fullparms_edit = fullparms.copy()
fullparms_edit[parmid] = parmval
corr.fit_parameters = fullparms_edit
Fit(corr)
return corr.fit_parameters[parmid]
def deviate_parameter(model, parmid):
"""
Returns a deviated version of the parameter with id `parmid`.
Performs model checks to ensure the new value is physical.
"""
val = model.default_values[parmid]
if val == 0:
val += .1
else:
val *= .9
return val
@pytest.mark.filterwarnings('ignore::pycorrfit.fit.StuckParameterWarning')
@pytest.mark.skipif(os.uname().machine == 'mips64',
reason='minor skew on mips causes test failure')
def test_fit_single_parameter():
"""
Deviate a single parameter and fit it back.
"""
allow_fail = [
[6082, "SP"],
]
faillist = list()
for model in pycorrfit.models.models:
fullparms = model.default_values
for ii, val in enumerate(fullparms):
newval = deviate_parameter(model, ii)
fitval = fit_single_parameter(model.id, fullparms, ii,
newval, noise=False)
# print(val-fitval)
if not np.allclose([val], [fitval]):
if not [model.id, model.parameters[0][ii]] in allow_fail:
faillist.append([model.id, model.parameters[0][ii],
val, fitval])
if faillist:
raise ValueError("Model tests failed for:\n", faillist)
def fit_single_parameter_with_noise(noise=0.005):
"""
Deviate a single parameter and fit it back.
"""
faillist = list()
succlist = list()
for model in pycorrfit.models.models:
fullparms = model.default_values
for ii, val in enumerate(fullparms):
newval = deviate_parameter(model, ii)
fitval = fit_single_parameter(model.id, fullparms, ii, newval,
noise=noise)
if not np.allclose([val], [fitval], atol=.1, rtol=.1):
faillist.append([model.id, model.parameters[0][ii],
val, fitval])
else:
succlist.append([model.id, model.parameters[0][ii],
val, fitval])
return succlist, faillist
@pytest.mark.filterwarnings('ignore::pycorrfit.fit.StuckParameterWarning')
@pytest.mark.skipif(os.uname().machine == 'mips64',
reason='minor skew on mips causes test failure')
def test_fit_single_parameter_with_noise_one_permille():
succlist, faillist = fit_single_parameter_with_noise(noise=0.001)
if len(faillist)/len(succlist) > .01:
raise ValueError("Model tests failed for:\n", faillist)
@pytest.mark.filterwarnings('ignore::pycorrfit.fit.StuckParameterWarning')
@pytest.mark.skipif(os.uname().machine == 'mips64',
reason='minor skew on mips causes test failure')
def test_fit_single_parameter_with_noise_two_percent():
succlist, faillist = fit_single_parameter_with_noise(noise=0.02)
if len(faillist)/len(succlist) > .05:
raise ValueError("Model tests failed for:\n", faillist)
@pytest.mark.filterwarnings('ignore::pycorrfit.fit.StuckParameterWarning')
@pytest.mark.skipif(os.uname().machine == 'mips64',
reason='minor skew on mips causes test failure')
def test_fit_single_parameter_with_noise_five_percent():
succlist, faillist = fit_single_parameter_with_noise(noise=0.05)
if len(faillist)/len(succlist) > .10:
raise ValueError("Model tests failed for:\n", faillist)
if __name__ == "__main__":
# Run all tests
loc = locals()
for key in list(loc.keys()):
if key.startswith("test_") and hasattr(loc[key], "__call__"):
loc[key]()
|
