def _compare_dist(self, dist):
import biggles
import esutil as eu
from biggles import plot_hist
r=dist.rvs(self.deep_pars.shape[0])
names=['c1','c2','M1','M2','T','I']
grid=eu.plotting.Grid(6)
tab=biggles.Table(grid.nrow, grid.ncol)
for i in xrange(6):
row,col=grid(i)
plt=plot_hist(r[:,i], nbin=100,
xlabel=names[i],
visible=False)
plt=plot_hist(self.deep_pars[:,i], nbin=100, color='red', plt=plt,
visible=False)
tab[row,col] = plt
pngfile=files.get_psample_summed_url(self['run'],
self['is2n'],
itrial=self['itrial'],
ext='png')
pngfile=pngfile.replace('.png','-hist.png')
print(pngfile)
tab.write_img(1200,800,pngfile)
key=raw_input('hit a key: ')
if key=='q':
stop
def make_r50_plot(r50):
print("r50:",r50.mean(), r50.std())
nbin=50
plt=plot_hist(r50,
nbin=nbin,
visible=False,
norm=1,
min=0,
max=6,
aspect_ratio=1,
xlabel=r'$r_{50} [pixels]$')
# the r50 from the parametric sims
p=ngmix.priors.LogNormal(1.54, 1.0)
r=p.sample(r50.size*10)
plot_hist(r,
nbin=nbin,
visible=False,
norm=1,
min=0,
max=6,
plt=plt,
color='red',
smooth=True,
type='dashed')
plt.y1.draw_ticklabels=False
return plt
def plot_resid_hist(self):
means=self.means
if means.size == 1:
return
else:
import biggles
fits=self.fits
args=self.args
#Q=calc_q(fits)
diff = means['shear'] - means['shear_true']
plt = biggles.plot_hist(diff, nbin=20, visible=False,
xlabel=r'$\gamma - \gamma_{True}$')
dmax=numpy.abs(diff).max()
plt.xrange=[-1.3*dmax, 1.3*dmax]
fname=self._get_resid_hist_file()
eu.ostools.makedirs_fromfile(fname)
print("writing:",fname)
plt.write_eps(fname)
if args.show:
plt.show(width=1000, height=1000)
def plot_resid_hist(self):
means = self.means
if means.size == 1:
return
else:
import biggles
fits = self.fits
args = self.args
#Q=calc_q(fits)
diff = means['shear'] - means['shear_true']
plt = biggles.plot_hist(diff,
nbin=20,
visible=False,
xlabel=r'$\gamma - \gamma_{True}$')
dmax = np.abs(diff).max()
plt.xrange = [-1.3 * dmax, 1.3 * dmax]
fname = self._get_resid_hist_file()
eu.ostools.makedirs_fromfile(fname)
print("writing:", fname)
plt.write_eps(fname)
if args.show:
plt.show(width=1000, height=1000)
def make_s2n_plot(s2n):
log_s2n=numpy.log10( s2n.clip(min=0.001,max=10000.0) )
plt=plot_hist(log_s2n,
nbin=100,
visible=False,
min=0.5,
max=2.5,
xlabel=r'$log_{10}(S/N)$',
aspect_ratio=1,
norm=1)
plt.y1.draw_ticklabels=False
return plt
def test_1gauss_T_recovery(noise,
T=8.0,
counts=1.0,
ntrial=100,
show=True,
png=None):
import biggles
T_true=T
T_meas=numpy.zeros(ntrial)
for i in xrange(ntrial):
while True:
try:
gm=test_1gauss(
noise=noise,
T=T_true,
counts=counts,
verbose=False,
)
T=gm.get_T()
T_meas[i]=T
break
except GMixRangeError:
pass
except GMixMaxIterEM:
pass
mean=T_meas.mean()
std=T_meas.std()
print("<T>:",mean,"sigma(T):",std)
binsize=0.2*std
plt=biggles.plot_hist(T_meas, binsize=binsize, visible=False)
p=biggles.Point(T_true, 0.0, type='filled circle', size=2, color='red')
plt.add(p)
plt.title='Flux: %g T: %g noise: %g' % (counts, T_true, noise)
xmin=mean-4.0*std
xmax=mean+4.0*std
plt.xrange=[xmin, xmax]
if show:
plt.show()
if png is not None:
print(png)
plt.write_img(800, 800, png)
def test_1gauss_T_recovery(noise,
T=8.0,
counts=1.0,
ntrial=100,
show=True,
png=None):
import biggles
T_true=T
T_meas=numpy.zeros(ntrial)
for i in xrange(ntrial):
while True:
try:
gm=test_1gauss(
noise=noise,
T=T_true,
counts=counts,
verbose=False,
)
T=gm.get_T()
T_meas[i]=T
break
except GMixRangeError:
pass
except GMixMaxIterEM:
pass
mean=T_meas.mean()
std=T_meas.std()
print("<T>:",mean,"sigma(T):",std)
binsize=0.2*std
plt=biggles.plot_hist(T_meas, binsize=binsize, visible=False)
p=biggles.Point(T_true, 0.0, type='filled circle', size=2, color='red')
plt.add(p)
plt.title='Flux: %g T: %g noise: %g' % (counts, T_true, noise)
xmin=mean-4.0*std
xmax=mean+4.0*std
plt.xrange=[xmin, xmax]
if show:
plt.show()
if png is not None:
print(png)
plt.write_img(800, 800, png)
def plothist_weights(weights, show=False):
import biggles
normweights = weights/weights.max()
w,=numpy.where(normweights > 0)
print("plothist_weights: greater than zero: %d/%d" % (w.size,normweights.size))
lw = numpy.log10(normweights[w])
#binsize = 0.1*lw.std()
plt=biggles.plot_hist(lw,
nbin=50,
#binsize=binsize,
xlabel=r'$log_{10}(relative weight)$',
visible=False)
if show:
plt.show()
return plt
def compare_images(self, type, nrand=None):
"""
compare images of the specified type
parameters
----------
type: string
Type of images to compare, e.g. 'image','weight' etc.
nrand: integer, optional
Compare at most this many images. A random subset will be
drawn
"""
if nrand is not None:
nrand = min(nrand, self.ind1.size)
ids = self._get_random_subset(nrand)
else:
ids = numpy.arange(self.ind1.size)
means = []
ntot = ids.size
for icount, index in enumerate(ids):
print(" %d/%d %d" % (icount + 1, ntot, index))
i1 = self.ind1[index]
i2 = self.ind2[index]
ncut = self._check_ncutout(i1, i2)
for icut in range(ncut):
im2 = self.m2.get_cutout(i2, icut, type=type)
im1 = self.m1.get_cutout(i1, icut, type=type)
#if type=='image':
# wt=self.m2.get_cutout(i2, icut, type='weight')
# err=numpy.sqrt(1.0/wt.max())
# print(" compare to noise:",err)
if type == 'image':
ok, mean, err, std = self._compare_images(im1, im2)
means.append(mean)
elif type == 'weight':
self._compare_weights(im1, im2)
else:
self._compare_images_exact(im1, im2)
if type == 'image' and self.png_prefix is not None:
import images
fname = self.png_prefix + '-imdiff-%06d-%03d.png' % (index,
icut)
print("writing:", fname)
if ok:
oks = 'OK'
else:
oks = 'Not OK'
images.compare_images(
im1,
im2,
dims=(1500, 1500),
#width=1500,
#height=1500,
title='rms: %.3g mean: %.3g +/- %.3g %s' %
(std, mean, err, oks),
file=fname,
)
#if 'q'==raw_input("hit a key: (q to quit)"):
# stop
if type == 'image':
import esutil as eu
print(len(means))
means = numpy.array(means)
mdiff = means.mean()
mdiff_err = means.std() / numpy.sqrt(means.size)
print("average mean diff: %g +/- %g" % (mdiff, mdiff_err))
mdiff, mdiff_std, mdiff_err = eu.stat.sigma_clip(means,
get_err=True)
print("clipped average mean diff: %g +/- %g" % (mdiff, mdiff_err))
if self.png_prefix is not None:
import biggles
binsize = mdiff_std * 0.1
fname = self.png_prefix + '-imdiff-means.png'
biggles.plot_hist(
means,
binsize=binsize,
visible=False,
file=fname,
)
def fit_prior(run, is2n=0, field='pars_noshear',show=False):
import biggles
import esutil as eu
from sklearn.mixture import GMM
import fitsio
gm=ngmix.gmix.GMixND()
alldata=files.read_output(run, is2n)
data=alldata[field]
# fit |g|
g=sqrt(data[:,2]**2 + data[:,3]**2)
gp=ngmix.priors.GPriorBA()
bs=eu.stat.Binner(g)
bs.dohist(nbin=100)
bs.calc_stats()
xvals=bs['center']
yvals=bs['hist'].astype('f8')
gp.dofit(xvals, yvals)
rg=gp.sample1d(1000000)
#eta1,eta2,good=ngmix.shape.g1g2_to_eta1eta2_array(data[:,2],
# data[:,3])
#eta = 2*numpy.arctanh(g)
#logg=log(g)
outfile=files.get_fitprior_url(run, is2n)
epsfile=files.get_fitprior_url(run, is2n, ext='eps')
# fit TF with n-dimensional gaussian
ngauss=20
gm.fit(data[:, 4:], ngauss, min_covar=1.0e-4)
'''
print("fitting log(g)")
g_gmm=GMM(n_components=10,
n_iter=5000,
min_covar=1.0e-4,
covariance_type='full')
g_gmm.fit(eta[:,newaxis])
if not g_gmm.converged_:
print("DID NOT CONVERGE")
rg=g_gmm.sample(1000000)
gplt=biggles.plot_hist(eta,nbin=100,xlabel='|g|',
norm=1,
visible=False)
'''
gplt=biggles.plot_hist(g,nbin=100,xlabel='|g|',
norm=1,
visible=False)
#gplt=biggles.plot_hist(rg[:,0],nbin=100,
gplt=biggles.plot_hist(rg,nbin=100,
plt=gplt,
norm=1,
color='red',
visible=False)
print("saving mixture to:",outfile)
gm.save_mixture(outfile)
with fitsio.FITS(outfile,'rw') as fits:
gout=numpy.zeros(1, dtype=[('sigma','f8')])
gout['sigma'] = gp.fit_pars[1]
fits.write(gout, extname='gfit')
#
# plots
#
r=gm.sample(1000000)
tab=biggles.Table(2,2)
Tplt=biggles.plot_hist(data[:,4], nbin=100,
xlabel='log(T)',
norm=1,
visible=False)
biggles.plot_hist(r[:,0], nbin=100,
color='red',
norm=1,
plt=Tplt,
visible=False)
Fplt=biggles.plot_hist(data[:,5], nbin=100,
xlabel='log(F)',
norm=1,
visible=False)
biggles.plot_hist(r[:,1], nbin=100,
color='red',
norm=1,
plt=Fplt,
visible=False)
'''
eta1plt=biggles.plot_hist(eta1, nbin=100,
xlabel='eta1',
norm=1,
visible=False)
eta2plt=biggles.plot_hist(eta2, nbin=100,
xlabel='eta2',
norm=1,
visible=False)
'''
tab[0,0] = Tplt
tab[0,1] = Fplt
#tab[1,0] = eta1plt
#tab[1,1] = eta2plt
tab[1,0] = gplt
tab[1,1] = gplt
tab.aspect_ratio=0.5
print("writing:",epsfile)
tab.write_eps(epsfile)
if show:
biggles.configure('screen','width',1200)
biggles.configure('screen','height',700)
tab.show()
def compare_images(self, type, nrand=None):
"""
compare images of the specified type
parameters
----------
type: string
Type of images to compare, e.g. 'image','weight' etc.
nrand: integer, optional
Compare at most this many images. A random subset will be
drawn
"""
if nrand is not None:
nrand = min(nrand, self.ind1.size)
ids = self._get_random_subset(nrand)
else:
ids = numpy.arange(self.ind1.size)
means=[]
ntot = ids.size
for icount, index in enumerate(ids):
print(" %d/%d %d" % (icount+1,ntot,index))
i1 = self.ind1[index]
i2 = self.ind2[index]
ncut=self._check_ncutout(i1, i2)
for icut in range(ncut):
im2 = self.m2.get_cutout(i2, icut, type=type)
im1 = self.m1.get_cutout(i1, icut, type=type)
#if type=='image':
# wt=self.m2.get_cutout(i2, icut, type='weight')
# err=numpy.sqrt(1.0/wt.max())
# print(" compare to noise:",err)
if type=='image':
ok, mean, err, std = self._compare_images(im1, im2)
means.append(mean)
elif type == 'weight':
self._compare_weights(im1, im2)
else:
self._compare_images_exact(im1, im2)
if type=='image' and self.png_prefix is not None:
import images
fname=self.png_prefix + '-imdiff-%06d-%03d.png' % (index, icut)
print("writing:",fname)
if ok:
oks='OK'
else:
oks='Not OK'
images.compare_images(
im1,
im2,
dims=(1500,1500),
#width=1500,
#height=1500,
title='rms: %.3g mean: %.3g +/- %.3g %s' % (std, mean, err, oks),
file=fname,
)
#if 'q'==raw_input("hit a key: (q to quit)"):
# stop
if type=='image':
import esutil as eu
print(len(means))
means = numpy.array(means)
mdiff = means.mean()
mdiff_err = means.std()/numpy.sqrt(means.size)
print("average mean diff: %g +/- %g" % (mdiff, mdiff_err))
mdiff, mdiff_std, mdiff_err = eu.stat.sigma_clip(means, get_err=True)
print("clipped average mean diff: %g +/- %g" % (mdiff, mdiff_err))
if self.png_prefix is not None:
import biggles
binsize=mdiff_std*0.1
fname=self.png_prefix + '-imdiff-means.png'
biggles.plot_hist(
means, binsize=binsize, visible=False, file=fname,
)
