def test_gmix_convolve(model):
row, col, g1, g2, T, flux = -0.5, -0.4, -0.2, 0.3, 2.8, 1
pars = np.array([row, col, g1, g2, T, flux])
gm = make_gmix_model(pars, model)
kernel = make_gmix_model([0, 0, 0.1, -0.1, 2.5, 1], 'turb')
gs_obj = galsim.Convolve(
[gm.make_galsim_object(),
kernel.make_galsim_object()])
cgm = gm.convolve(kernel)
assert len(cgm) == 3 * len(gm)
# test making an image
# only works for scale=1 because ngmix uses pixel and galsim is in
# world coords
dims = (13, 11)
jac = DiagonalJacobian(row=6, col=5, scale=1)
gs_jac = jac.get_galsim_wcs()
im = cgm.make_image(dims, jacobian=jac)
gs_im = gs_obj.drawImage(nx=dims[1],
ny=dims[0],
wcs=gs_jac,
method='no_pixel').array
# not perfect but close
assert np.allclose(im,
gs_im,
atol=3e-6 if model == 'dev' else 1e-6,
rtol=0)
def test_gmix_get_pars(model, row, col, flux, g1, g2, T):
e1, e2 = g1g2_to_e1e2(g1, g2)
sigma = np.sqrt(T / 2)
pars = np.array([row, col, g1, g2, T, flux])
gm = make_gmix_model(pars, model)
# this will always get coverage
assert model in ['gauss', 'exp', 'dev', 'turb']
if model == 'gauss':
assert len(gm) == 1
elif model == 'exp':
assert len(gm) == 6
elif model == 'dev':
assert len(gm) == 10
elif model == 'turb':
assert len(gm) == 3
assert np.allclose(gm.get_cen(), np.array([row, col]))
assert np.allclose(gm.get_flux(), flux)
assert np.allclose(gm.get_T(), T)
assert np.allclose(gm.get_sigma(), sigma)
assert np.allclose(gm.get_e1e2T(), [e1, e2, T])
assert np.allclose(gm.get_g1g2T(), [g1, g2, T])
assert np.allclose(gm.get_e1e2sigma(), [e1, e2, sigma])
assert np.allclose(gm.get_g1g2sigma(), [g1, g2, sigma])
# we are going to simply check a few things here since the tests
# above use all of this data too
data = gm.get_data()
full_pars = gm.get_full_pars()
if model == 'gauss':
assert len(data) == 1
assert len(full_pars) == 6
elif model == 'exp':
assert len(data) == 6
assert len(full_pars) == 6 * 6
elif model == 'dev':
assert len(data) == 10
assert len(full_pars) == 10 * 6
elif model == 'turb':
assert len(data) == 3
assert len(full_pars) == 3 * 6
assert np.allclose(np.sum(full_pars[::6]), flux)
assert np.allclose(
np.sum(full_pars[::6] * full_pars[1::6]) / np.sum(full_pars[::6]), row)
assert np.allclose(
np.sum(full_pars[::6] * full_pars[2::6]) / np.sum(full_pars[::6]), col)
assert np.allclose(np.sum(data['p']), flux)
assert np.allclose(
np.sum(data['p'] * data['row']) / np.sum(data['p']), row)
assert np.allclose(
np.sum(data['p'] * data['col']) / np.sum(data['p']), col)
gmfull = make_gmix_model(full_pars, 'full')
assert np.all(gmfull.get_full_pars() == full_pars)
def test_gmix_make_galsim_make_image(model):
row, col, g1, g2, T, flux = -0.5, -0.4, -0.2, 0.3, 0.8, 56
pars = np.array([row, col, g1, g2, T, flux])
gm = make_gmix_model(pars, model)
gs_obj = gm.make_galsim_object()
assert np.allclose(gs_obj.flux, 56)
assert np.allclose(gs_obj.centroid.x, -0.4)
assert np.allclose(gs_obj.centroid.y, -0.5)
if isinstance(gs_obj, galsim.Sum):
obj_list = gs_obj.obj_list
data = gm.get_data()
assert len(obj_list) == len(data)
for i in range(len(obj_list)):
assert np.allclose(data['p'][i], obj_list[i].flux)
assert np.allclose(obj_list[i].centroid.x, -0.4)
assert np.allclose(obj_list[i].centroid.y, -0.5)
# test making an image
# only works for scale=1 because ngmix uses pixel and galsim is in
# world coords
dims = (213, 113)
jac = DiagonalJacobian(row=106, col=56, scale=1)
gs_jac = jac.get_galsim_wcs()
im = gm.make_image(dims, jacobian=jac)
gs_im = gs_obj.drawImage(nx=dims[1],
ny=dims[0],
wcs=gs_jac,
method='no_pixel').array
assert np.allclose(im, gs_im)
def test_gmix_get_sheared(model):
row, col, g1, g2, T, flux = -0.5, -0.4, 0, 0, 0.8, 56
pars = np.array([row, col, g1, g2, T, flux])
gm = make_gmix_model(pars, model)
assert np.allclose(gm.get_flux(), flux)
assert np.allclose(gm.get_g1g2T(), [g1, g2, T])
assert np.allclose(gm.get_cen(), [row, col])
new_gm = gm.get_sheared(-0.8, 0.01)
assert np.allclose(gm.get_flux(), flux)
assert np.allclose(gm.get_g1g2T(), [g1, g2, T])
assert np.allclose(gm.get_cen(), [row, col])
assert np.allclose(new_gm.get_flux(), flux)
assert np.allclose(new_gm.get_cen(), [row, col])
assert np.allclose(new_gm.get_g1g2T()[0:2], [-0.8, 0.01])
shear = Shape(g1=-0.8, g2=0.01)
new_gm = gm.get_sheared(shear)
assert np.allclose(gm.get_flux(), flux)
assert np.allclose(gm.get_g1g2T(), [g1, g2, T])
assert np.allclose(gm.get_cen(), [row, col])
assert np.allclose(new_gm.get_flux(), flux)
assert np.allclose(new_gm.get_cen(), [row, col])
assert np.allclose(new_gm.get_g1g2T()[0:2], [-0.8, 0.01])
def test_gmix_set_cen(model):
row, col, g1, g2, T, flux = -0.5, -0.4, -0.2, 0.3, 0.8, 56
pars = np.array([row, col, g1, g2, T, flux])
gm = make_gmix_model(pars, model)
assert np.allclose(gm.get_cen(), [row, col])
gm.set_cen(0.1, -0.1)
assert np.allclose(gm.get_cen(), [0.1, -0.1])
assert np.allclose(gm.get_g1g2T(), [g1, g2, T])
assert np.allclose(gm.get_flux(), flux)
def test_gmix_convolve_gauss():
row, col, g1, g2, T, flux = -0.5, -0.4, -0.2, 0.3, 2.8, 5
pars = np.array([row, col, g1, g2, T, flux])
gm = make_gmix_model(pars, 'gauss')
krow, kcol, kg1, kg2, kT, kflux = 0, 1.0, 0.5, 0.4, 1.2, 6
kpars = np.array([krow, kcol, kg1, kg2, kT, kflux])
kernel = make_gmix_model(kpars, 'gauss')
cgm = gm.convolve(kernel)
mom = ngmix.moments.g2mom(g1, g2, T)
kmom = ngmix.moments.g2mom(kg1, kg2, kT)
cmom = (mom[0] + kmom[0], mom[1] + kmom[1], mom[2] + kmom[2])
cg1, cg2, cT = ngmix.moments.mom2g(*cmom)
# convolve ignores the PSF center and flux
assert np.allclose(cgm.get_cen()[0], row)
assert np.allclose(cgm.get_cen()[1], col)
assert np.allclose(cgm.get_flux(), flux)
assert np.allclose(cgm.get_g1g2T(), [cg1, cg2, cT])
def test_gmix_reset(model):
row, col, g1, g2, T, flux = -0.5, -0.4, -0.2, 0.3, 0.8, 56
pars = np.array([row, col, g1, g2, T, flux])
gm = make_gmix_model(pars, model)
assert np.allclose(gm.get_flux(), flux)
assert np.allclose(gm.get_g1g2T(), [g1, g2, T])
assert np.allclose(gm.get_cen(), [row, col])
gm.reset()
assert np.allclose(gm.get_full_pars(), 0)
def test_gmix_fill_pars(model):
row, col, g1, g2, T, flux = -0.5, -0.4, -0.2, 0.3, 0.8, 56
pars = np.array([row, col, g1, g2, T, flux])
gm = make_gmix_model(pars, model)
assert np.allclose(gm.get_flux(), flux)
assert np.allclose(gm.get_g1g2T(), [g1, g2, T])
assert np.allclose(gm.get_cen(), [row, col])
new_pars = [-0.1, 0.1, 0.5, -0.6, 0.7, 99.5]
gm.fill(new_pars)
assert np.allclose(gm.get_flux(), 99.5)
assert np.allclose(gm.get_g1g2T(), [0.5, -0.6, 0.7])
assert np.allclose(gm.get_cen(), [-0.1, 0.1])
def test_gmix_loglike_fdiff(start):
"""
test that the approx fdiff calculation agrees well enough with that
calculated with the exp() value
restrict comparisons to the valid range, which is +/- 5 sigma
"""
rtol = 4.0e-5
row, col, g1, g2, T, flux = -0.5, -0.4, 0, 0, 2.8, 5
sigma = np.sqrt(T / 2)
pars = np.array([row, col, g1, g2, T, flux])
gm = make_gmix_model(pars, 'gauss')
jacob = DiagonalJacobian(row=6.5, col=7.1, scale=0.25)
area = jacob.area
rng = np.random.RandomState(seed=10)
dims = (13, 15)
obs = Observation(
image=rng.normal(size=dims),
weight=np.exp(rng.normal(size=dims)),
jacobian=jacob,
)
fdiff = np.zeros(dims[0] * dims[1] + start, dtype=np.float64)
gm.fill_fdiff(obs, fdiff, start=start)
cen = (np.array(dims) - 1) / 2
row_range = int(cen[0] - 2 * sigma), int(cen[0] + 2 * sigma) + 1
col_range = int(cen[1] - 2 * sigma), int(cen[1] + 2 * sigma) + 1
loc = start
pnorm = 5 / 2.0 / np.pi / sigma**2
for r in range(row_range[0], row_range[1]):
for c in range(col_range[0], col_range[1]):
loc = start + r * dims[1] + c
v, u = obs.jacobian(r, c)
chi2 = ((u - col)**2 + (v - row)**2) / sigma**2
model = pnorm * np.exp(-0.5 * chi2) * area
_fdiff = (model - obs.image[r, c]) * np.sqrt(obs.weight[r, c])
print(_fdiff, fdiff[loc])
assert np.allclose(_fdiff, fdiff[loc], rtol=rtol)
def test_gmix_copy(model):
row, col, g1, g2, T, flux = -0.5, -0.4, -0.2, 0.3, 0.8, 56
pars = np.array([row, col, g1, g2, T, flux])
gm = make_gmix_model(pars, model)
# we make one, copy it, adjust the copy and then test again
assert np.allclose(gm.get_flux(), flux)
assert np.allclose(gm.get_g1g2T(), [g1, g2, T])
assert np.allclose(gm.get_cen(), [row, col])
new_gm = gm.copy()
new_gm.reset()
assert np.allclose(new_gm.get_full_pars(), 0)
assert np.allclose(gm.get_flux(), flux)
assert np.allclose(gm.get_g1g2T(), [g1, g2, T])
assert np.allclose(gm.get_cen(), [row, col])
def test_gmix_make_round(model, preserve_size):
row, col, g1, g2, T, flux = -0.5, -0.4, -0.2, 0.3, 0.8, 56
pars = np.array([row, col, g1, g2, T, flux])
gm = make_gmix_model(pars, model)
irr, irc, icc = ngmix.moments.e2mom(*gm.get_e1e2T())
max_eig = np.linalg.eigvals(np.array([[irr, irc], [irc, icc]])).max()
# we make one, copy it, adjust the copy and then test again
assert np.allclose(gm.get_flux(), flux)
assert np.allclose(gm.get_g1g2T(), [g1, g2, T])
assert np.allclose(gm.get_cen(), [row, col])
rgm = gm.make_round(preserve_size=preserve_size)
assert np.allclose(gm.get_flux(), flux)
assert np.allclose(gm.get_g1g2T(), [g1, g2, T])
assert np.allclose(gm.get_cen(), [row, col])
assert np.allclose(rgm.get_cen(), [row, col])
assert np.allclose(rgm.get_flux(), flux)
assert np.allclose(rgm.get_g1g2T()[0:2], [0, 0])
if preserve_size:
half_T = rgm.get_T() / 2 + 1e-10 # small constant for float stuff
assert half_T >= max_eig