"""Runs commands to produce convolved predicted counts map in current directory."""
import numpy as np
import matplotlib.pyplot as plt
from astropy.io import fits
from gammapy.stats import significance
from gammapy.image.utils import disk_correlate
from aplpy import FITSFigure
from npred_general import prepare_images
model, gtmodel, ratio, counts, header = prepare_images()
# Top hat correlation
correlation_radius = 3
correlated_gtmodel = disk_correlate(gtmodel, correlation_radius)
correlated_counts = disk_correlate(counts, correlation_radius)
correlated_model = disk_correlate(model, correlation_radius)
# Fermi significance
fermi_significance = np.nan_to_num(significance(correlated_counts, gtmodel,
method='lima'))
# Gammapy significance
significance = np.nan_to_num(significance(correlated_counts, correlated_model,
method='lima'))
titles = ['Gammapy Significance', 'Fermi Tools Significance']
# Plot
fig = plt.figure(figsize=(10, 5))
hdu1 = fits.ImageHDU(significance, header)
def make_model_image(psf='Fermi', resolution=0.1, center=[0, 0], lat_range=[0, 180], lon_range=[0, 360], catalog='1FHL', total_flux='False', filename='1fhl_fermi_psf.fits'):
from gammapy.image import coordinates
from astropy.convolution import convolve
from astropy.modeling.models import Gaussian2D, Disk2D
from gammapy.image import images_to_cube
from gammapy.image import paste_cutout_into_image
from gammapy.image.utils import solid_angle
reference = make_reference_survey_image(resolution, lat_range, lon_range, center, units='ph/cm2/s/sr') #Check these units!
lons, lats = coordinates(reference)
source_table = fits.open('simulated_galaxy_1.fits')[1].data
sources = np.arange(len(source_table['flux'])).astype(int)
if psf == 'None':
# If there is no PSF defined, sources will be modelled as Gaussians
source_kernel = Gaussian2D(0, 0, 0, 0.1, 0.1)
new_image = im_1 = im_2 = im_3 = source_kernel(lats, lons)
else:
# Otherwise, all of the flux will be placed into the reference pixel to be later PSF convolved with the defined PSF
# Hence original reference empty image is called
new_image = im_1 = im_2 = im_3 = reference.data
from gammapy.image import wcs_histogram2d
header = reference.header
lon = source_table['GLON']
lat = source_table['GLAT']
flux = source_table['flux']
total_point_image = wcs_histogram2d(header, lon, lat, weights=flux)
new_image = total_point_image.data
#import IPython; IPython.embed(); 1 /0
"""
print len(sources)
total_point_image = fits.ImageHDU(header=reference.header, data=new_image)
from astropy.wcs import WCS
wcs = WCS(total_point_image.header)
new_image = np.zeros_like(total_point_image.data, dtype=np.float64)
for source in sources:
source_type = 'PointSource'#source_table['Source_Type'][source]
print source
if source_type == 'ExtendedSource':
raise NotImplementedError
# This needs more work...
#image = source_table['Image'][source]
#image.data = (image.data * solid_angle(image).value.mean()) / 1000000000000 # TODO: fix this hack... units???
#resample_factor1 = np.round(reference.header['CDELT1'] / image.header['CDELT1'])
#resample_factor2 = np.round(reference.header['CDELT2'] / image.header['CDELT2'])
#block_factors = np.array([resample_factor1, resample_factor2]) # TODO: fix this approximation... kapteyn image utils reprojectto?
#resampled_image = block_reduce_hdu(image, block_factors, np.sum)
#paste_cutout_into_image(total_point_image, resampled_image)
elif source_type == 'PointSource':
lon = source_table['GLON'][source]
lat = source_table['GLAT'][source]
flux = source_table['flux'][source]
precise = False
if precise:
raise NotImplementedError
else:
#print(lon, lat)
#import IPython; IPython.embed()
print lon, lat
x, y = wcs.wcs_world2pix(lon, lat, 0)
print x, y
# TODO: check if this is 0.5 or 1 pix off
xi, yi = x.astype(int), y.astype(int)
new_image[yi, xi] += flux
total_point_image = fits.ImageHDU(header=reference.header, data=new_image)
"""
# Ensure flux or counts remain the same
if total_flux == 'True':
factor = source_table['Flux'].sum()
else:
factor = total_flux
if psf == 'None':
new_image = (new_image/new_image.sum()) * factor
elif psf == 'Gaussian':
new_image = correlate_gauss((new_image/new_image.sum()), 2)
elif psf == 'Disk':
new_image = disk_correlate((new_image/new_image.sum()), 2)
elif psf == 'Fermi':
print "Original Flux"
print new_image.sum()
new_image = correlate_fermi_psf(new_image, 5)
print "PSF Convolved Flux"
print new_image.sum()
header = reference.header
image = fits.ImageHDU(data=new_image, header=header)
image.writeto(filename, clobber=True)
"""Runs commands to produce convolved predicted counts map in current directory."""
import numpy as np
import matplotlib.pyplot as plt
from astropy.io import fits
from gammapy.stats import significance
from gammapy.image.utils import disk_correlate
from npred_general import prepare_images
from aplpy import FITSFigure
model, gtmodel, ratio, counts, header = prepare_images()
# Top hat correlation
correlation_radius = 3
correlated_gtmodel = disk_correlate(gtmodel, correlation_radius)
correlated_counts = disk_correlate(counts, correlation_radius)
correlated_model = disk_correlate(model, correlation_radius)
# Fermi significance
fermi_significance = np.nan_to_num(significance(correlated_counts, gtmodel,
method='lima'))
# Gammapy significance
significance = np.nan_to_num(significance(correlated_counts, correlated_model,
method='lima'))
titles = ['Gammapy Significance', 'Fermi Tools Significance']
# Plot
fig = plt.figure(figsize=(10, 5))
hdu1 = fits.ImageHDU(significance, header)
def make_model_image(psf='Fermi',
resolution=0.1,
center=[0, 0],
lat_range=[0, 180],
lon_range=[0, 360],
catalog='1FHL',
total_flux='False',
filename='1fhl_fermi_psf.fits'):
from gammapy.image import coordinates
from astropy.convolution import convolve
from astropy.modeling.models import Gaussian2D, Disk2D
from gammapy.image import images_to_cube
from gammapy.image import paste_cutout_into_image
from gammapy.image.utils import solid_angle
reference = make_reference_survey_image(
resolution, lat_range, lon_range, center,
units='ph/cm2/s/sr') #Check these units!
lons, lats = coordinates(reference)
source_table = fits.open('simulated_galaxy_1.fits')[1].data
sources = np.arange(len(source_table['flux'])).astype(int)
if psf == 'None':
# If there is no PSF defined, sources will be modelled as Gaussians
source_kernel = Gaussian2D(0, 0, 0, 0.1, 0.1)
new_image = im_1 = im_2 = im_3 = source_kernel(lats, lons)
else:
# Otherwise, all of the flux will be placed into the reference pixel to be later PSF convolved with the defined PSF
# Hence original reference empty image is called
new_image = im_1 = im_2 = im_3 = reference.data
from gammapy.image import wcs_histogram2d
header = reference.header
lon = source_table['GLON']
lat = source_table['GLAT']
flux = source_table['flux']
total_point_image = wcs_histogram2d(header, lon, lat, weights=flux)
new_image = total_point_image.data
#import IPython; IPython.embed(); 1 /0
"""
print len(sources)
total_point_image = fits.ImageHDU(header=reference.header, data=new_image)
from astropy.wcs import WCS
wcs = WCS(total_point_image.header)
new_image = np.zeros_like(total_point_image.data, dtype=np.float64)
for source in sources:
source_type = 'PointSource'#source_table['Source_Type'][source]
print source
if source_type == 'ExtendedSource':
raise NotImplementedError
# This needs more work...
#image = source_table['Image'][source]
#image.data = (image.data * solid_angle(image).value.mean()) / 1000000000000 # TODO: fix this hack... units???
#resample_factor1 = np.round(reference.header['CDELT1'] / image.header['CDELT1'])
#resample_factor2 = np.round(reference.header['CDELT2'] / image.header['CDELT2'])
#block_factors = np.array([resample_factor1, resample_factor2]) # TODO: fix this approximation... kapteyn image utils reprojectto?
#resampled_image = block_reduce_hdu(image, block_factors, np.sum)
#paste_cutout_into_image(total_point_image, resampled_image)
elif source_type == 'PointSource':
lon = source_table['GLON'][source]
lat = source_table['GLAT'][source]
flux = source_table['flux'][source]
precise = False
if precise:
raise NotImplementedError
else:
#print(lon, lat)
#import IPython; IPython.embed()
print lon, lat
x, y = wcs.wcs_world2pix(lon, lat, 0)
print x, y
# TODO: check if this is 0.5 or 1 pix off
xi, yi = x.astype(int), y.astype(int)
new_image[yi, xi] += flux
total_point_image = fits.ImageHDU(header=reference.header, data=new_image)
"""
# Ensure flux or counts remain the same
if total_flux == 'True':
factor = source_table['Flux'].sum()
else:
factor = total_flux
if psf == 'None':
new_image = (new_image / new_image.sum()) * factor
elif psf == 'Gaussian':
new_image = correlate_gauss((new_image / new_image.sum()), 2)
elif psf == 'Disk':
new_image = disk_correlate((new_image / new_image.sum()), 2)
elif psf == 'Fermi':
print "Original Flux"
print new_image.sum()
new_image = correlate_fermi_psf(new_image, 5)
print "PSF Convolved Flux"
print new_image.sum()
header = reference.header
image = fits.ImageHDU(data=new_image, header=header)
image.writeto(filename, clobber=True)