|
1
|
|
|
# Licensed under a 3-clause BSD style license - see LICENSE.rst |
|
2
|
|
|
import pytest |
|
3
|
|
|
import numpy as np |
|
4
|
|
|
from numpy.testing import assert_allclose |
|
5
|
|
|
import astropy.units as u |
|
6
|
|
|
from gammapy.datasets import MapDataset, MapDatasetOnOff |
|
7
|
|
|
from gammapy.estimators import ExcessMapEstimator |
|
8
|
|
|
from gammapy.maps import Map, MapAxis, WcsGeom |
|
9
|
|
|
from gammapy.irf import PSFMap |
|
10
|
|
|
from gammapy.modeling.models import ( |
|
11
|
|
|
GaussianSpatialModel, |
|
12
|
|
|
PowerLawSpectralModel, |
|
13
|
|
|
SkyModel, |
|
14
|
|
|
) |
|
15
|
|
|
from gammapy.utils.testing import requires_data |
|
16
|
|
|
from gammapy.estimators.utils import estimate_exposure_reco_energy |
|
17
|
|
|
|
|
18
|
|
|
|
|
19
|
|
|
def image_to_cube(input_map, energy_min, energy_max): |
|
20
|
|
|
energy_min = u.Quantity(energy_min) |
|
21
|
|
|
energy_max = u.Quantity(energy_max) |
|
22
|
|
|
axis = MapAxis.from_energy_bounds(energy_min, energy_max, nbin=1) |
|
23
|
|
|
geom = input_map.geom.to_cube([axis]) |
|
24
|
|
|
return Map.from_geom(geom, data=input_map.data[np.newaxis, :, :]) |
|
25
|
|
|
|
|
26
|
|
|
|
|
27
|
|
|
@pytest.fixture |
|
28
|
|
|
def simple_dataset(): |
|
29
|
|
|
axis = MapAxis.from_energy_bounds(0.1, 10, 1, unit="TeV") |
|
30
|
|
|
geom = WcsGeom.create(npix=20, binsz=0.02, axes=[axis]) |
|
31
|
|
|
dataset = MapDataset.create(geom) |
|
32
|
|
|
dataset.mask_safe += np.ones(dataset.data_shape, dtype=bool) |
|
33
|
|
|
dataset.counts += 2 |
|
34
|
|
|
dataset.background += 1 |
|
35
|
|
|
return dataset |
|
36
|
|
|
|
|
37
|
|
|
|
|
38
|
|
|
@pytest.fixture |
|
39
|
|
|
def simple_dataset_on_off(): |
|
40
|
|
|
axis = MapAxis.from_energy_bounds(0.1, 10, 2, unit="TeV") |
|
41
|
|
|
geom = WcsGeom.create(npix=20, binsz=0.02, axes=[axis]) |
|
42
|
|
|
dataset = MapDatasetOnOff.create(geom) |
|
43
|
|
|
dataset.mask_safe += np.ones(dataset.data_shape, dtype=bool) |
|
44
|
|
|
dataset.counts += 2 |
|
45
|
|
|
dataset.counts_off += 1 |
|
46
|
|
|
dataset.acceptance += 1 |
|
47
|
|
|
dataset.acceptance_off += 1 |
|
48
|
|
|
dataset.exposure.data += 1e6 |
|
49
|
|
|
dataset.psf = None |
|
50
|
|
|
|
|
51
|
|
|
mask_fit = Map.from_geom(geom, data=1, dtype=bool) |
|
52
|
|
|
mask_fit.data[:, :, 10] = False |
|
53
|
|
|
mask_fit.data[:, 10, :] = False |
|
54
|
|
|
dataset.mask_fit = mask_fit |
|
55
|
|
|
return dataset |
|
56
|
|
|
|
|
57
|
|
|
|
|
58
|
|
|
@requires_data() |
|
59
|
|
|
def test_compute_lima_image(): |
|
60
|
|
|
""" |
|
61
|
|
|
Test Li & Ma image against TS image for Tophat kernel |
|
62
|
|
|
""" |
|
63
|
|
|
filename = "$GAMMAPY_DATA/tests/unbundled/poisson_stats_image/input_all.fits.gz" |
|
64
|
|
|
counts = Map.read(filename, hdu="counts") |
|
65
|
|
|
counts = image_to_cube(counts, "1 GeV", "100 GeV") |
|
66
|
|
|
background = Map.read(filename, hdu="background") |
|
67
|
|
|
background = image_to_cube(background, "1 GeV", "100 GeV") |
|
68
|
|
|
dataset = MapDataset(counts=counts, background=background) |
|
69
|
|
|
|
|
70
|
|
|
estimator = ExcessMapEstimator("0.1 deg") |
|
71
|
|
|
result_lima = estimator.run(dataset) |
|
72
|
|
|
|
|
73
|
|
|
assert_allclose(result_lima["sqrt_ts"].data[:, 100, 100], 30.814916, atol=1e-3) |
|
74
|
|
|
assert_allclose(result_lima["sqrt_ts"].data[:, 1, 1], 0.164, atol=1e-3) |
|
75
|
|
|
|
|
76
|
|
|
|
|
77
|
|
|
@requires_data() |
|
78
|
|
|
def test_compute_lima_on_off_image(): |
|
79
|
|
|
""" |
|
80
|
|
|
Test Li & Ma image with snippet from the H.E.S.S. survey data. |
|
81
|
|
|
""" |
|
82
|
|
|
filename = "$GAMMAPY_DATA/tests/unbundled/hess/survey/hess_survey_snippet.fits.gz" |
|
83
|
|
|
n_on = Map.read(filename, hdu="ON") |
|
84
|
|
|
counts = image_to_cube(n_on, "1 TeV", "100 TeV") |
|
85
|
|
|
n_off = Map.read(filename, hdu="OFF") |
|
86
|
|
|
counts_off = image_to_cube(n_off, "1 TeV", "100 TeV") |
|
87
|
|
|
a_on = Map.read(filename, hdu="ONEXPOSURE") |
|
88
|
|
|
acceptance = image_to_cube(a_on, "1 TeV", "100 TeV") |
|
89
|
|
|
a_off = Map.read(filename, hdu="OFFEXPOSURE") |
|
90
|
|
|
acceptance_off = image_to_cube(a_off, "1 TeV", "100 TeV") |
|
91
|
|
|
dataset = MapDatasetOnOff( |
|
92
|
|
|
counts=counts, |
|
93
|
|
|
counts_off=counts_off, |
|
94
|
|
|
acceptance=acceptance, |
|
95
|
|
|
acceptance_off=acceptance_off, |
|
96
|
|
|
) |
|
97
|
|
|
|
|
98
|
|
|
significance = Map.read(filename, hdu="SIGNIFICANCE") |
|
99
|
|
|
significance = image_to_cube(significance, "1 TeV", "10 TeV") |
|
100
|
|
|
estimator = ExcessMapEstimator("0.1 deg", correlate_off=False) |
|
101
|
|
|
results = estimator.run(dataset) |
|
102
|
|
|
|
|
103
|
|
|
# Reproduce safe significance threshold from HESS software |
|
104
|
|
|
results["sqrt_ts"].data[results["npred"].data < 5] = 0 |
|
105
|
|
|
|
|
106
|
|
|
# crop the image at the boundaries, because the reference image |
|
107
|
|
|
# is cut out from a large map, there is no way to reproduce the |
|
108
|
|
|
# result with regular boundary handling |
|
109
|
|
|
actual = results["sqrt_ts"].crop((11, 11)).data |
|
110
|
|
|
desired = significance.crop((11, 11)).data |
|
111
|
|
|
|
|
112
|
|
|
# Set boundary to NaN in reference image |
|
113
|
|
|
# The absolute tolerance is low because the method used here is slightly different from the one used in HGPS |
|
114
|
|
|
# n_off is convolved as well to ensure the method applies to true ON-OFF datasets |
|
115
|
|
|
assert_allclose(actual, desired, atol=0.2) |
|
116
|
|
|
|
|
117
|
|
|
|
|
118
|
|
|
def test_significance_map_estimator_map_dataset(simple_dataset): |
|
119
|
|
|
simple_dataset.exposure = None |
|
120
|
|
|
estimator = ExcessMapEstimator(0.1 * u.deg, selection_optional=["all"]) |
|
121
|
|
|
result = estimator.run(simple_dataset) |
|
122
|
|
|
|
|
123
|
|
|
assert_allclose(result["npred"].data[0, 10, 10], 162) |
|
124
|
|
|
assert_allclose(result["npred_excess"].data[0, 10, 10], 81) |
|
125
|
|
|
assert_allclose(result["sqrt_ts"].data[0, 10, 10], 7.910732, atol=1e-5) |
|
126
|
|
|
|
|
127
|
|
|
assert_allclose(result["npred_excess_err"].data[0, 10, 10], 12.727922, atol=1e-3) |
|
128
|
|
|
assert_allclose(result["npred_excess_errp"].data[0, 10, 10], 13.063328, atol=1e-3) |
|
129
|
|
|
assert_allclose(result["npred_excess_errn"].data[0, 10, 10], 12.396716, atol=1e-3) |
|
130
|
|
|
assert_allclose(result["npred_excess_ul"].data[0, 10, 10], 107.806275, atol=1e-3) |
|
131
|
|
|
|
|
132
|
|
|
|
|
133
|
|
|
def test_significance_map_estimator_map_dataset_exposure(simple_dataset): |
|
134
|
|
|
simple_dataset.exposure += 1e10 * u.cm ** 2 * u.s |
|
135
|
|
|
axis = simple_dataset.exposure.geom.axes[0] |
|
136
|
|
|
simple_dataset.psf = PSFMap.from_gauss(axis, sigma="0.05 deg") |
|
137
|
|
|
|
|
138
|
|
|
model = SkyModel( |
|
139
|
|
|
PowerLawSpectralModel(amplitude="1e-9 cm-2 s-1 TeV-1"), |
|
140
|
|
|
GaussianSpatialModel( |
|
141
|
|
|
lat_0=0.0 * u.deg, lon_0=0.0 * u.deg, sigma=0.1 * u.deg, frame="icrs" |
|
142
|
|
|
), |
|
143
|
|
|
name="sky_model", |
|
144
|
|
|
) |
|
145
|
|
|
|
|
146
|
|
|
simple_dataset.models = [model] |
|
147
|
|
|
simple_dataset.npred() |
|
148
|
|
|
|
|
149
|
|
|
estimator = ExcessMapEstimator(0.1 * u.deg, selection_optional="all") |
|
150
|
|
|
result = estimator.run(simple_dataset) |
|
151
|
|
|
|
|
152
|
|
|
assert_allclose(result["npred_excess"].data.sum(), 19733.602, rtol=1e-3) |
|
153
|
|
|
assert_allclose(result["sqrt_ts"].data[0, 10, 10], 4.217129, rtol=1e-3) |
|
154
|
|
|
|
|
155
|
|
|
|
|
156
|
|
View Code Duplication |
def test_excess_map_estimator_map_dataset_on_off_no_correlation( |
|
|
|
|
|
|
157
|
|
|
simple_dataset_on_off, |
|
158
|
|
|
): |
|
159
|
|
|
# Test with exposure |
|
160
|
|
|
estimator_image = ExcessMapEstimator( |
|
161
|
|
|
0.11 * u.deg, energy_edges=[0.1, 1] * u.TeV, correlate_off=False |
|
162
|
|
|
) |
|
163
|
|
|
|
|
164
|
|
|
result_image = estimator_image.run(simple_dataset_on_off) |
|
165
|
|
|
assert result_image["npred"].data.shape == (1, 20, 20) |
|
166
|
|
|
assert_allclose(result_image["npred"].data[0, 10, 10], 194) |
|
167
|
|
|
assert_allclose(result_image["npred_excess"].data[0, 10, 10], 97) |
|
168
|
|
|
assert_allclose(result_image["sqrt_ts"].data[0, 10, 10], 0.780125, atol=1e-3) |
|
169
|
|
|
assert_allclose(result_image["flux"].data[:, 10, 10], 9.7e-9, atol=1e-5) |
|
170
|
|
|
|
|
171
|
|
|
|
|
172
|
|
|
def test_excess_map_estimator_map_dataset_on_off_with_correlation_no_exposure( |
|
173
|
|
|
simple_dataset_on_off, |
|
174
|
|
|
): |
|
175
|
|
|
# First without exposure |
|
176
|
|
|
simple_dataset_on_off.exposure = None |
|
177
|
|
|
simple_dataset_on_off.mask_fit = None |
|
178
|
|
|
|
|
179
|
|
|
estimator = ExcessMapEstimator( |
|
180
|
|
|
0.11 * u.deg, energy_edges=[0.1, 1, 10] * u.TeV, correlate_off=True |
|
181
|
|
|
) |
|
182
|
|
|
result = estimator.run(simple_dataset_on_off) |
|
183
|
|
|
|
|
184
|
|
|
assert result["npred"].data.shape == (2, 20, 20) |
|
185
|
|
|
assert_allclose(result["npred"].data[:, 10, 10], 194) |
|
186
|
|
|
assert_allclose(result["npred_excess"].data[:, 10, 10], 97) |
|
187
|
|
|
assert_allclose(result["sqrt_ts"].data[:, 10, 10], 5.741116, atol=1e-5) |
|
188
|
|
|
|
|
189
|
|
|
|
|
190
|
|
View Code Duplication |
def test_excess_map_estimator_map_dataset_on_off_with_correlation( |
|
|
|
|
|
|
191
|
|
|
simple_dataset_on_off, |
|
192
|
|
|
): |
|
193
|
|
|
simple_dataset_on_off.exposure = None |
|
194
|
|
|
simple_dataset_on_off.mask_fit = None |
|
195
|
|
|
|
|
196
|
|
|
estimator_image = ExcessMapEstimator( |
|
197
|
|
|
0.11 * u.deg, energy_edges=[0.1, 1] * u.TeV, correlate_off=True |
|
198
|
|
|
) |
|
199
|
|
|
|
|
200
|
|
|
result_image = estimator_image.run(simple_dataset_on_off) |
|
201
|
|
|
assert result_image["npred"].data.shape == (1, 20, 20) |
|
202
|
|
|
assert_allclose(result_image["npred"].data[0, 10, 10], 194) |
|
203
|
|
|
assert_allclose(result_image["npred_excess"].data[0, 10, 10], 97) |
|
204
|
|
|
assert_allclose(result_image["sqrt_ts"].data[0, 10, 10], 5.741116, atol=1e-3) |
|
205
|
|
|
assert_allclose(result_image["flux"].data[:, 10, 10], 9.7e-9, atol=1e-5) |
|
206
|
|
|
|
|
207
|
|
|
|
|
208
|
|
|
def test_excess_map_estimator_map_dataset_on_off_with_correlation_mask_fit( |
|
209
|
|
|
simple_dataset_on_off, |
|
210
|
|
|
): |
|
211
|
|
|
estimator_image = ExcessMapEstimator( |
|
212
|
|
|
0.11 * u.deg, apply_mask_fit=True, correlate_off=True |
|
213
|
|
|
) |
|
214
|
|
|
|
|
215
|
|
|
result_image = estimator_image.run(simple_dataset_on_off) |
|
216
|
|
|
assert result_image["npred"].data.shape == (1, 20, 20) |
|
217
|
|
|
|
|
218
|
|
|
assert_allclose(result_image["sqrt_ts"].data[0, 10, 10], np.nan, atol=1e-3) |
|
219
|
|
|
assert_allclose(result_image["npred"].data[0, 10, 10], np.nan) |
|
220
|
|
|
assert_allclose(result_image["npred_excess"].data[0, 10, 10], np.nan) |
|
221
|
|
|
assert_allclose(result_image["sqrt_ts"].data[0, 9, 9], 7.186745, atol=1e-3) |
|
222
|
|
|
assert_allclose(result_image["npred"].data[0, 9, 9], 304) |
|
223
|
|
|
assert_allclose(result_image["npred_excess"].data[0, 9, 9], 152) |
|
224
|
|
|
|
|
225
|
|
|
assert result_image["flux"].unit == u.Unit("cm-2s-1") |
|
226
|
|
|
assert_allclose(result_image["flux"].data[0, 9, 9], 1.190928e-08, rtol=1e-3) |
|
227
|
|
|
|
|
228
|
|
|
|
|
229
|
|
|
def test_excess_map_estimator_map_dataset_on_off_with_correlation_model( |
|
230
|
|
|
simple_dataset_on_off, |
|
231
|
|
|
): |
|
232
|
|
|
model = SkyModel( |
|
233
|
|
|
PowerLawSpectralModel(amplitude="1e-9 cm-2 s-1TeV-1"), |
|
234
|
|
|
GaussianSpatialModel( |
|
235
|
|
|
lat_0=0.0 * u.deg, lon_0=0.0 * u.deg, sigma=0.1 * u.deg, frame="icrs" |
|
236
|
|
|
), |
|
237
|
|
|
name="sky_model", |
|
238
|
|
|
) |
|
239
|
|
|
|
|
240
|
|
|
simple_dataset_on_off.models = [model] |
|
241
|
|
|
|
|
242
|
|
|
estimator_mod = ExcessMapEstimator( |
|
243
|
|
|
0.11 * u.deg, apply_mask_fit=False, correlate_off=True |
|
244
|
|
|
) |
|
245
|
|
|
result_mod = estimator_mod.run(simple_dataset_on_off) |
|
246
|
|
|
assert result_mod["npred"].data.shape == (1, 20, 20) |
|
247
|
|
|
|
|
248
|
|
|
assert_allclose(result_mod["sqrt_ts"].data[0, 10, 10], 8.899278, atol=1e-3) |
|
249
|
|
|
|
|
250
|
|
|
assert_allclose(result_mod["npred"].data[0, 10, 10], 388) |
|
251
|
|
|
assert_allclose(result_mod["npred_excess"].data[0, 10, 10], 190.68057) |
|
252
|
|
|
|
|
253
|
|
|
assert result_mod["flux"].unit == "cm-2s-1" |
|
254
|
|
|
assert_allclose(result_mod["flux"].data[0, 10, 10], 1.906806e-08, rtol=1e-3) |
|
255
|
|
|
assert_allclose(result_mod["flux"].data.sum(), 5.920642e-06, rtol=1e-3) |
|
256
|
|
|
|
|
257
|
|
|
|
|
258
|
|
|
def test_excess_map_estimator_map_dataset_on_off_reco_exposure( |
|
259
|
|
|
simple_dataset_on_off, |
|
260
|
|
|
): |
|
261
|
|
|
|
|
262
|
|
|
# TODO: this has never worked... |
|
263
|
|
|
model = SkyModel( |
|
264
|
|
|
PowerLawSpectralModel(amplitude="1e-9 cm-2 s-1TeV-1"), |
|
265
|
|
|
GaussianSpatialModel( |
|
266
|
|
|
lat_0=0.0 * u.deg, lon_0=0.0 * u.deg, sigma=0.1 * u.deg, frame="icrs" |
|
267
|
|
|
), |
|
268
|
|
|
name="sky_model", |
|
269
|
|
|
) |
|
270
|
|
|
|
|
271
|
|
|
simple_dataset_on_off.models = [model] |
|
272
|
|
|
|
|
273
|
|
|
spectral_model = PowerLawSpectralModel(index=15) |
|
274
|
|
|
estimator_mod = ExcessMapEstimator( |
|
275
|
|
|
0.11 * u.deg, |
|
276
|
|
|
apply_mask_fit=False, |
|
277
|
|
|
correlate_off=True, |
|
278
|
|
|
spectral_model=spectral_model, |
|
279
|
|
|
) |
|
280
|
|
|
result_mod = estimator_mod.run(simple_dataset_on_off) |
|
281
|
|
|
|
|
282
|
|
|
assert result_mod["flux"].unit == "cm-2s-1" |
|
283
|
|
|
assert_allclose(result_mod["flux"].data.sum(), 5.920642e-06, rtol=1e-3) |
|
284
|
|
|
|
|
285
|
|
|
reco_exposure = estimate_exposure_reco_energy( |
|
286
|
|
|
simple_dataset_on_off, spectral_model=spectral_model |
|
287
|
|
|
) |
|
288
|
|
|
assert_allclose(reco_exposure.data.sum(), 7.977796e12, rtol=0.001) |
|
289
|
|
|
|
|
290
|
|
|
|
|
291
|
|
|
def test_incorrect_selection(): |
|
292
|
|
|
with pytest.raises(ValueError): |
|
293
|
|
|
ExcessMapEstimator(0.11 * u.deg, selection_optional=["bad"]) |
|
294
|
|
|
|
|
295
|
|
|
with pytest.raises(ValueError): |
|
296
|
|
|
ExcessMapEstimator(0.11 * u.deg, selection_optional=["ul", "bad"]) |
|
297
|
|
|
|
|
298
|
|
|
estimator = ExcessMapEstimator(0.11 * u.deg) |
|
299
|
|
|
with pytest.raises(ValueError): |
|
300
|
|
|
estimator.selection_optional = "bad" |
|
301
|
|
|
|
|
302
|
|
|
|
|
303
|
|
|
def test_significance_map_estimator_incorrect_dataset(): |
|
304
|
|
|
with pytest.raises(ValueError): |
|
305
|
|
|
ExcessMapEstimator("bad") |
|
306
|
|
|
|
gammapy /
gammapy
Loading history...