Coverage for metadetect / tests / test_fitting.py: 5%
456 statements
« prev ^ index » next coverage.py v7.13.5, created at 2026-05-06 08:51 +0000
1import numpy as np
2import galsim
3import ngmix
5import pytest
7from ngmix.gaussmom import GaussMom
8from ngmix.moments import fwhm_to_T
10from .sim import make_mbobs_sim
11from metadetect.fitting import (
12 fit_mbobs_wavg,
13 _combine_fit_results_wavg,
14 symmetrize_obs_weights,
15 fit_all_psfs,
16 _sum_bands_wavg,
17 MOMNAME,
18 _make_mom_res,
19 combine_fit_res,
20)
21from metadetect import procflags
24def _print_res(res):
25 print("", flush=True)
26 for name in res.dtype.names:
27 if "flag" in name:
28 if len(np.shape(res[name])) > 0 and np.shape(res[name])[0] > 1:
29 for i, f in enumerate(res[name]):
30 print(
31 " %s[%d]: %d (%s)" % (
32 name,
33 i,
34 res[name][i],
35 procflags.get_procflags_str(res[name][i]),
36 ),
37 flush=True,
38 )
39 else:
40 print(
41 " %s: %d (%s)" % (
42 name,
43 res[name],
44 procflags.get_procflags_str(res[name]),
45 ),
46 flush=True,
47 )
48 else:
49 print(" %s:" % name, res[name], flush=True)
52def test_fit_all_psfs_same():
53 mbobs1 = make_mbobs_sim(45, 4)
54 fit_all_psfs(mbobs1, np.random.RandomState(seed=10))
56 mbobs2 = make_mbobs_sim(45, 4)
57 fit_all_psfs(mbobs2, np.random.RandomState(seed=10))
59 for i in range(4):
60 for key in mbobs1[i][0].psf.meta["result"]:
61 assert np.all(
62 mbobs1[i][0].psf.meta["result"][key]
63 == mbobs2[i][0].psf.meta["result"][key]
64 )
67def test_fitting_fit_mbobs_wavg_flagging_nodata():
68 mbobs = make_mbobs_sim(45, 4)
69 mbobs[1] = ngmix.ObsList()
70 res = fit_mbobs_wavg(
71 mbobs=mbobs,
72 fitter=GaussMom(1.2),
73 bmask_flags=0,
74 )
75 _print_res(res[0])
76 assert np.all((res["wmom_flags"] & procflags.MISSING_BAND) != 0)
77 assert not np.any(np.isfinite(res["wmom_g_cov"]))
78 assert np.all((res["wmom_band_flux_flags"][:, 1] & procflags.MISSING_BAND) != 0)
79 for i in [0, 2, 3]:
80 assert np.all(res["wmom_band_flux_flags"][:, i] == 0)
81 for tail in ["", "_err"]:
82 assert not np.any(np.isfinite(res["wmom_band_flux" + tail][:, 1]))
83 for i in [0, 2, 3]:
84 assert np.all(np.isfinite(res["wmom_band_flux" + tail][:, i]))
85 assert np.all(res["shear_bands"] == "0123")
87 mbobs = make_mbobs_sim(45, 7)
88 mbobs[1] = ngmix.ObsList()
89 shear_bands = [0, 1, 2, 3]
90 res = fit_mbobs_wavg(
91 mbobs=mbobs,
92 fitter=GaussMom(1.2),
93 bmask_flags=0,
94 shear_bands=shear_bands,
95 )
96 _print_res(res[0])
97 assert np.all((res["wmom_flags"] & procflags.MISSING_BAND) != 0)
98 assert np.all((res["wmom_band_flux_flags"][:, 1] & procflags.MISSING_BAND) != 0)
99 for i in [0, 2, 3, 4, 5, 6]:
100 assert np.all(res["wmom_band_flux_flags"][:, i] == 0)
101 assert not np.any(np.isfinite(res["wmom_g_cov"]))
102 for tail in ["", "_err"]:
103 assert not np.any(np.isfinite(res["wmom_band_flux" + tail][:, 1]))
104 for i in [0, 2, 3, 4, 5, 6]:
105 assert np.all(np.isfinite(res["wmom_band_flux" + tail][:, i]))
106 assert np.all(res["shear_bands"] == "0123")
108 mbobs = make_mbobs_sim(45, 7)
109 mbobs[5] = ngmix.ObsList()
110 shear_bands = [0, 1, 2, 3]
111 res = fit_mbobs_wavg(
112 mbobs=mbobs,
113 fitter=GaussMom(1.2),
114 bmask_flags=0,
115 shear_bands=shear_bands,
116 )
117 _print_res(res[0])
118 assert np.all((res["wmom_flags"] & procflags.MISSING_BAND) != 0)
119 assert np.all((res["wmom_band_flux_flags"][:, 5] & procflags.MISSING_BAND) != 0)
120 for i in [0, 1, 2, 3, 4, 6]:
121 assert np.all(res["wmom_band_flux_flags"][:, i] == 0)
122 assert np.all(np.isfinite(res["wmom_g_cov"]))
123 for tail in ["", "_err"]:
124 assert not np.any(np.isfinite(res["wmom_band_flux" + tail][:, 5]))
125 for i in [0, 1, 2, 3, 4, 6]:
126 assert np.all(np.isfinite(res["wmom_band_flux" + tail][:, i]))
127 assert np.all(res["shear_bands"] == "0123")
130def test_fitting_fit_mbobs_wavg_flagging_edge():
131 bmask_flags = 2**8
132 other_flags = 2**3
133 mbobs = make_mbobs_sim(45, 4)
134 with mbobs[1][0].writeable():
135 mbobs[1][0].bmask[2, 3] = bmask_flags
136 mbobs[1][0].bmask[3, 1] = other_flags
137 res = fit_mbobs_wavg(
138 mbobs=mbobs,
139 fitter=GaussMom(1.2),
140 bmask_flags=bmask_flags,
141 )
142 _print_res(res[0])
143 assert np.all((res["wmom_flags"] & procflags.EDGE_HIT) != 0)
144 for f in [procflags.EDGE_HIT, procflags.MISSING_BAND]:
145 assert np.all((res["wmom_band_flux_flags"][:, 1] & f) != 0)
146 for i in [0, 2, 3]:
147 assert np.all(res["wmom_band_flux_flags"][:, i] == 0)
148 assert not np.any(np.isfinite(res["wmom_g_cov"]))
149 for tail in ["", "_err"]:
150 assert not np.any(np.isfinite(res["wmom_band_flux" + tail][:, 1]))
151 for i in [0, 2, 3]:
152 assert np.all(np.isfinite(res["wmom_band_flux" + tail][:, i]))
153 assert np.all(res["shear_bands"] == "0123")
155 mbobs = make_mbobs_sim(45, 7)
156 with mbobs[1][0].writeable():
157 mbobs[1][0].bmask[2, 3] = bmask_flags
158 mbobs[1][0].bmask[3, 1] = other_flags
159 res = fit_mbobs_wavg(
160 mbobs=mbobs,
161 fitter=GaussMom(1.2),
162 bmask_flags=bmask_flags,
163 shear_bands=list(range(4)),
164 )
165 _print_res(res[0])
166 assert np.all((res["wmom_flags"] & procflags.EDGE_HIT) != 0)
167 for f in [procflags.EDGE_HIT, procflags.MISSING_BAND]:
168 assert np.all((res["wmom_band_flux_flags"][:, 1] & f) != 0)
169 for i in [0, 2, 3, 4, 5, 6]:
170 assert np.all(res["wmom_band_flux_flags"][:, i] == 0)
171 assert not np.any(np.isfinite(res["wmom_g_cov"]))
172 for tail in ["", "_err"]:
173 assert not np.any(np.isfinite(res["wmom_band_flux" + tail][:, 1]))
174 for i in [0, 2, 3, 4, 5, 6]:
175 assert np.all(np.isfinite(res["wmom_band_flux" + tail][:, i]))
176 assert np.all(res["shear_bands"] == "0123")
178 mbobs = make_mbobs_sim(45, 7)
179 with mbobs[5][0].writeable():
180 mbobs[5][0].bmask[2, 3] = bmask_flags
181 mbobs[5][0].bmask[3, 1] = other_flags
182 res = fit_mbobs_wavg(
183 mbobs=mbobs,
184 fitter=GaussMom(1.2),
185 bmask_flags=bmask_flags,
186 shear_bands=list(range(4)),
187 )
188 _print_res(res[0])
189 assert np.all((res["wmom_flags"] & procflags.EDGE_HIT) != 0)
190 for f in [procflags.EDGE_HIT, procflags.MISSING_BAND]:
191 assert np.all((res["wmom_band_flux_flags"][:, 5] & f) != 0)
192 for i in [0, 1, 2, 3, 4, 6]:
193 assert np.all(res["wmom_band_flux_flags"][:, i] == 0)
194 assert np.all(np.isfinite(res["wmom_g_cov"]))
195 for tail in ["", "_err"]:
196 assert not np.any(np.isfinite(res["wmom_band_flux" + tail][:, 5]))
197 for i in [0, 1, 2, 3, 4, 6]:
198 assert np.all(np.isfinite(res["wmom_band_flux" + tail][:, i]))
199 assert np.all(res["shear_bands"] == "0123")
202def test_fitting_fit_mbobs_wavg_flagging_zeroweight():
203 mbobs = make_mbobs_sim(45, 4)
204 with mbobs[1][0].writeable():
205 mbobs[1][0].ignore_zero_weight = False
206 mbobs[1][0].weight[:, :] = 0
207 res = fit_mbobs_wavg(
208 mbobs=mbobs,
209 fitter=GaussMom(1.2),
210 bmask_flags=0,
211 )
212 _print_res(res[0])
213 assert np.all((res["wmom_flags"] & procflags.ZERO_WEIGHTS) != 0)
214 for f in [procflags.ZERO_WEIGHTS, procflags.MISSING_BAND]:
215 assert np.all((res["wmom_band_flux_flags"][:, 1] & f) != 0)
216 for i in [0, 2, 3]:
217 assert np.all(res["wmom_band_flux_flags"][:, i] == 0)
218 assert not np.any(np.isfinite(res["wmom_g_cov"]))
219 for tail in ["", "_err"]:
220 assert not np.any(np.isfinite(res["wmom_band_flux" + tail][:, 1]))
221 for i in [0, 2, 3]:
222 assert np.all(np.isfinite(res["wmom_band_flux" + tail][:, i]))
223 assert np.all(res["shear_bands"] == "0123")
225 mbobs = make_mbobs_sim(45, 7)
226 with mbobs[1][0].writeable():
227 mbobs[1][0].ignore_zero_weight = False
228 mbobs[1][0].weight[:, :] = 0
229 res = fit_mbobs_wavg(
230 mbobs=mbobs,
231 fitter=GaussMom(1.2),
232 bmask_flags=0,
233 shear_bands=list(range(4)),
234 )
235 _print_res(res[0])
236 assert np.all((res["wmom_flags"] & procflags.ZERO_WEIGHTS) != 0)
237 for f in [procflags.ZERO_WEIGHTS, procflags.MISSING_BAND]:
238 assert np.all((res["wmom_band_flux_flags"][:, 1] & f) != 0)
239 for i in [0, 2, 3, 4, 5, 6]:
240 assert np.all(res["wmom_band_flux_flags"][:, i] == 0)
241 assert not np.any(np.isfinite(res["wmom_g_cov"]))
242 for tail in ["", "_err"]:
243 assert not np.any(np.isfinite(res["wmom_band_flux" + tail][:, 1]))
244 for i in [0, 2, 3, 4, 5, 6]:
245 assert np.all(np.isfinite(res["wmom_band_flux" + tail][:, i]))
246 assert np.all(res["shear_bands"] == "0123")
248 mbobs = make_mbobs_sim(45, 7)
249 with mbobs[5][0].writeable():
250 mbobs[5][0].ignore_zero_weight = False
251 mbobs[5][0].weight[:, :] = 0
252 res = fit_mbobs_wavg(
253 mbobs=mbobs,
254 fitter=GaussMom(1.2),
255 bmask_flags=0,
256 shear_bands=list(range(4)),
257 )
258 _print_res(res[0])
259 assert np.all((res["wmom_flags"] & procflags.ZERO_WEIGHTS) != 0)
260 for f in [procflags.ZERO_WEIGHTS, procflags.MISSING_BAND]:
261 assert np.all((res["wmom_band_flux_flags"][:, 5] & f) != 0)
262 for i in [0, 1, 2, 3, 4, 6]:
263 assert np.all(res["wmom_band_flux_flags"][:, i] == 0)
264 assert np.all(np.isfinite(res["wmom_g_cov"]))
265 for tail in ["", "_err"]:
266 assert not np.any(np.isfinite(res["wmom_band_flux" + tail][:, 5]))
267 for i in [0, 1, 2, 3, 4, 6]:
268 assert np.all(np.isfinite(res["wmom_band_flux" + tail][:, i]))
269 assert np.all(res["shear_bands"] == "0123")
272def test_fitting_fit_mbobs_wavg_flagging_combined():
273 mbobs = make_mbobs_sim(45, 4)
274 bmask_flags = 2**8
275 other_flags = 2**3
276 mbobs[0] = ngmix.ObsList()
277 with mbobs[1][0].writeable():
278 mbobs[1][0].bmask[2, 3] = bmask_flags
279 mbobs[1][0].bmask[3, 1] = other_flags
280 with mbobs[2][0].writeable():
281 mbobs[2][0].ignore_zero_weight = False
282 mbobs[2][0].weight[:, :] = 0
283 res = fit_mbobs_wavg(
284 mbobs=mbobs,
285 fitter=GaussMom(1.2),
286 bmask_flags=bmask_flags,
287 )
288 _print_res(res[0])
289 assert np.all((res["wmom_flags"] & procflags.MISSING_BAND) != 0)
290 assert np.all((res["wmom_flags"] & procflags.ZERO_WEIGHTS) != 0)
291 assert np.all((res["wmom_flags"] & procflags.EDGE_HIT) != 0)
292 assert np.all(
293 (
294 res["wmom_band_flux_flags"][:, 0]
295 & (procflags.MISSING_BAND)
296 ) != 0
297 )
298 for f in [procflags.EDGE_HIT, procflags.MISSING_BAND]:
299 assert np.all((res["wmom_band_flux_flags"][:, 1] & f) != 0)
300 for f in [procflags.ZERO_WEIGHTS, procflags.MISSING_BAND]:
301 assert np.all((res["wmom_band_flux_flags"][:, 2] & f) != 0)
302 assert np.all(res["wmom_band_flux_flags"][:, 3:] == 0)
303 assert not np.any(np.isfinite(res["wmom_g_cov"]))
304 for tail in ["", "_err"]:
305 assert np.all(np.isfinite(res["wmom_band_flux" + tail][:, 3]))
306 for i in [0, 1, 2]:
307 assert not np.any(np.isfinite(res["wmom_band_flux" + tail][:, i]))
308 assert np.all(res["shear_bands"] == "0123")
310 mbobs = make_mbobs_sim(45, 7)
311 bmask_flags = 2**8
312 other_flags = 2**3
313 mbobs[0] = ngmix.ObsList()
314 with mbobs[1][0].writeable():
315 mbobs[1][0].bmask[2, 3] = bmask_flags
316 mbobs[1][0].bmask[3, 1] = other_flags
317 with mbobs[2][0].writeable():
318 mbobs[2][0].ignore_zero_weight = False
319 mbobs[2][0].weight[:, :] = 0
320 res = fit_mbobs_wavg(
321 mbobs=mbobs,
322 fitter=GaussMom(1.2),
323 bmask_flags=bmask_flags,
324 shear_bands=list(range(4)),
325 )
326 _print_res(res[0])
327 assert np.all((res["wmom_flags"] & procflags.MISSING_BAND) != 0)
328 assert np.all((res["wmom_flags"] & procflags.ZERO_WEIGHTS) != 0)
329 assert np.all((res["wmom_flags"] & procflags.EDGE_HIT) != 0)
330 assert np.all(
331 (
332 res["wmom_band_flux_flags"][:, 0]
333 & (procflags.MISSING_BAND)
334 ) != 0
335 )
336 for f in [procflags.EDGE_HIT, procflags.MISSING_BAND]:
337 assert np.all((res["wmom_band_flux_flags"][:, 1] & f) != 0)
338 for f in [procflags.ZERO_WEIGHTS, procflags.MISSING_BAND]:
339 assert np.all((res["wmom_band_flux_flags"][:, 2] & f) != 0)
340 assert np.all(res["wmom_band_flux_flags"][:, 3:] == 0)
341 assert not np.any(np.isfinite(res["wmom_g_cov"]))
342 for tail in ["", "_err"]:
343 for i in [3, 4, 5, 6]:
344 assert np.all(np.isfinite(res["wmom_band_flux" + tail][:, i]))
345 for i in [0, 1, 2]:
346 assert not np.any(np.isfinite(res["wmom_band_flux" + tail][:, i]))
347 assert np.all(res["shear_bands"] == "0123")
349 mbobs = make_mbobs_sim(45, 7)
350 bmask_flags = 2**8
351 other_flags = 2**3
352 mbobs[4] = ngmix.ObsList()
353 with mbobs[5][0].writeable():
354 mbobs[5][0].ignore_zero_weight = False
355 mbobs[5][0].weight[:, :] = 0
356 with mbobs[6][0].writeable():
357 mbobs[6][0].bmask[2, 3] = bmask_flags
358 mbobs[6][0].bmask[3, 1] = other_flags
359 res = fit_mbobs_wavg(
360 mbobs=mbobs,
361 fitter=GaussMom(1.2),
362 bmask_flags=bmask_flags,
363 shear_bands=list(range(4)),
364 )
365 _print_res(res[0])
366 assert np.all((res["wmom_flags"] & procflags.MISSING_BAND) != 0)
367 assert np.all((res["wmom_flags"] & procflags.ZERO_WEIGHTS) != 0)
368 assert np.all((res["wmom_flags"] & procflags.EDGE_HIT) != 0)
369 assert np.all(
370 (
371 res["wmom_band_flux_flags"][:, 4]
372 & (procflags.MISSING_BAND)
373 ) != 0
374 )
375 for f in [procflags.ZERO_WEIGHTS, procflags.MISSING_BAND]:
376 assert np.all((res["wmom_band_flux_flags"][:, 5] & f) != 0)
377 for f in [procflags.EDGE_HIT, procflags.MISSING_BAND]:
378 assert np.all((res["wmom_band_flux_flags"][:, 6] & f) != 0)
379 assert np.all(res["wmom_band_flux_flags"][:, :4] == 0)
380 assert np.all(np.isfinite(res["wmom_g_cov"]))
381 for tail in ["", "_err"]:
382 for i in [0, 1, 2, 3]:
383 assert np.all(np.isfinite(res["wmom_band_flux" + tail][:, i]))
384 for i in [4, 5, 6]:
385 assert not np.any(np.isfinite(res["wmom_band_flux" + tail][:, i]))
386 assert np.all(res["shear_bands"] == "0123")
389@pytest.mark.parametrize("purpose,kwargs,psf_flags,model_flags,flux_flags", [
390 (
391 "no data at all",
392 dict(
393 all_res=[],
394 all_psf_res=[],
395 all_is_shear_band=[],
396 all_wgts=[],
397 all_flags=[],
398 ),
399 ["MISSING_BAND"],
400 ["MISSING_BAND", "PSF_FAILURE"],
401 ["MISSING_BAND"],
402 ),
403 (
404 "everything failed",
405 dict(
406 all_res=[None, None, None, None],
407 all_psf_res=[None, None, None, None],
408 all_is_shear_band=[True, True, False, False],
409 all_wgts=[0, 0, 0, 0],
410 all_flags=[0, 0, 0, 0],
411 ),
412 ["MISSING_BAND", "ZERO_WEIGHTS"],
413 ["MISSING_BAND", "PSF_FAILURE", "ZERO_WEIGHTS"],
414 [["MISSING_BAND"]] * 4,
415 ),
417 (
418 "everything failed w/ input flags that should be in the output",
419 dict(
420 all_res=[None, None, None, None],
421 all_psf_res=[None, None, None, None],
422 all_is_shear_band=[True, True, False, False],
423 all_wgts=[0, 0, 0, 0],
424 all_flags=[1, 0, 1, 0],
425 ),
426 ["MISSING_BAND", "ZERO_WEIGHTS", "NO_ATTEMPT"],
427 ["MISSING_BAND", "PSF_FAILURE", "NO_ATTEMPT", "ZERO_WEIGHTS"],
428 [
429 ["MISSING_BAND", "NO_ATTEMPT"],
430 ["MISSING_BAND"],
431 ["MISSING_BAND", "NO_ATTEMPT"],
432 ["MISSING_BAND"]
433 ],
434 ),
435 (
436 "we mark weights zero vs not for failures",
437 dict(
438 all_res=[None, None, None, None],
439 all_psf_res=[None, None, None, None],
440 all_is_shear_band=[True, True, False, False],
441 all_wgts=[1, 1, 1, 1],
442 all_flags=[0, 0, 0, 0],
443 ),
444 ["MISSING_BAND"],
445 ["MISSING_BAND", "PSF_FAILURE"],
446 [["MISSING_BAND"]] * 4,
447 ),
448 (
449 "everything is fine one band",
450 dict(
451 all_res=[{
452 "flux_flags": 0,
453 "flux": 1,
454 "flux_err": 1,
455 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
456 MOMNAME+"_cov": np.diag(np.ones(6))
457 }] * 1,
458 all_psf_res=[
459 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
460 ] * 1,
461 all_is_shear_band=[True],
462 all_wgts=[1],
463 all_flags=[0],
464 ),
465 [],
466 [],
467 [],
468 ),
469 (
470 "everything is fine for more than one band",
471 dict(
472 all_res=[{
473 "flux_flags": 0,
474 "flux": 1,
475 "flux_err": 1,
476 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
477 MOMNAME+"_cov": np.diag(np.ones(6))
478 }] * 4,
479 all_psf_res=[
480 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
481 ] * 4,
482 all_is_shear_band=[True, True, False, False],
483 all_wgts=[1, 1, 1, 1],
484 all_flags=[0, 0, 0, 0],
485 ),
486 [],
487 [],
488 [[]] * 4,
489 ),
490 (
491 "extra shear bands",
492 dict(
493 all_res=[{
494 "flux_flags": 0,
495 "flux": 1,
496 "flux_err": 1,
497 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
498 MOMNAME+"_cov": np.diag(np.ones(6))
499 }] * 2,
500 all_psf_res=[
501 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
502 ] * 1,
503 all_is_shear_band=[True],
504 all_wgts=[1],
505 all_flags=[0],
506 ),
507 ["INCONSISTENT_BANDS"],
508 ["INCONSISTENT_BANDS", "PSF_FAILURE"],
509 [["INCONSISTENT_BANDS"], ["INCONSISTENT_BANDS"]],
510 ),
511 (
512 "extra PSF bands",
513 dict(
514 all_res=[{
515 "flux_flags": 0,
516 "flux": 1,
517 "flux_err": 1,
518 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
519 MOMNAME+"_cov": np.diag(np.ones(6))
520 }],
521 all_psf_res=[
522 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
523 ] * 2,
524 all_is_shear_band=[True],
525 all_wgts=[1],
526 all_flags=[0],
527 ),
528 ["INCONSISTENT_BANDS"],
529 ["INCONSISTENT_BANDS", "PSF_FAILURE"],
530 ["INCONSISTENT_BANDS"],
531 ),
532 (
533 "extra weights",
534 dict(
535 all_res=[{
536 "flux_flags": 0,
537 "flux": 1,
538 "flux_err": 1,
539 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
540 MOMNAME+"_cov": np.diag(np.ones(6))
541 }],
542 all_psf_res=[{MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}],
543 all_is_shear_band=[True],
544 all_wgts=[1, 1],
545 all_flags=[0],
546 ),
547 ["INCONSISTENT_BANDS"],
548 ["INCONSISTENT_BANDS", "PSF_FAILURE"],
549 ["INCONSISTENT_BANDS"],
550 ),
551 (
552 "extra flags",
553 dict(
554 all_res=[{
555 "flux_flags": 0,
556 "flux": 1,
557 "flux_err": 1,
558 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
559 MOMNAME+"_cov": np.diag(np.ones(6))
560 }],
561 all_psf_res=[{MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}],
562 all_is_shear_band=[True],
563 all_wgts=[1],
564 all_flags=[0, 0],
565 ),
566 ["INCONSISTENT_BANDS"],
567 ["INCONSISTENT_BANDS", "PSF_FAILURE"],
568 ["INCONSISTENT_BANDS"],
569 ),
570 (
571 "extra shear bands",
572 dict(
573 all_res=[{
574 "flux_flags": 0,
575 "flux": 1,
576 "flux_err": 1,
577 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
578 MOMNAME+"_cov": np.diag(np.ones(6))
579 }],
580 all_psf_res=[{MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}],
581 all_is_shear_band=[True, False],
582 all_wgts=[1],
583 all_flags=[0, 0],
584 ),
585 ["INCONSISTENT_BANDS"],
586 ["INCONSISTENT_BANDS", "PSF_FAILURE"],
587 ["INCONSISTENT_BANDS"],
588 ),
589 (
590 "flag a single shear",
591 dict(
592 all_res=[{
593 "flux_flags": 0,
594 "flux": 1,
595 "flux_err": 1,
596 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
597 MOMNAME+"_cov": np.diag(np.ones(6))
598 }] * 4,
599 all_psf_res=[
600 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
601 ] * 4,
602 all_is_shear_band=[True, True, False, False],
603 all_wgts=[1, 1, 1, 1],
604 all_flags=[1, 0, 0, 0],
605 ),
606 ["NO_ATTEMPT"],
607 ["NO_ATTEMPT", "PSF_FAILURE"],
608 [["NO_ATTEMPT"], [], [], []],
609 ),
610 (
611 "flag a shear res is fine",
612 dict(
613 all_res=[{
614 "flux_flags": 1,
615 "flux": 1,
616 "flux_err": 1,
617 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
618 MOMNAME+"_cov": np.diag(np.ones(6))
619 }] + [{
620 "flux_flags": 0,
621 "flux": 1,
622 "flux_err": 1,
623 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
624 MOMNAME+"_cov": np.diag(np.ones(6))
625 }] * 3,
626 all_psf_res=[
627 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
628 ] * 4,
629 all_is_shear_band=[True, True, False, False],
630 all_wgts=[1, 1, 1, 1],
631 all_flags=[0, 0, 0, 0],
632 ),
633 [],
634 [],
635 [["NO_ATTEMPT"], [], [], []],
636 ),
637 (
638 "zero weight a shear",
639 dict(
640 all_res=[{
641 "flux_flags": 0,
642 "flux": 1,
643 "flux_err": 1,
644 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
645 MOMNAME+"_cov": np.diag(np.ones(6))
646 }] * 4,
647 all_psf_res=[
648 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
649 ] * 4,
650 all_is_shear_band=[True, True, False, False],
651 all_wgts=[0, 1, 1, 1],
652 all_flags=[0, 0, 0, 0],
653 ),
654 ["ZERO_WEIGHTS"],
655 ["ZERO_WEIGHTS", "PSF_FAILURE"],
656 [[]] * 4,
657 ),
658 (
659 "missing a shear res",
660 dict(
661 all_res=[None] + [{
662 "flux_flags": 0,
663 "flux": 1,
664 "flux_err": 1,
665 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
666 MOMNAME+"_cov": np.diag(np.ones(6))
667 }] * 3,
668 all_psf_res=[
669 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
670 ] * 4,
671 all_is_shear_band=[True, True, False, False],
672 all_wgts=[1, 1, 1, 1],
673 all_flags=[0, 0, 0, 0],
674 ),
675 ["MISSING_BAND"],
676 ["MISSING_BAND", "PSF_FAILURE"],
677 [["MISSING_BAND"], [], [], []],
678 ),
679 (
680 "zero weight a flux",
681 dict(
682 all_res=[{
683 "flux_flags": 0,
684 "flux": 1,
685 "flux_err": 1,
686 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
687 MOMNAME+"_cov": np.diag(np.ones(6))
688 }] * 4,
689 all_psf_res=[
690 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
691 ] * 4,
692 all_is_shear_band=[True, True, False, False],
693 all_wgts=[1, 1, 1, 0],
694 all_flags=[0, 0, 0, 0],
695 ),
696 [],
697 [],
698 [[]] * 4,
699 ),
700 (
701 "flag a flux",
702 dict(
703 all_res=[{
704 "flux_flags": 0,
705 "flux": 1,
706 "flux_err": 1,
707 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
708 MOMNAME+"_cov": np.diag(np.ones(6))
709 }] * 4,
710 all_psf_res=[
711 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
712 ] * 4,
713 all_is_shear_band=[True, True, False, False],
714 all_wgts=[1, 1, 1, 1],
715 all_flags=[0, 0, 0, 1],
716 ),
717 [],
718 [],
719 [[], [], [], ["NO_ATTEMPT"]],
720 ),
721 (
722 "flag a flux in res",
723 dict(
724 all_res=[{
725 "flux_flags": 0,
726 "flux": 1,
727 "flux_err": 1,
728 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
729 MOMNAME+"_cov": np.diag(np.ones(6))
730 }] * 3 + [
731 {
732 "flux_flags": 1,
733 "flux": 1,
734 "flux_err": 1,
735 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
736 MOMNAME+"_cov": np.diag(np.ones(6))
737 }
738 ],
739 all_psf_res=[
740 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
741 ] * 4,
742 all_is_shear_band=[True, True, False, False],
743 all_wgts=[1, 1, 1, 1],
744 all_flags=[0, 0, 0, 0],
745 ),
746 [],
747 [],
748 [[], [], [], ["NO_ATTEMPT"]],
749 ),
750 (
751 "missing a flux res",
752 dict(
753 all_res=[{
754 "flux_flags": 0,
755 "flux": 1,
756 "flux_err": 1,
757 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
758 MOMNAME+"_cov": np.diag(np.ones(6))
759 }] * 3 + [None],
760 all_psf_res=[
761 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
762 ] * 4,
763 all_is_shear_band=[True, True, False, False],
764 all_wgts=[1, 1, 1, 1],
765 all_flags=[0, 0, 0, 0],
766 ),
767 [],
768 [],
769 [[], [], [], ["MISSING_BAND"]],
770 ),
771 (
772 "missing flux",
773 dict(
774 all_res=[{
775 "flux_flags": 0,
776 "flux": 1,
777 "flux_err": 1,
778 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
779 MOMNAME+"_cov": np.diag(np.ones(6))
780 }] * 3 + [{
781 "flux_flags": 0,
782 "flux_err": 1,
783 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
784 MOMNAME+"_cov": np.diag(np.ones(6))
785 }],
786 all_psf_res=[
787 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
788 ] * 4,
789 all_is_shear_band=[True, True, False, False],
790 all_wgts=[1, 1, 1, 1],
791 all_flags=[0, 0, 0, 0],
792 ),
793 [],
794 [],
795 [[], [], [], ["NOMOMENTS_FAILURE"]],
796 ),
797 (
798 "missing flux_err",
799 dict(
800 all_res=[{
801 "flux_flags": 0,
802 "flux": 1,
803 "flux_err": 1,
804 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
805 MOMNAME+"_cov": np.diag(np.ones(6))
806 }] * 3 + [{
807 "flux_flags": 0,
808 "flux": 1,
809 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
810 MOMNAME+"_cov": np.diag(np.ones(6))
811 }],
812 all_psf_res=[
813 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
814 ] * 4,
815 all_is_shear_band=[True, True, False, False],
816 all_wgts=[1, 1, 1, 1],
817 all_flags=[0, 0, 0, 0],
818 ),
819 [],
820 [],
821 [[], [], [], ["NOMOMENTS_FAILURE"]],
822 ),
823 (
824 "missing flux_flags",
825 dict(
826 all_res=[{
827 "flux_flags": 0,
828 "flux": 1,
829 "flux_err": 1,
830 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
831 MOMNAME+"_cov": np.diag(np.ones(6))
832 }] * 3 + [{
833 "flux": 1,
834 "flux_err": 1,
835 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
836 MOMNAME+"_cov": np.diag(np.ones(6))
837 }],
838 all_psf_res=[
839 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
840 ] * 4,
841 all_is_shear_band=[True, True, False, False],
842 all_wgts=[1, 1, 1, 1],
843 all_flags=[0, 0, 0, 0],
844 ),
845 [],
846 [],
847 [[], [], [], ["NOMOMENTS_FAILURE"]],
848 ),
849 (
850 "missing mom",
851 dict(
852 all_res=[{
853 "flux_flags": 0,
854 "flux": 1,
855 "flux_err": 1,
856 MOMNAME+"_cov": np.diag(np.ones(6))
857 }] + [{
858 "flux_flags": 0,
859 "flux": 1,
860 "flux_err": 1,
861 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
862 MOMNAME+"_cov": np.diag(np.ones(6))
863 }] * 3,
864 all_psf_res=[
865 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
866 ] * 4,
867 all_is_shear_band=[True, True, False, False],
868 all_wgts=[1, 1, 1, 1],
869 all_flags=[0, 0, 0, 0],
870 ),
871 ["NOMOMENTS_FAILURE"],
872 ["NOMOMENTS_FAILURE", "PSF_FAILURE"],
873 [[]] * 4,
874 ),
875 (
876 "missing mom_cov",
877 dict(
878 all_res=[{
879 "flux_flags": 0,
880 "flux": 1,
881 "flux_err": 1,
882 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1])
883 }] + [{
884 "flux_flags": 0,
885 "flux": 1,
886 "flux_err": 1,
887 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
888 MOMNAME+"_cov": np.diag(np.ones(6))
889 }] * 3,
890 all_psf_res=[
891 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
892 ] * 4,
893 all_is_shear_band=[True, True, False, False],
894 all_wgts=[1, 1, 1, 1],
895 all_flags=[0, 0, 0, 0],
896 ),
897 ["NOMOMENTS_FAILURE"],
898 ["NOMOMENTS_FAILURE", "PSF_FAILURE"],
899 [[]] * 4,
900 ),
901 (
902 "missing psf mom",
903 dict(
904 all_res=[{
905 "flux_flags": 0,
906 "flux": 1,
907 "flux_err": 1,
908 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
909 MOMNAME+"_cov": np.diag(np.ones(6))
910 }] * 4,
911 all_psf_res=[
912 {MOMNAME+"_cov": np.diag(np.ones(6))}
913 ] + [
914 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
915 ] * 3,
916 all_is_shear_band=[True, True, False, False],
917 all_wgts=[1, 1, 1, 1],
918 all_flags=[0, 0, 0, 0],
919 ),
920 ["NOMOMENTS_FAILURE"],
921 ["PSF_FAILURE"],
922 [[]] * 4,
923 ),
924 (
925 "missing psf mom cov",
926 dict(
927 all_res=[{
928 "flux_flags": 0,
929 "flux": 1,
930 "flux_err": 1,
931 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
932 MOMNAME+"_cov": np.diag(np.ones(6))
933 }] * 4,
934 all_psf_res=[
935 {MOMNAME: np.ones(6)}
936 ] + [
937 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
938 ] * 3,
939 all_is_shear_band=[True, True, False, False],
940 all_wgts=[1, 1, 1, 1],
941 all_flags=[0, 0, 0, 0],
942 ),
943 ["NOMOMENTS_FAILURE"],
944 ["PSF_FAILURE"],
945 [[]] * 4,
946 ),
947 (
948 "missing psf mom for flux",
949 dict(
950 all_res=[{
951 "flux_flags": 0,
952 "flux": 1,
953 "flux_err": 1,
954 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
955 MOMNAME+"_cov": np.diag(np.ones(6))
956 }] * 4,
957 all_psf_res=[
958 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
959 ] * 3 + [
960 {MOMNAME+"_cov": np.diag(np.ones(6))}
961 ],
962 all_is_shear_band=[True, True, False, False],
963 all_wgts=[1, 1, 1, 1],
964 all_flags=[0, 0, 0, 0],
965 ),
966 [],
967 [],
968 [[]] * 4,
969 ),
970 (
971 "missing psf mom for flux",
972 dict(
973 all_res=[{
974 "flux_flags": 0,
975 "flux": 1,
976 "flux_err": 1,
977 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
978 MOMNAME+"_cov": np.diag(np.ones(6))
979 }] * 4,
980 all_psf_res=[
981 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
982 ] * 3 + [
983 {MOMNAME: np.diag(np.ones(6))}
984 ],
985 all_is_shear_band=[True, True, False, False],
986 all_wgts=[1, 1, 1, 1],
987 all_flags=[0, 0, 0, 0],
988 ),
989 [],
990 [],
991 [[]] * 4,
992 ),
993 (
994 "negative/cancelling weights somehow",
995 dict(
996 all_res=[{
997 "flux_flags": 0,
998 "flux": 1,
999 "flux_err": 1,
1000 MOMNAME: np.array([0, 0, 0.5, 0.5, 1, 1]),
1001 MOMNAME+"_cov": np.diag(np.ones(6))
1002 }] * 5,
1003 all_psf_res=[
1004 {MOMNAME: np.ones(6), MOMNAME+"_cov": np.diag(np.ones(6))}
1005 ] * 5,
1006 all_is_shear_band=[True, True, True, False, False],
1007 all_wgts=[-1, 1, 0, 1, 1],
1008 all_flags=[0, 0, 0, 0, 0],
1009 ),
1010 ["ZERO_WEIGHTS"],
1011 ["ZERO_WEIGHTS", "PSF_FAILURE"],
1012 [[]] * 5,
1013 ),
1014])
1015def test_fitting_combine_fit_results_wavg_flagging(
1016 purpose, kwargs, psf_flags, model_flags, flux_flags
1017):
1018 def _check_flags(val, flags):
1019 if np.shape(val) == tuple():
1020 fval = 0
1021 for flag in flags:
1022 fval |= getattr(procflags, flag)
1023 if val != fval:
1024 for flag in flags:
1025 assert (val & getattr(procflags, flag)) != 0, (
1026 "%s: flag val %s failed!" % (purpose, flag)
1027 )
1028 assert val == fval, purpose
1029 else:
1030 for i, _val in enumerate(val):
1031 fval = 0
1032 for flag in flags[i]:
1033 fval |= getattr(procflags, flag)
1035 if _val != fval:
1036 for flag in flags:
1037 assert (_val & getattr(procflags, flag)) != 0, (
1038 "%s: flag val %s failed!" % (purpose, flag)
1039 )
1040 assert _val == fval, purpose
1042 model = "wwmom"
1043 shear_bands = [i for i, b in enumerate(kwargs["all_is_shear_band"]) if b]
1045 data = _combine_fit_results_wavg(
1046 model=model, shear_bands=shear_bands, fwhm_reg=0, **kwargs
1047 )
1048 print()
1049 _print_res(data[0])
1050 _check_flags(data[model + "_psf_flags"][0], psf_flags)
1051 _check_flags(data[model + "_obj_flags"][0], model_flags)
1052 _check_flags(data[model + "_band_flux_flags"][0], flux_flags)
1053 if len(flux_flags) > 1 and isinstance(flux_flags[0], list):
1054 dff = 0
1055 for f in data[model + "_band_flux_flags"][0]:
1056 dff |= f
1057 else:
1058 dff = data[model + "_band_flux_flags"][0]
1059 assert (
1060 data[model + "_flags"][0] ==
1061 (data[model + "_obj_flags"][0] | dff)
1062 ), purpose
1063 if data[model + "_psf_flags"][0] != 0:
1064 assert (data[model + "_obj_flags"][0] & procflags.PSF_FAILURE) != 0, purpose
1067@pytest.mark.parametrize("mom_norm", [None, [0.3, 0.9, 0.8, 0.6]])
1068def test_fitting_sum_bands_wavg_weighting(mom_norm):
1069 all_is_shear_band = [True, True]
1070 all_res = [
1071 {
1072 MOMNAME: np.ones(6) * 3,
1073 MOMNAME+"_cov": np.diag(np.ones(6)) * 3.1,
1074 },
1075 {
1076 MOMNAME: np.ones(6) * 7,
1077 MOMNAME+"_cov": np.diag(np.ones(6)) * 7.1,
1078 },
1079 ]
1080 all_wgts = [0.2, 0.5]
1081 all_flags = [0, 0]
1082 all_wgt_res = [
1083 {
1084 MOMNAME: np.ones(6) * 6,
1085 MOMNAME + "_cov": np.diag(np.ones(6)) * 6.1,
1086 },
1087 {
1088 MOMNAME: np.ones(6) * 2,
1089 MOMNAME+"_cov": np.diag(np.ones(6)) * 2.1,
1090 },
1091 ]
1093 if mom_norm is not None:
1094 all_res[0][MOMNAME+"_norm"] = mom_norm[0]
1095 all_res[1][MOMNAME+"_norm"] = mom_norm[1]
1096 all_wgt_res[0][MOMNAME+"_norm"] = mom_norm[2]
1097 all_wgt_res[1][MOMNAME+"_norm"] = mom_norm[3]
1099 # return value is
1100 # raw_mom, raw_mom_cov, wgt_sum, final_flags, used_shear_bands,
1101 # flux, flux_var, flux_wgt_sum
1102 sums_wgt = _sum_bands_wavg(
1103 all_res=all_res,
1104 all_is_shear_band=all_is_shear_band,
1105 all_wgts=all_wgts,
1106 all_flags=all_flags,
1107 all_wgt_res=all_wgt_res,
1108 )
1109 if mom_norm is None:
1110 fac0 = 6/3
1111 fac1 = 2/7
1112 else:
1113 fac0 = (6 / mom_norm[2]) / (3 / mom_norm[0])
1114 fac1 = (2 / mom_norm[3]) / (7 / mom_norm[1])
1116 assert sums_wgt["wgt_sum"] == 0.7
1117 if mom_norm is None:
1118 np.testing.assert_allclose(
1119 sums_wgt["raw_mom"],
1120 [0.2 * 3 * fac0 + 0.5 * 7 * fac1] * 6
1121 )
1122 else:
1123 np.testing.assert_allclose(
1124 sums_wgt["raw_mom"],
1125 [0.2 * 3 * fac0 / mom_norm[0] + 0.5 * 7 * fac1 / mom_norm[1]] * 6
1126 )
1127 if mom_norm is None:
1128 np.testing.assert_allclose(
1129 np.diag(sums_wgt["raw_mom_cov"]),
1130 [0.2**2 * 3.1 * fac0**2 + 0.5**2 * 7.1 * fac1**2] * 6
1131 )
1132 else:
1133 np.testing.assert_allclose(
1134 np.diag(sums_wgt["raw_mom_cov"]),
1135 [
1136 0.2**2 * 3.1 * fac0**2 / mom_norm[0]**2
1137 + 0.5**2 * 7.1 * fac1**2 / mom_norm[1]**2
1138 ] * 6
1139 )
1141 sums = _sum_bands_wavg(
1142 all_res=all_res,
1143 all_is_shear_band=all_is_shear_band,
1144 all_wgts=all_wgts,
1145 all_flags=all_flags,
1146 all_wgt_res=None,
1147 )
1148 assert sums["wgt_sum"] == 0.7
1149 if mom_norm is None:
1150 np.testing.assert_allclose(
1151 sums["raw_mom"],
1152 [0.2 * 3 + 0.5 * 7] * 6
1153 )
1154 else:
1155 np.testing.assert_allclose(
1156 sums["raw_mom"],
1157 [0.2 * 3 / mom_norm[0] + 0.5 * 7 / mom_norm[1]] * 6
1158 )
1159 if mom_norm is None:
1160 np.testing.assert_allclose(
1161 np.diag(sums["raw_mom_cov"]),
1162 [0.2**2 * 3.1 + 0.5**2 * 7.1] * 6
1163 )
1164 else:
1165 np.testing.assert_allclose(
1166 np.diag(sums["raw_mom_cov"]),
1167 [0.2**2 * 3.1 / mom_norm[0]**2 + 0.5**2 * 7.1 / mom_norm[1]**2] * 6
1168 )
1170 # everything but the moments should be the same
1171 for key in [
1172 "wgt_sum", "final_flags", "used_shear_bands", "flux", "flux_var",
1173 ]:
1174 assert sums[key] == sums_wgt[key], (key, sums[key])
1177def test_fitting_symmetrize_obs_weights_all_zero():
1178 obs = ngmix.Observation(
1179 image=np.zeros((13, 13)),
1180 weight=np.zeros((13, 13)),
1181 ignore_zero_weight=False,
1182 )
1183 sym_obs = symmetrize_obs_weights(obs)
1184 assert sym_obs is not obs
1185 assert sym_obs.ignore_zero_weight is False
1186 assert np.all(sym_obs.weight == 0)
1188 wgt = np.zeros((13, 13))
1189 wgt[:, :2] = 1
1190 obs = ngmix.Observation(
1191 image=np.zeros((13, 13)),
1192 weight=wgt,
1193 )
1194 sym_obs = symmetrize_obs_weights(obs)
1195 assert sym_obs is not obs
1196 assert sym_obs.ignore_zero_weight is False
1197 assert np.all(sym_obs.weight == 0)
1198 assert not np.array_equal(sym_obs.weight, obs.weight)
1201def test_fitting_symmetrize_obs_weights_none():
1202 obs = ngmix.Observation(
1203 image=np.zeros((13, 13)),
1204 weight=np.ones((13, 13)),
1205 )
1206 sym_obs = symmetrize_obs_weights(obs)
1207 assert sym_obs is not obs
1208 assert sym_obs.ignore_zero_weight is True
1209 assert np.all(sym_obs.weight == 1)
1210 assert np.array_equal(sym_obs.weight, obs.weight)
1213def test_fitting_symmetrize_obs_weights():
1214 wgt = np.array([
1215 [1, 1, 0],
1216 [1, 1, 1],
1217 [1, 1, 1],
1218 ])
1219 obs = ngmix.Observation(
1220 image=np.zeros((3, 3)),
1221 weight=wgt,
1222 )
1223 sym_obs = symmetrize_obs_weights(obs)
1224 assert sym_obs is not obs
1225 assert sym_obs.ignore_zero_weight is True
1226 assert not np.array_equal(sym_obs.weight, obs.weight)
1227 sym_wgt = np.array([
1228 [0, 1, 0],
1229 [1, 1, 1],
1230 [0, 1, 0],
1231 ])
1232 assert np.array_equal(sym_obs.weight, sym_wgt)
1235def test_fitting_fit_mbobs_wavg_wmom_tratio():
1236 fitter = GaussMom(1.2)
1237 seed = 10
1238 nband = 3
1239 bmask_flags = 0
1241 mbobs = make_mbobs_sim(
1242 seed,
1243 nband,
1244 simulate_star=True,
1245 noise_scale=1e-4,
1246 band_flux_factors=[0.1, 2.0, 5.0],
1247 band_image_sizes=[39, 45, 67],
1248 )
1249 res = fit_mbobs_wavg(mbobs=mbobs, fitter=fitter, bmask_flags=bmask_flags)
1250 _print_res(res[0])
1251 assert np.allclose(res["wmom_T_ratio"], 1.0)
1253 mbobs = make_mbobs_sim(
1254 seed,
1255 nband,
1256 simulate_star=False,
1257 noise_scale=1e-4,
1258 band_flux_factors=[0.1, 2.0, 5.0],
1259 band_image_sizes=[39, 45, 67],
1260 )
1261 res = fit_mbobs_wavg(mbobs=mbobs, fitter=fitter, bmask_flags=bmask_flags)
1262 _print_res(res[0])
1263 assert not np.allclose(res["wmom_T_ratio"], 1.0)
1264 assert res["wmom_T_ratio"][0] > 1.5
1267@pytest.mark.parametrize("fwhm_reg", [0, 0.8])
1268@pytest.mark.parametrize("has_nan", [True, False])
1269@pytest.mark.parametrize("zero_flux", [True, False])
1270@pytest.mark.parametrize("neg_flux_var", [True, False])
1271def test_make_mom_res(fwhm_reg, has_nan, zero_flux, neg_flux_var):
1272 fwhm = 0.9
1273 image_size = 107
1274 cen = (image_size - 1)/2
1275 gs_wcs = galsim.ShearWCS(
1276 0.125, galsim.Shear(g1=0, g2=0)).jacobian()
1278 obj = galsim.Gaussian(
1279 fwhm=fwhm
1280 ).shear(
1281 g1=-0.1, g2=0.3
1282 ).withFlux(
1283 400)
1284 im = obj.drawImage(
1285 nx=image_size,
1286 ny=image_size,
1287 wcs=gs_wcs,
1288 method='no_pixel').array
1289 noise = np.sqrt(np.sum(im**2)) / 1e2
1290 wgt = np.ones_like(im) / noise**2
1292 fitter = GaussMom(fwhm=1.2)
1294 # get true flux
1295 jac = ngmix.Jacobian(
1296 y=cen, x=cen,
1297 dudx=gs_wcs.dudx, dudy=gs_wcs.dudy,
1298 dvdx=gs_wcs.dvdx, dvdy=gs_wcs.dvdy)
1299 obs = ngmix.Observation(
1300 image=im,
1301 jacobian=jac,
1302 weight=wgt,
1303 )
1304 res = fitter.go(obs=obs)
1306 if has_nan:
1307 res["sums"][0] = np.nan
1308 res["sums"][1] = np.nan
1310 raw_mom = res["sums"].copy()
1311 raw_mom_cov = res["sums_cov"].copy()
1312 raw_flux = res["flux"] / 1.15
1313 raw_flux_var = res["sums_cov"][5, 5] / 1.15**2
1315 if zero_flux:
1316 raw_flux = -1
1317 if neg_flux_var:
1318 raw_flux_var = -1
1320 res_reg = _make_mom_res(
1321 raw_mom=raw_mom,
1322 raw_mom_cov=raw_mom_cov,
1323 raw_flux=raw_flux,
1324 raw_flux_var=raw_flux_var,
1325 fwhm_reg=fwhm_reg,
1326 )
1328 if has_nan:
1329 assert np.isnan(res_reg["sums"][0])
1330 assert np.isnan(res_reg["sums"][1])
1331 assert np.all(np.isfinite(res_reg["sums"][2:]))
1333 T_reg = fwhm_to_T(fwhm_reg)
1335 if not has_nan:
1336 assert np.allclose(res["sums"][[0, 1]], res_reg["sums"][[0, 1]])
1337 assert np.allclose(res["sums"][4] + T_reg * res["sums"][5], res_reg["sums"][4])
1338 if fwhm_reg > 0:
1339 assert not np.allclose(res["sums"][4], res_reg["sums"][4])
1340 assert np.allclose(res["sums"][[2, 3, 5]], res_reg["sums"][[2, 3, 5]])
1341 for col in ["T", "T_err", "T_flags"]:
1342 assert np.allclose(res[col], res_reg[col])
1343 for col in ["e1", "e2", "e", "e_err", "e_cov"]:
1344 if fwhm_reg > 0:
1345 assert not np.allclose(res[col], res_reg[col])
1346 else:
1347 assert np.allclose(res[col], res_reg[col])
1349 if zero_flux:
1350 assert (res_reg["flags"] & ngmix.flags.NONPOS_FLUX) != 0
1351 else:
1352 assert (res_reg["flags"] & ngmix.flags.NONPOS_FLUX) == 0
1353 assert not np.allclose(res["flux"], res_reg["flux"])
1355 if neg_flux_var:
1356 assert (res_reg["flags"] & ngmix.flags.NONPOS_VAR) != 0
1357 assert (res_reg["flux_flags"] & ngmix.flags.NONPOS_VAR) != 0
1358 assert np.isnan(res_reg["flux_err"])
1359 assert np.isnan(res_reg["s2n"])
1360 else:
1361 assert (res_reg["flags"] & ngmix.flags.NONPOS_VAR) == 0
1362 assert (res_reg["flux_flags"] & ngmix.flags.NONPOS_VAR) == 0
1363 assert not np.allclose(res["flux_err"], res_reg["flux_err"])
1365 if not zero_flux and not neg_flux_var:
1366 assert np.allclose(res["s2n"], res_reg["s2n"])
1367 else:
1368 assert not np.allclose(res["s2n"], res_reg["s2n"])
1371@pytest.mark.parametrize("all_res,expected,raises", [
1372 [[None], None, ""],
1373 [[None, None], None, ""],
1374 [
1375 [
1376 None,
1377 np.zeros(0, dtype=[("a_flags", "i8"), ("a", "f4")]),
1378 np.zeros(0, dtype=[("b_flags", "i8"), ("b", "f8")]),
1379 ],
1380 np.zeros(0, dtype=[
1381 ("a_flags", "i8"),
1382 ("a", "f4"),
1383 ("b_flags", "i8"),
1384 ("b", "f8")
1385 ]),
1386 "",
1387 ],
1388 [
1389 [
1390 None,
1391 np.zeros(1, dtype=[("a_flags", "i8"), ("a", "f4")]),
1392 np.zeros(0, dtype=[("b_flags", "i8"), ("b", "f4")]),
1393 ],
1394 None,
1395 "All fit results must be the same length!",
1396 ],
1397 [
1398 [
1399 np.zeros(1, dtype=[("flags", "i8"), ("a", "f4")]),
1400 None,
1401 ],
1402 None,
1403 "All fit results must zero length if one is None!",
1404 ],
1405 [
1406 [
1407 None,
1408 np.zeros(1, dtype=[("flags", "i8"), ("a", "f4")]),
1409 ],
1410 None,
1411 "All fit results must be the same length!",
1412 ],
1413 [
1414 [
1415 np.zeros(1, dtype=[("a_flags", "i8"), ("a", "f4"), ("shear_bands", "i4")]),
1416 np.ones(1, dtype=[("b_flags", "i8"), ("b", "f8"), ("shear_bands", "i4")]),
1417 ],
1418 None,
1419 "Inconsistent column values for shear_bands when combining results!",
1420 ],
1421])
1422def test_combine_fit_res_nodata(all_res, expected, raises):
1423 if raises == "":
1424 res = combine_fit_res(all_res)
1425 if expected is not None:
1426 np.testing.assert_array_equal(expected, res)
1427 else:
1428 assert res is None
1429 else:
1430 with pytest.raises(RuntimeError) as e:
1431 combine_fit_res(all_res)
1433 assert raises in str(e.value)
1436def test_combine_fit_res():
1437 all_res = [
1438 np.zeros(2, dtype=[("a", "f4"), ("shear_bands", "i4")]),
1439 np.zeros(2, dtype=[("b", "f8"), ("shear_bands", "i4")]),
1440 ]
1441 all_res[0]["a"] = [0.3, 2.3]
1442 all_res[1]["b"] = [1.4, 4.6]
1444 all_res[0]["shear_bands"] = [0, 2]
1445 all_res[1]["shear_bands"] = [0, 2]
1447 res = combine_fit_res(all_res)
1449 np.testing.assert_array_equal(res["a"], np.array([0.3, 2.3], dtype="f4"))
1450 np.testing.assert_array_equal(res["b"], np.array([1.4, 4.6], dtype="f8"))
1451 np.testing.assert_array_equal(res["shear_bands"], np.array([0, 2], dtype="i4"))