def image_data():
dims = (13, 15)
rng = np.random.RandomState(seed=10)
data = {}
for key in ['image', 'weight', 'bmask', 'ormask', 'noise']:
data[key] = rng.normal(size=dims)
if key == 'weight':
data[key] = np.exp(data[key])
elif key in ['bmask', 'ormask']:
data[key] = np.clip(data[key] * 100, 0, np.inf).astype(np.int32)
data['jacobian'] = DiagonalJacobian(x=7, y=6, scale=0.25)
data['meta'] = {'pi': 3.14}
data['psf'] = Observation(image=rng.normal(size=dims),
meta={'ispsf': True})
data['psf_gmix'] = Observation(image=rng.normal(size=dims),
meta={'ispsf': True},
gmix=GMix(pars=rng.uniform(size=6)))
data['gmix'] = GMix(pars=rng.uniform(size=6))
return data
def test_observation_set(image_data):
obs = Observation(image=image_data['image'],
weight=image_data['weight'],
bmask=image_data['bmask'],
ormask=image_data['ormask'],
noise=image_data['noise'],
jacobian=image_data['jacobian'],
gmix=image_data['gmix'],
psf=image_data['psf'],
meta=image_data['meta'])
rng = np.random.RandomState(seed=11)
new_arr = rng.normal(size=image_data['image'].shape)
assert np.all(obs.image != new_arr)
obs.image = new_arr
assert np.all(obs.image == new_arr)
new_arr = np.exp(rng.normal(size=image_data['image'].shape))
assert np.all(obs.weight != new_arr)
obs.weight = new_arr
assert np.all(obs.weight == new_arr)
new_arr = (np.exp(rng.normal(size=image_data['image'].shape)) *
100).astype(np.int32)
assert np.all(obs.bmask != new_arr)
obs.bmask = new_arr
assert np.all(obs.bmask == new_arr)
new_arr = (np.exp(rng.normal(size=image_data['image'].shape)) *
100).astype(np.int32)
assert np.all(obs.ormask != new_arr)
obs.ormask = new_arr
assert np.all(obs.ormask == new_arr)
new_arr = rng.normal(size=image_data['image'].shape)
assert np.all(obs.noise != new_arr)
obs.noise = new_arr
assert np.all(obs.noise == new_arr)
new_jac = DiagonalJacobian(x=8, y=13, scale=1.2)
assert new_jac.get_galsim_wcs() != obs.jacobian.get_galsim_wcs()
obs.jacobian = new_jac
assert new_jac.get_galsim_wcs() == obs.jacobian.get_galsim_wcs()
new_meta = {'new': 5}
assert obs.meta != new_meta
obs.meta = new_meta
assert obs.meta == new_meta
new_meta = {'blue': 10}
new_meta.update(obs.meta)
assert obs.meta != new_meta
obs.update_meta_data({'blue': 10})
assert obs.meta == new_meta
new_gmix = GMix(pars=rng.uniform(size=6))
assert np.all(obs.gmix.get_full_pars() != new_gmix.get_full_pars())
obs.gmix = new_gmix
assert np.all(obs.gmix.get_full_pars() == new_gmix.get_full_pars())
new_psf = Observation(image=rng.normal(size=obs.psf.image.shape),
meta={'ispsf': True})
assert np.all(obs.psf.image != new_psf.image)
obs.psf = new_psf
assert np.all(obs.psf.image == new_psf.image)
def _get_nbrs_data(self,cen_id,band,cutout_index):
"""
return cen and nbrs info for a given cen_id
Parameters
----------
cen_id: id of central object
band: integer for band (e.g., 0 for griz MOF)
cutout_index: index of epoch in MEDS file for nbrs
Returns
-------
cen_ind: index of central intp returned fit_data
cen_flags: flags for central from nbrs_data
cen_psf_gmix: an ngmix GMix for PSF of central
cen_jac: an ngmix Jacobian for the central
nbrs_inds: list of indexes into returned fit_data of nbrs
nbrs_flags: flags of nbrs (only use flags == 0)
nbrs_psf_gmixes: a list of ngmix GMixes of the PSF models for the nbrs
nbrs_jacs: a list of ngmix Jacobians for the nbrs
fit_data: entries of input fit_data corresponding to cen_ind and nbrs_inds
pixel_scale: the pixel_scale of the jacobian associated with the image
will return None if thete is an error or if not all quantities are found
"""
fit_inds = []
# get cen stuff
q, = numpy.where(self.fit_data['id'] == cen_id)
if len(q) != 1:
return None
cen_ind = 0
fit_inds.append(q[0])
q, = numpy.where(
(self.nbrs_data['id'] == cen_id)
& (self.nbrs_data['nbr_id'] == cen_id)
& (self.nbrs_data['band_num'] == band)
& (self.nbrs_data['cutout_index'] == cutout_index)
)
if len(q) != 1:
return None
ind = q[0]
cen_flags = self.nbrs_data['nbr_flags'][ind]
if cen_flags == 0:
cen_jac = Jacobian(
row=self.nbrs_data['nbr_jac_row0'][ind],
col=self.nbrs_data['nbr_jac_col0'][ind],
dudrow=self.nbrs_data['nbr_jac_dudrow'][ind],
dudcol=self.nbrs_data['nbr_jac_dudcol'][ind],
dvdrow=self.nbrs_data['nbr_jac_dvdrow'][ind],
dvdcol=self.nbrs_data['nbr_jac_dvdcol'][ind]
)
cen_psf_gmix = GMix(pars=self.nbrs_data['nbr_psf_fit_pars'][ind,:])
else:
cen_psf_gmix = None
cen_jac = None
pixel_scale = self.nbrs_data['pixel_scale'][ind]
# get nbr ids
q, = numpy.where(
(self.nbrs_data['id'] == cen_id)
& (self.nbrs_data['nbr_id'] != cen_id)
& (self.nbrs_data['band_num'] == band)
& (self.nbrs_data['cutout_index'] == cutout_index)
)
if len(q) == 0:
return None
n1 = len(numpy.unique(self.nbrs_data['nbr_id'][q]))
n2 = len(self.nbrs_data['nbr_id'][q])
mess=("nbrs list for given band %d and cutout_index "
"%d for object %d is not unique!" % (band,cutout_index,cen_id))
assert n1==n2, mess
nbrs_inds = [i+1 for i in xrange(len(q))]
# find location of nbrs in fit_data
for nbr_id in self.nbrs_data['nbr_id'][q]:
qq, = numpy.where(self.fit_data['id'] == nbr_id)
if len(qq) != 1:
return None
fit_inds.append(qq[0])
# doing fit data
fit_data = self.fit_data[fit_inds]
# build flag, psf and jacobian lists
nbrs_flags = []
nbrs_psf_gmixes = []
nbrs_jacs = []
for i,ind in enumerate(q):
check=fit_data['id'][i+1] == self.nbrs_data['nbr_id'][ind]
assert check,"Matching of nbr_ids to ids in fit_data did not work!"
nbrs_flags.append(self.nbrs_data['nbr_flags'][ind])
if nbrs_flags[-1] == 0:
nbrs_jacs.append(
Jacobian(
row=self.nbrs_data['nbr_jac_row0'][ind],
col=self.nbrs_data['nbr_jac_col0'][ind],
dudrow=self.nbrs_data['nbr_jac_dudrow'][ind],
dudcol=self.nbrs_data['nbr_jac_dudcol'][ind],
dvdrow=self.nbrs_data['nbr_jac_dvdrow'][ind],
dvdcol=self.nbrs_data['nbr_jac_dvdcol'][ind]
)
)
tgm = GMix(pars=self.nbrs_data['nbr_psf_fit_pars'][ind,:])
nbrs_psf_gmixes.append(tgm)
else:
nbrs_jacs.append(None)
nbrs_psf_gmixes.append(None)
return cen_ind, cen_flags, cen_psf_gmix, cen_jac, \
nbrs_inds, nbrs_flags, nbrs_psf_gmixes, nbrs_jacs, \
fit_data, pixel_scale
def render_central(self,
cen_id,
meds_data,
mindex,
cutout_index,
model,
band,
image_shape):
"""
just render the central. This is needed since Matt does not
store data for the central i the nbrs data structure if there
are no neighbors
you must send epoch_data= to the constructor
"""
import esutil as eu
#
# set and check some parameters
#
if self.epoch_data is None:
raise RuntimeError(
"send epoch_data= to the constructor to render centrals"
)
pars_tag = '%s_max_pars' % model
if model == 'cm':
fracdev_tag = 'cm_fracdev'
TdByTe_tag = 'cm_TdByTe'
else:
fracdev_tag=None
TdByTe_tag=None
#
# get data needed to render the central
#
we,=numpy.where(
(self.epoch_data['id'] == cen_id)
& (self.epoch_data['cutout_index'] == cutout_index)
)
if we.size == 0:
print(" central not in epochs data")
return None
psf_pars=self.epoch_data['psf_fit_pars'][we[0],:]
psf_gmix = GMix(pars=psf_pars)
e1,e2,T=psf_gmix.get_e1e2T()
if T <= 0.0:
print(" bad psf fit for central")
return None
pixel_scale=self.epoch_data['pixel_scale'][we[0]]
jdict=meds_data.get_jacobian(mindex, cutout_index)
jac=Jacobian(
row=jdict['row0'],
col=jdict['col0'],
dudrow=jdict['dudrow'],
dudcol=jdict['dudcol'],
dvdrow=jdict['dvdrow'],
dvdcol=jdict['dvdcol'],
)
# fit information for central
wfit, = numpy.where(self.fit_data['id'] == cen_id)
if wfit.size != 1:
# not sure why this would happen
print(" Central not found in fit_data")
return None
fit_data = self.fit_data[wfit]
cen_flags=fit_data['flags'][0]
if cen_flags != 0:
print(" bad central fit:",cen_flags)
return None
cen_img = RenderNGmixNbrs._render_single(
model,
band,
image_shape,
pars_tag,
fit_data,
psf_gmix,
jac,
fracdev_tag=fracdev_tag,TdByTe_tag=TdByTe_tag,
)
if cen_img is None:
return None
return cen_img, pixel_scale
def go(self, gmix_guess, sky):
"""
Run the em algorithm from the input starting guesses
parameters
----------
gmix_guess: GMix
A gaussian mixture (GMix or child class) representing a starting
guess for the algorithm. This should be *before* psf convolution.
sky: number
The sky value added to the image
"""
if hasattr(self, '_gm'):
del self._gm
del self._gm_conv
obs = self._obs
# makes a copy
if not obs.has_psf() or not obs.psf.has_gmix():
logger.debug('NO PSF SET')
gmix_psf = ngmix.GMixModel([0., 0., 0., 0., 0., 1.0], 'gauss')
else:
gmix_psf = obs.psf.gmix
gmix_psf.set_flux(1.0)
conf = self._make_conf()
conf['sky'] = sky
gm = gmix_guess.copy()
gm_conv = gm.convolve(gmix_psf)
sums = self._make_sums(len(gm))
pixels = obs.pixels.copy()
if np.any(pixels['ierr'] <= 0.0):
fill_zero_weight = True
else:
fill_zero_weight = False
flags = 0
try:
numiter, fdiff, sky = self._runner(
conf,
pixels,
sums,
gm.get_data(),
gmix_psf.get_data(),
gm_conv.get_data(),
fill_zero_weight=fill_zero_weight,
)
pars = gm.get_full_pars()
pars_conv = gm_conv.get_full_pars()
self._gm = GMix(pars=pars)
self._gm_conv = GMix(pars=pars_conv)
if numiter >= self.maxiter:
flags = EM_MAXITER
message = 'maxit'
else:
message = 'OK'
result = {
'flags': flags,
'numiter': numiter,
'fdiff': fdiff,
'sky': sky,
'message': message,
}
except (GMixRangeError, ZeroDivisionError) as err:
message = str(err)
logger.info(message)
result = {
'flags': EM_RANGE_ERROR,
'message': message,
}
self._result = result
def _get_nbrs_data(self, cen_id, band, cutout_index):
"""
return cen and nbrs info for a given cen_id
Parameters
----------
cen_id: id of central object
band: integer for band (e.g., 0 for griz MOF)
cutout_index: index of epoch in MEDS file for nbrs
Returns
-------
cen_ind: index of central intp returned fit_data
cen_flags: flags for central from nbrs_data
cen_psf_gmix: an ngmix GMix for PSF of central
cen_jac: an ngmix Jacobian for the central
nbrs_inds: list of indexes into returned fit_data of nbrs
nbrs_flags: flags of nbrs (only use flags == 0)
nbrs_psf_gmixes: a list of ngmix GMixes of the PSF models for the nbrs
nbrs_jacs: a list of ngmix Jacobians for the nbrs
fit_data: entries of input fit_data corresponding to cen_ind and nbrs_inds
pixel_scale: the pixel_scale of the jacobian associated with the image
will return None if thete is an error or if not all quantities are found
"""
fit_inds = []
# get cen stuff
qids, = numpy.where(self.fit_ids == cen_id)
if len(qids) != 1:
print(" no cen match in fits")
return None
cen_ind = 0
fit_inds.append(qids[0])
qnbrs_ids, = numpy.where(self.nbrs_ids == cen_id)
if len(qnbrs_ids) == 0:
return None
# extract just the data we need
nbrs_data = self._fits['nbrs_data'][qnbrs_ids]
qcen, = numpy.where((nbrs_data['nbr_id'] == cen_id)
& (nbrs_data['band_num'] == band)
& (nbrs_data['cutout_index'] == cutout_index))
if len(qcen) != 1:
return None
ind = qcen[0]
cen_flags = nbrs_data['nbr_flags'][ind]
if cen_flags == 0:
cen_jac = Jacobian(row=nbrs_data['nbr_jac_row0'][ind],
col=nbrs_data['nbr_jac_col0'][ind],
dudrow=nbrs_data['nbr_jac_dudrow'][ind],
dudcol=nbrs_data['nbr_jac_dudcol'][ind],
dvdrow=nbrs_data['nbr_jac_dvdrow'][ind],
dvdcol=nbrs_data['nbr_jac_dvdcol'][ind])
cen_psf_gmix = GMix(pars=nbrs_data['nbr_psf_fit_pars'][ind, :])
else:
cen_psf_gmix = None
cen_jac = None
pixel_scale = nbrs_data['pixel_scale'][ind]
# get nbr ids
qnbr, = numpy.where((nbrs_data['nbr_id'] != cen_id)
& (nbrs_data['band_num'] == band)
& (nbrs_data['cutout_index'] == cutout_index))
if len(qnbr) == 0:
return None
n1 = len(numpy.unique(nbrs_data['nbr_id'][qnbr]))
n2 = len(nbrs_data['nbr_id'][qnbr])
if n1 != n2:
# not sure why this happens, but it seems to be very rare.
# For now ignore
mess = ("nbrs list for given band %d and cutout_index "
"%d for object %d is not unique! %d:%d" %
(band, cutout_index, cen_id, n1, n2))
print(nbrs_data['nbr_id'][qnbr])
print(mess)
return None
#raise RuntimeError(mess)
nbrs_inds = [i + 1 for i in xrange(len(qnbr))]
# find location of nbrs in fit_data
for nbr_id in nbrs_data['nbr_id'][qnbr]:
qq, = numpy.where(self.fit_ids == nbr_id)
if len(qq) != 1:
print(" nbr", nbr_id, "not found in fit data")
return None
fit_inds.append(qq[0])
# doing fit data
# the data come out in row order, we need to reorder
fit_data_list = []
for tmp_ind in fit_inds:
fit_data_list.append(self._fits['model_fits'][tmp_ind])
fit_data = eu.numpy_util.combine_arrlist(fit_data_list)
# build flag, psf and jacobian lists
nbrs_flags = []
nbrs_psf_gmixes = []
nbrs_jacs = []
for i, ind in enumerate(qnbr):
check = fit_data['id'][i + 1] == nbrs_data['nbr_id'][ind]
assert check, "Matching of nbr_ids to ids in fit_data did not work!"
nbrs_flags.append(nbrs_data['nbr_flags'][ind])
if nbrs_flags[-1] == 0:
nbrs_jacs.append(
Jacobian(row=nbrs_data['nbr_jac_row0'][ind],
col=nbrs_data['nbr_jac_col0'][ind],
dudrow=nbrs_data['nbr_jac_dudrow'][ind],
dudcol=nbrs_data['nbr_jac_dudcol'][ind],
dvdrow=nbrs_data['nbr_jac_dvdrow'][ind],
dvdcol=nbrs_data['nbr_jac_dvdcol'][ind]))
tgm = GMix(pars=nbrs_data['nbr_psf_fit_pars'][ind, :])
nbrs_psf_gmixes.append(tgm)
else:
nbrs_jacs.append(None)
nbrs_psf_gmixes.append(None)
return cen_ind, cen_flags, cen_psf_gmix, cen_jac, \
nbrs_inds, nbrs_flags, nbrs_psf_gmixes, nbrs_jacs, \
fit_data, pixel_scale