1
|
|
|
# -*- coding: utf-8 -*- |
2
|
|
|
# vim:fileencoding=utf-8 |
3
|
|
|
# |
4
|
|
|
# Copyright (c) 2018-2022 Stefan Bender |
5
|
|
|
# |
6
|
|
|
# This module is part of sciapy. |
7
|
|
|
# sciapy is free software: you can redistribute it or modify |
8
|
|
|
# it under the terms of the GNU General Public License as published |
9
|
|
|
# by the Free Software Foundation, version 2. |
10
|
|
|
# See accompanying LICENSE file or http://www.gnu.org/licenses/gpl-2.0.html. |
11
|
|
|
"""SCIAMACHY regression module tests |
12
|
|
|
""" |
13
|
|
|
import numpy as np |
14
|
|
|
|
15
|
|
|
import pytest |
16
|
|
|
|
17
|
|
|
try: |
18
|
|
|
import pymc3 as pm |
19
|
|
|
import arviz as az |
20
|
|
|
except (ImportError, ModuleNotFoundError): |
21
|
|
|
pytest.skip("Theano/PyMC3 packages not installed", allow_module_level=True) |
22
|
|
|
|
23
|
|
|
try: |
24
|
|
|
from sciapy.regress.models_theano import ( |
25
|
|
|
HarmonicModelCosineSine, |
26
|
|
|
HarmonicModelAmpPhase, |
27
|
|
|
) |
28
|
|
|
except (ImportError, ModuleNotFoundError): |
29
|
|
|
pytest.skip("Theano/PyMC3 interface not installed", allow_module_level=True) |
30
|
|
|
|
31
|
|
|
|
32
|
|
|
@pytest.fixture(scope="module") |
33
|
|
|
def xs(): |
34
|
|
|
_xs = np.linspace(0., 11.1, 2048) |
35
|
|
|
return np.ascontiguousarray(_xs, dtype=np.float64) |
36
|
|
|
|
37
|
|
|
|
38
|
|
|
def ys(xs, c, s): |
39
|
|
|
_ys = c * np.cos(2 * np.pi * xs) + s * np.sin(2 * np.pi * xs) |
40
|
|
|
return np.ascontiguousarray(_ys, dtype=np.float64) |
41
|
|
|
|
42
|
|
|
|
43
|
|
|
@pytest.mark.parametrize( |
44
|
|
|
"c, s", |
45
|
|
|
[ |
46
|
|
|
(0.5, 2.0), |
47
|
|
|
(1.0, 0.5), |
48
|
|
|
(1.0, 1.0), |
49
|
|
|
] |
50
|
|
|
) |
51
|
|
|
def test_harmonics_theano(xs, c, s): |
52
|
|
|
# Initialize random number generator |
53
|
|
|
np.random.seed(93457) |
54
|
|
|
yp = ys(xs, c, s) |
55
|
|
|
yp += 0.5 * np.random.randn(xs.shape[0]) |
56
|
|
|
|
57
|
|
|
with pm.Model() as model1: |
58
|
|
|
cos = pm.Normal("cos", mu=0.0, sd=4.0) |
59
|
|
|
sin = pm.Normal("sin", mu=0.0, sd=4.0) |
60
|
|
|
harm1 = HarmonicModelCosineSine(1., cos, sin) |
61
|
|
|
wave1 = harm1.get_value(xs) |
62
|
|
|
# add amplitude and phase for comparison |
63
|
|
|
pm.Deterministic("amp", harm1.get_amplitude()) |
64
|
|
|
pm.Deterministic("phase", harm1.get_phase()) |
65
|
|
|
resid1 = yp - wave1 |
66
|
|
|
pm.Normal("obs", mu=0.0, observed=resid1) |
67
|
|
|
trace1 = pm.sample(tune=800, draws=800, chains=2, return_inferencedata=True) |
68
|
|
|
|
69
|
|
|
with pm.Model() as model2: |
70
|
|
|
amp2 = pm.HalfNormal("amp", sigma=4.0) |
71
|
|
|
phase2 = pm.Normal("phase", mu=0.0, sd=4.0) |
72
|
|
|
harm2 = HarmonicModelAmpPhase(1., amp2, phase2) |
73
|
|
|
wave2 = harm2.get_value(xs) |
74
|
|
|
resid2 = yp - wave2 |
75
|
|
|
pm.Normal("obs", mu=0.0, observed=resid2) |
76
|
|
|
trace2 = pm.sample(tune=800, draws=800, chains=2, return_inferencedata=True) |
77
|
|
|
|
78
|
|
|
np.testing.assert_allclose( |
79
|
|
|
trace1.posterior.median(dim=("chain", "draw"))[["cos", "sin"]].to_array(), |
80
|
|
|
(c, s), |
81
|
|
|
atol=1e-2, |
82
|
|
|
) |
83
|
|
|
np.testing.assert_allclose( |
84
|
|
|
trace1.posterior.median(dim=("chain", "draw"))[["amp", "phase"]].to_array(), |
85
|
|
|
trace2.posterior.median(dim=("chain", "draw"))[["amp", "phase"]].to_array(), |
86
|
|
|
atol=3e-3, |
87
|
|
|
) |
88
|
|
|
|