Galaxy.py 20.7 KB
Newer Older
Fang Yuedong's avatar
Fang Yuedong committed
1
2
3
4
5
6
7
import numpy as np
import galsim
import os, sys
import astropy.constants as cons
from astropy.table import Table
from scipy import interpolate

8
9
10
11
from ObservationSim.MockObject._util import eObs, integrate_sed_bandpass, getNormFactorForSpecWithABMAG, getObservedSED, getABMAG,convolveGaussXorders
from ObservationSim.MockObject.SpecDisperser import SpecDisperser
from ObservationSim.MockObject.MockObject import MockObject

Fang Yuedong's avatar
Fang Yuedong committed
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
class Galaxy(MockObject):
    def __init__(self, param, rotation=None):
        super().__init__(param)
        self.thetaR = self.param["theta"]
        self.bfrac = self.param["bfrac"]
        self.hlr_disk = self.param["hlr_disk"]
        self.hlr_bulge = self.param["hlr_bulge"]

        # Extract ellipticity components
        self.e_disk = galsim.Shear(g=self.param["ell_disk"], beta=self.thetaR*galsim.degrees)
        self.e_bulge = galsim.Shear(g=self.param["ell_bulge"], beta=self.thetaR*galsim.degrees)
        self.e_total = galsim.Shear(g=self.param["ell_tot"], beta=self.thetaR*galsim.degrees)
        self.e1_disk, self.e2_disk = self.e_disk.g1, self.e_disk.g2
        self.e1_bulge, self.e2_bulge = self.e_bulge.g1, self.e_bulge.g2
        self.e1_total, self.e2_total = self.e_total.g1, self.e_total.g2

        if rotation is not None:
            self.rotateEllipticity(rotation)

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
    # def load_SED(self, survey_type, sed_path=None, cosids=None, objtypes=None, sed_templates=None, normFilter=None, target_filt=None):
    #     if survey_type == "photometric":
    #         norm_thr_rang_ids = normFilter['SENSITIVITY'] > 0.001
    #         if sed_templates is None:
    #             # Read SED data directly
    #             itype = objtypes[cosids==self.sed_type][0]
    #             sed_file = os.path.join(sed_path, itype + "_ID%s.sed"%(self.sed_type))
    #             if not os.path.exists(sed_file):
    #                 raise ValueError("!!! No SED found.")
    #             sed_data = Table.read(sed_file, format="ascii")
    #             wave, flux = sed_data["observedLambda"].data, sed_data["observedFlux"].data
    #         else:
    #             # Load SED from templates
    #             sed_data = sed_templates[self.sed_type]
    #             # redshift, intrinsic extinction
    #             sed_data = getObservedSED(
    #                 sedCat=sed_data, 
    #                 redshift=self.z, 
    #                 av=self.param['av'], 
    #                 redden=self.param['redden'])
    #             wave, flux = sed_data[0], sed_data[1] 
    #         flux_photon = flux * (wave / (cons.h.value * cons.c.value)) * 1e-13
    #         sed_photon = Table(np.array([wave, flux_photon]).T, names=('WAVELENGTH', 'FLUX'))
    #         # Get scaling factor for SED
    #         sedNormFactor = getNormFactorForSpecWithABMAG(ABMag=self.param['mag_use_normal'],
    #             spectrum=sed_photon,
    #             norm_thr=normFilter,
    #             sWave=np.floor(normFilter[norm_thr_rang_ids][0][0]),
    #             eWave=np.ceil(normFilter[norm_thr_rang_ids][-1][0]))
    #         sed_photon = np.array([sed_photon['WAVELENGTH'], sed_photon['FLUX']*sedNormFactor]).T
    #         # Convert to galsim.SED object
    #         spec = galsim.LookupTable(x=np.array(sed_photon[:, 0]), f=np.array(sed_photon[:, 1]), interpolant='nearest')
    #         self.sed = galsim.SED(spec, wave_type='A', flux_type='1', fast=False)
    #         # Get magnitude
    #         interFlux = integrate_sed_bandpass(sed=self.sed, bandpass=target_filt.bandpass_full)
    #         self.param['mag_%s'%target_filt.filter_type] = getABMAG(
    #             interFlux=interFlux, 
    #             bandpass=target_filt.bandpass_full)
    #         # print('mag_use_normal = ', self.param['mag_use_normal'])
    #         # print('mag_%s = '%target_filt.filter_type, self.param['mag_%s'%target_filt.filter_type])
    #         # print('redshift = %.3f'%(self.z))
    #         # print('sed_type = %d, av = %.2f, redden = %d'%(self.sed_type, self.param['av'], self.param['redden']))
Fang Yuedong's avatar
Fang Yuedong committed
73

74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
    #     elif survey_type == "spectroscopic":
    #         if sed_templates is None:
    #             self.sedPhotons(sed_path=sed_path, cosids=cosids, objtypes=objtypes)
    #         else:
    #             sed_data = sed_templates[self.sed_type]
    #             sed_data = getObservedSED(
    #                 sedCat=sed_data, 
    #                 redshift=self.z, 
    #                 av=self.param['av'], 
    #                 redden=self.param['redden'])
    #             speci = interpolate.interp1d(sed_data[0], sed_data[1])
    #             lamb = np.arange(2500, 10001 + 0.5, 0.5)
    #             y = speci(lamb)
    #             # erg/s/cm2/A --> photo/s/m2/A
    #             all_sed = y * lamb / (cons.h.value * cons.c.value) * 1e-13
    #             self.sed = Table(np.array([lamb, all_sed]).T, names=('WAVELENGTH', 'FLUX'))
Fang Yuedong's avatar
Fang Yuedong committed
90
91
92
93
94
95
96


    def unload_SED(self):
        """(Test) free up SED memory
        """
        del self.sed

97
    # def sedPhotons(self, sed_path, cosids, objtypes):
Fang Yuedong's avatar
Fang Yuedong committed
98

99
100
101
102
103
104
105
106
    #     itype = objtypes[cosids == self.sed_type][0]
    #     sed_file = os.path.join(sed_path, itype + "_ID%s.sed" % (self.sed_type))
    #     if not os.path.exists(sed_file):
    #         raise ValueError("!!! No SED found.")
    #     sed = Table.read(sed_file, format="ascii")
    #     spec_data = {}
    #     f_orig = sed["observedFlux"].data
    #     w_orig = sed["observedLambda"].data
Fang Yuedong's avatar
Fang Yuedong committed
107

108
109
110
111
112
113
    #     speci = interpolate.interp1d(w_orig, f_orig)
    #     lamb = np.arange(2500, 10001 + 0.5, 0.5)
    #     y = speci(lamb)
    #     # erg/s/cm2/A --> photo/s/m2/A
    #     all_sed = y * lamb / (cons.h.value * cons.c.value) * 1e-13
    #     self.sed = Table(np.array([lamb, all_sed]).T, names=('WAVELENGTH', 'FLUX'))
Fang Yuedong's avatar
Fang Yuedong committed
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
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260

    def getGSObj_multiband(self, tel, psf_list, bandpass_list, filt, nphotons_tot=None, g1=0, g2=0, exptime=150.):
        if len(psf_list) != len(bandpass_list):
            raise ValueError("!!!The number of PSF profiles and the number of bandpasses must be equal.")
        objs = []
        if nphotons_tot == None:
            nphotons_tot = self.getElectronFluxFilt(filt, tel, exptime)
        # print("nphotons_tot = ", nphotons_tot)

        try:
            full = integrate_sed_bandpass(sed=self.sed, bandpass=filt.bandpass_full)
        except Exception as e:
            print(e)
            return -1
        for i in range(len(bandpass_list)):
            bandpass = bandpass_list[i]
            try:
                sub = integrate_sed_bandpass(sed=self.sed, bandpass=bandpass)
            except Exception as e:
                print(e)
                return -1
            
            ratio = sub/full
            if not (ratio == -1 or (ratio != ratio)):
                nphotons = ratio * nphotons_tot
            else:
                return -1

            psf = psf_list[i]
            disk = galsim.Sersic(n=1.0, half_light_radius=self.hlr_disk, flux=1.0)
            disk_shape = galsim.Shear(g1=self.e1_disk, g2=self.e2_disk)
            disk = disk.shear(disk_shape)
            bulge = galsim.Sersic(n=4.0, half_light_radius=self.hlr_bulge, flux=1.0)
            bulge_shape = galsim.Shear(g1=self.e1_bulge, g2=self.e2_bulge)
            bulge = bulge.shear(bulge_shape)

            gal = self.bfrac * bulge + (1.0 - self.bfrac) * disk
            gal = gal.withFlux(nphotons)
            gal_shear = galsim.Shear(g1=g1, g2=g2)
            gal = gal.shear(gal_shear)
            gal = galsim.Convolve(psf, gal)
            objs.append(gal)
        final = galsim.Sum(objs)
        return final

    def drawObj_multiband(self, tel, pos_img, psf_model, bandpass_list, filt, chip, nphotons_tot=None, g1=0, g2=0, exptime=150.):
        if nphotons_tot == None:
            nphotons_tot = self.getElectronFluxFilt(filt, tel, exptime)
        # print("nphotons_tot = ", nphotons_tot)

        try:
            full = integrate_sed_bandpass(sed=self.sed, bandpass=filt.bandpass_full)
        except Exception as e:
            print(e)
            return False

        nphotons_sum = 0
        photons_list = []
        xmax, ymax = 0, 0

        # print('hlr_disk = %.4f, hlr_bulge = %.4f'%(self.hlr_disk, self.hlr_bulge))
        big_galaxy = False
        if self.hlr_disk > 3.0: # Very big galaxy
            big_galaxy = True

        # (TEST) Galsim Parameters
        if self.getMagFilter(filt) <= 15 and (not big_galaxy):
            folding_threshold = 5.e-4
        else:
            folding_threshold = 5.e-3
        gsp = galsim.GSParams(folding_threshold=folding_threshold)

        for i in range(len(bandpass_list)):
            bandpass = bandpass_list[i]

            try:
                sub = integrate_sed_bandpass(sed=self.sed, bandpass=bandpass)
            except Exception as e:
                print(e)
                # return False
                continue
            
            ratio = sub/full
            if not (ratio == -1 or (ratio != ratio)):
                nphotons = ratio * nphotons_tot
            else:
                # return False
                continue
            nphotons_sum += nphotons
            # print("nphotons_sub-band_%d = %.2f"%(i, nphotons))

            psf, pos_shear = psf_model.get_PSF(chip=chip, pos_img=pos_img, bandpass=bandpass, folding_threshold=folding_threshold)
            disk = galsim.Sersic(n=1.0, half_light_radius=self.hlr_disk, flux=1.0, gsparams=gsp)
            disk_shape = galsim.Shear(g1=self.e1_disk, g2=self.e2_disk)
            disk = disk.shear(disk_shape)
            bulge = galsim.Sersic(n=4.0, half_light_radius=self.hlr_bulge, flux=1.0, gsparams=gsp)
            bulge_shape = galsim.Shear(g1=self.e1_bulge, g2=self.e2_bulge)
            bulge = bulge.shear(bulge_shape)

            gal = self.bfrac * bulge + (1.0 - self.bfrac) * disk
            gal = gal.withFlux(nphotons)
            gal_shear = galsim.Shear(g1=g1, g2=g2)
            gal = gal.shear(gal_shear)

            if self.hlr_disk < 10.0: # Not apply PSF for very big galaxy
                gal = galsim.Convolve(psf, gal)

            # Use (explicit) stamps to draw
            stamp = gal.drawImage(wcs=self.localWCS, method='phot', offset=self.offset, save_photons=True)
            xmax = max(xmax, stamp.xmax)
            ymax = max(ymax, stamp.ymax)
            photons = stamp.photons
            photons.x += self.x_nominal
            photons.y += self.y_nominal
            photons_list.append(photons)

        # print('xmax = %d, ymax = %d '%(xmax, ymax))

        stamp = galsim.ImageF(int(xmax*1.1), int(ymax*1.1))
        stamp.wcs = self.localWCS
        stamp.setCenter(self.x_nominal, self.y_nominal)
        bounds = stamp.bounds & chip.img.bounds
        stamp[bounds] = chip.img[bounds]

        if not big_galaxy:
            for i in range(len(photons_list)):
                if i == 0:
                    chip.sensor.accumulate(photons_list[i], stamp)
                else:
                    chip.sensor.accumulate(photons_list[i], stamp, resume=True)
        else:
            sensor = galsim.Sensor()
            for i in range(len(photons_list)):
                if i == 0:
                    sensor.accumulate(photons_list[i], stamp)
                else:
                    sensor.accumulate(photons_list[i], stamp, resume=True)

        # print(stamp.array.sum())
        # chip.img[bounds] += stamp[bounds]
        chip.img[bounds] = stamp[bounds]
        # print("nphotons_sum = ", nphotons_sum)
        del photons_list
        del stamp
        return True, pos_shear

    def drawObj_slitless(self, tel, pos_img, psf_model, bandpass_list, filt, chip, nphotons_tot=None, g1=0, g2=0,
Xin Zhang's avatar
Xin Zhang committed
261
                         exptime=150., normFilter=None, grating_split_pos=3685):
Fang Yuedong's avatar
Fang Yuedong committed
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282

        norm_thr_rang_ids = normFilter['SENSITIVITY'] > 0.001
        sedNormFactor = getNormFactorForSpecWithABMAG(ABMag=self.param['mag_use_normal'], spectrum=self.sed,
                                                      norm_thr=normFilter,
                                                      sWave=np.floor(normFilter[norm_thr_rang_ids][0][0]),
                                                      eWave=np.ceil(normFilter[norm_thr_rang_ids][-1][0]))
        if sedNormFactor == 0:
            return False
        normalSED = Table(np.array([self.sed['WAVELENGTH'], self.sed['FLUX'] * sedNormFactor]).T,
                          names=('WAVELENGTH', 'FLUX'))

        big_galaxy = False
        if self.hlr_disk > 3.0: # Very big galaxy
            big_galaxy = True

        if self.getMagFilter(filt) <= 15 and (not big_galaxy):
            folding_threshold = 5.e-4
        else:
            folding_threshold = 5.e-3
        gsp = galsim.GSParams(folding_threshold=folding_threshold)
        # nphotons_sum = 0
Zhang Xin's avatar
Zhang Xin committed
283
        xOrderSigPlus = {'A':1.3909419820029296,'B':1.4760376591236062,'C':4.035447379743442,'D':5.5684364343742825,'E':16.260021029735388}
Xin Zhang's avatar
Xin Zhang committed
284
        grating_split_pos_chip = chip.bound.xmin + grating_split_pos
Fang Yuedong's avatar
Fang Yuedong committed
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
        for i in range(len(bandpass_list)):
            bandpass = bandpass_list[i]

            psf, pos_shear = psf_model.get_PSF(chip=chip, pos_img=pos_img, bandpass=bandpass, folding_threshold=folding_threshold)
            disk = galsim.Sersic(n=1.0, half_light_radius=self.hlr_disk, flux=1.0, gsparams=gsp)
            disk_shape = galsim.Shear(g1=self.e1_disk, g2=self.e2_disk)
            disk = disk.shear(disk_shape)
            bulge = galsim.Sersic(n=4.0, half_light_radius=self.hlr_bulge, flux=1.0, gsparams=gsp)
            bulge_shape = galsim.Shear(g1=self.e1_bulge, g2=self.e2_bulge)
            bulge = bulge.shear(bulge_shape)

            gal = self.bfrac * bulge + (1.0 - self.bfrac) * disk
            gal = gal.withFlux(tel.pupil_area * exptime)
            gal_shear = galsim.Shear(g1=g1, g2=g2)
            gal = gal.shear(gal_shear)
            gal = galsim.Convolve(psf, gal)

            starImg = gal.drawImage(wcs=self.localWCS)

            origin_star = [self.y_nominal - (starImg.center.y - starImg.ymin),
                           self.x_nominal - (starImg.center.x - starImg.xmin)]

Xin Zhang's avatar
Xin Zhang committed
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
            gal_origin = [origin_star[0], origin_star[1]]
            gal_end = [origin_star[0] + starImg.array.shape[0] - 1, origin_star[1] + starImg.array.shape[1] - 1]

            if gal_origin[1] < grating_split_pos_chip < gal_end[1]:
                subSlitPos = int(grating_split_pos_chip - gal_origin[1] + 1)
                ## part img disperse

                subImg_p1 = starImg.array[:, 0:subSlitPos]
                star_p1 = galsim.Image(subImg_p1)
                origin_p1 = origin_star
                xcenter_p1 = min(self.x_nominal,grating_split_pos_chip-1) - chip.bound.xmin
                ycenter_p1 = self.y_nominal-chip.bound.ymin

                sdp_p1 = SpecDisperser(orig_img=star_p1, xcenter=xcenter_p1,
                                    ycenter=ycenter_p1, origin=origin_p1,
                                    tar_spec=normalSED,
                                    band_start=bandpass.blue_limit * 10, band_end=bandpass.red_limit * 10,
                                    conf=chip.sls_conf[0],
                                    isAlongY=0)

                self.addSLStoChipImage(sdp=sdp_p1, chip=chip, xOrderSigPlus = xOrderSigPlus)

                subImg_p2 = starImg.array[:, subSlitPos+1:starImg.array.shape[1]]
                star_p2 = galsim.Image(subImg_p2)
                origin_p2 = [origin_star[0], grating_split_pos_chip]
                xcenter_p2 = max(self.x_nominal, grating_split_pos_chip - 1) - chip.bound.xmin
                ycenter_p2 = self.y_nominal - chip.bound.ymin

                sdp_p2 = SpecDisperser(orig_img=star_p2, xcenter=xcenter_p2,
                                       ycenter=ycenter_p2, origin=origin_p2,
                                       tar_spec=normalSED,
                                       band_start=bandpass.blue_limit * 10, band_end=bandpass.red_limit * 10,
                                       conf=chip.sls_conf[1],
                                       isAlongY=0)

                self.addSLStoChipImage(sdp=sdp_p2, chip=chip, xOrderSigPlus = xOrderSigPlus)

                del sdp_p1
                del sdp_p2
            elif grating_split_pos_chip<=gal_origin[1]:
                sdp = SpecDisperser(orig_img=starImg, xcenter=self.x_nominal - chip.bound.xmin,
                                    ycenter=self.y_nominal - chip.bound.ymin, origin=origin_star,
                                    tar_spec=normalSED,
                                    band_start=bandpass.blue_limit * 10, band_end=bandpass.red_limit * 10,
                                    conf=chip.sls_conf[1],
                                    isAlongY=0)
                self.addSLStoChipImage(sdp=sdp, chip=chip, xOrderSigPlus = xOrderSigPlus)
                del sdp
            elif grating_split_pos_chip>=gal_end[1]:
                sdp = SpecDisperser(orig_img=starImg, xcenter=self.x_nominal - chip.bound.xmin,
                                    ycenter=self.y_nominal - chip.bound.ymin, origin=origin_star,
                                    tar_spec=normalSED,
                                    band_start=bandpass.blue_limit * 10, band_end=bandpass.red_limit * 10,
                                    conf=chip.sls_conf[0],
                                    isAlongY=0)
                self.addSLStoChipImage(sdp=sdp, chip=chip, xOrderSigPlus = xOrderSigPlus)
                del sdp
Fang Yuedong's avatar
Fang Yuedong committed
364

Xin Zhang's avatar
Xin Zhang committed
365
            # print(self.y_nominal, starImg.center.y, starImg.ymin)
Fang Yuedong's avatar
Fang Yuedong committed
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
            del psf
        return True, pos_shear

    def getGSObj(self, psf, g1=0, g2=0, flux=None, filt=None, tel=None, exptime=150.):
        if flux == None:
            flux = self.getElectronFluxFilt(filt, tel, exptime)
        disk = galsim.Sersic(n=1.0, half_light_radius=self.hlr_disk, flux=1.0)
        disk_shape = galsim.Shear(g1=self.e1_disk, g2=self.e2_disk)
        disk = disk.shear(disk_shape)

        bulge = galsim.Sersic(n=4.0, half_light_radius=self.hlr_bulge, flux=1.0)
        bulge_shape = galsim.Shear(g1=self.e1_bulge, g2=self.e2_bulge)
        bulge = bulge.shear(bulge_shape)

        gal = self.bfrac * bulge + (1.0 - self.bfrac) * disk
        gal = gal.withFlux(flux)
        gal_shear = galsim.Shear(g1=g1, g2=g2)
        gal = gal.shear(gal_shear)
        final = galsim.Convolve(psf, gal)
        return final

    def rotateEllipticity(self, rotation):
        if rotation == 1:
            self.e1_disk, self.e2_disk, self.e1_bulge, self.e2_bulge, self.e1_total, self.e2_total = -self.e2_disk, self.e1_disk, -self.e2_bulge, self.e1_bulge, -self.e2_total, self.e1_total
        if rotation == 2:
            self.e1_disk, self.e2_disk, self.e1_bulge, self.e2_bulge, self.e1_total, self.e2_total = -self.e1_disk, -self.e2_disk, -self.e1_bulge, -self.e2_bulge, -self.e1_total, -self.e2_total
        if rotation == 3:
            self.e1_disk, self.e2_disk, self.e1_bulge, self.e2_bulge, self.e1_total, self.e2_total = self.e2_disk, -self.e1_disk, self.e2_bulge, -self.e1_bulge, self.e2_total, -self.e1_total

    def drawObject(self, img, final, noise_level=0.0, flux=None, filt=None, tel=None, exptime=150.):
        """ Override the method in parent class 
        Need to constrain the size of image stamp for extended objects
        """
        isUpdated = True
        if flux == None:
            flux = self.getElectronFluxFilt(filt, tel, exptime)
        stamp = final.drawImage(wcs=self.localWCS, offset=self.offset)
        stamp_arr = stamp.array
        mask = (stamp_arr >= 0.001*noise_level) # why 0.001?
        err = int(np.sqrt(mask.sum()))
        if np.mod(err, 2) == 1:
            err += 1
        # if err == 1:
        if err == 0:
            subSize = 16 # why 16?
        else:
            subSize = max([err, 16])
            fluxRatio = flux / stamp_arr[mask].sum()
            final = final.withScaledFlux(fluxRatio)

        imgSub = galsim.ImageF(subSize, subSize)

        # Draw with FFT
        # stamp = final.drawImage(image=imgSub, wcs=self.localWCS, offset=self.offset)

        # Draw with Photon Shoot
        stamp = final.drawImage(image=imgSub, wcs=self.localWCS, method='phot', offset=self.offset)
        
        stamp.setCenter(self.x_nominal, self.y_nominal)
        if np.sum(np.isnan(stamp.array)) >= 1:
            stamp.setZero()
        bounds = stamp.bounds & img.bounds
        if bounds.area() == 0:
            isUpdated = False
        else:
            img[bounds] += stamp[bounds]
        return img, stamp, isUpdated


    def getObservedEll(self, g1=0, g2=0):
        e1_obs, e2_obs, e_obs, theta = eObs(self.e1_total, self.e2_total, g1, g2)
Zhang Xin's avatar
Zhang Xin committed
437
        return self.e1_total, self.e2_total, g1, g2, e1_obs, e2_obs