def test_roi():
""" Testing ugali.observation.roi """
import ugali.observation.roi
roi = ugali.observation.roi.ROI(CONFIG, LON, LAT)
# testing roi location
np.testing.assert_equal(roi.lon, LON)
np.testing.assert_equal(roi.lat, LAT)
# testing the roi regions
np.testing.assert_equal(len(roi.pixels),61327)
np.testing.assert_equal(len(roi.pixels_interior),3828)
np.testing.assert_equal(len(roi.pixels_annulus),57499)
np.testing.assert_equal(len(roi.pixels_target),256)
# test points
np.testing.assert_equal(roi.inROI(POINTS[0],POINTS[1]),
np.array([True,True,True,True]))
np.testing.assert_equal(roi.inAnnulus(POINTS[0],POINTS[1]),
np.array([False,False,False,True]))
np.testing.assert_equal(roi.inInterior(POINTS[0],POINTS[1]),
np.array([True,True,True,False]))
np.testing.assert_equal(roi.inTarget(POINTS[0],POINTS[1]),
np.array([True,True,False,False]))
np.testing.assert_equal(roi.indexPixels(POINTS[0],POINTS[1],roi.pixels),
np.array([31034,30327,27527,54866]))
def run(self, outfile=None, size=None):
if size is None: size = self.config['simulate']['size']
data = self.generate(size)
dtype = [('kernel', 'S18'), ('ts', '>f4'), ('fit_kernel', 'S18'),
('fit_ts', '>f4'), ('fit_mass', '>f4'),
('fit_mass_err', '>f4'), ('fit_distance', '>f4'),
('fit_distance_err', '>f4')]
results = np.array(np.nan * np.ones(size), dtype=dtype)
results = recfuncs.merge_arrays([data, results],
flatten=True,
asrecarray=False,
usemask=False)
self.results = results
if outfile: self.write(outfile, results)
for i, d in enumerate(data):
params = dict(zip(data.dtype.names, d))
lon, lat = params['lon'], params['lat']
distance_modulus = params['distance_modulus']
logger.info('\n(%i/%i); (lon, lat) = (%.2f, %.2f)' %
(i + 1, len(data), lon, lat))
roi = ugali.analysis.loglike.createROI(self.config, lon, lat)
mask = ugali.analysis.loglike.createMask(self.config, roi)
isochrone = ugali.analysis.loglike.createIsochrone(self.config)
kernel = ugali.analysis.loglike.createKernel(self.config, lon, lat)
pix = roi.indexTarget(lon, lat)
simulator = Simulator(self.config, roi)
#catalog = simulator.simulate(seed=self.seed, **params)
catalog = simulator.simulate(**params)
#print "Catalog annulus contains:",roi.inAnnulus(simulator.catalog.lon,simulator.catalog.lat).sum()
logger.info("Simulated catalog annulus contains %i stars" %
roi.inAnnulus(catalog.lon, catalog.lat).sum())
if len(catalog.lon) < 1000:
logger.error(
"Simulation contains too few objects; skipping...")
continue
"""
like = ugali.analysis.loglike.LogLikelihood(self.config, roi, mask, catalog, isochrone, kernel)
like.set_params(distance_modulus=params['distance_modulus'])
like.sync_params()
results[i]['ts'] = 2*like.fit_richness()[0]
print 'TS=',results[i]['ts']
like2 = ugali.analysis.loglike.LogLikelihood(self.config, roi, mask, simulator.catalog, isochrone, kernel)
like2.set_params(distance_modulus=params['distance_modulus'])
like2.sync_params()
print 'TS=',2*like2.fit_richness()[0]
"""
#return simulator,like,like2
# Index of closest distance modulus
grid = ugali.analysis.scan.GridSearch(self.config, roi, mask,
catalog, isochrone, kernel)
self.catalog = catalog
self.simulator = simulator
self.grid = grid
self.loglike = self.grid.loglike
# ADW: Should allow fit_distance to float in order to model search procedure
#fit_distance = float(distance_modulus)
distance_idx = np.fabs(grid.distance_modulus_array -
params['distance_modulus']).argmin()
fit_distance = grid.distance_modulus_array[distance_idx]
grid.search(coords=(lon, lat), distance_modulus=fit_distance)
logger.info(str(self.loglike))
mle = grid.mle()
results[i]['kernel'] = simulator.kernel.name
results[i]['fit_kernel'] = grid.loglike.kernel.name
results[i][
'ts'] = 2 * grid.log_likelihood_sparse_array[distance_idx][pix]
results[i]['fit_ts'] = 2 * np.max(
grid.log_likelihood_sparse_array[:, pix])
results[i][
'fit_mass'] = grid.stellar_mass_conversion * mle['richness']
results[i]['fit_distance'] = fit_distance #mle['distance_modulus']
err = grid.err()
richness_err = (err['richness'][1] - err['richness'][0]) / 2.
results[i][
'fit_mass_err'] = grid.stellar_mass_conversion * richness_err
distance_modulus_err = (err['distance_modulus'][1] -
err['distance_modulus'][0]) / 2.
results[i]['fit_distance_err'] = distance_modulus_err
for d in dtype:
logger.info('\t%s: %s' % (d[0], results[i][d[0]]))
if i % self.config['simulate']['save'] == 0 and outfile:
self.write(outfile, results)
if outfile: self.write(outfile, results)
return results
def run(self, outfile=None, size=None):
if size is None: size = self.config['simulate']['size']
data = self.generate(size)
dtype=[('kernel','S18'),('ts','>f4'),('fit_kernel','S18'),('fit_ts','>f4'),
('fit_mass','>f4'),('fit_mass_err','>f4'),
('fit_distance','>f4'),('fit_distance_err','>f4')]
results = np.array(np.nan*np.ones(size),dtype=dtype)
results = recfuncs.merge_arrays([data,results],flatten=True,asrecarray=False,usemask=False)
self.results = results
if outfile: self.write(outfile,results)
for i,d in enumerate(data):
params = dict(zip(data.dtype.names,d))
lon,lat = params['lon'],params['lat']
distance_modulus = params['distance_modulus']
logger.info('\n(%i/%i); (lon, lat) = (%.2f, %.2f)'%(i+1,len(data),lon,lat))
roi = ugali.analysis.loglike.createROI(self.config,lon,lat)
mask = ugali.analysis.loglike.createMask(self.config,roi)
isochrone = ugali.analysis.loglike.createIsochrone(self.config)
kernel = ugali.analysis.loglike.createKernel(self.config,lon,lat)
pix = roi.indexTarget(lon,lat)
simulator = Simulator(self.config,roi)
#catalog = simulator.simulate(seed=self.seed, **params)
catalog = simulator.simulate(**params)
#print "Catalog annulus contains:",roi.inAnnulus(simulator.catalog.lon,simulator.catalog.lat).sum()
logger.info("Simulated catalog annulus contains %i stars"%roi.inAnnulus(catalog.lon,catalog.lat).sum())
if len(catalog.lon) < 1000:
logger.error("Simulation contains too few objects; skipping...")
continue
"""
like = ugali.analysis.loglike.LogLikelihood(self.config, roi, mask, catalog, isochrone, kernel)
like.set_params(distance_modulus=params['distance_modulus'])
like.sync_params()
results[i]['ts'] = 2*like.fit_richness()[0]
print 'TS=',results[i]['ts']
like2 = ugali.analysis.loglike.LogLikelihood(self.config, roi, mask, simulator.catalog, isochrone, kernel)
like2.set_params(distance_modulus=params['distance_modulus'])
like2.sync_params()
print 'TS=',2*like2.fit_richness()[0]
"""
#return simulator,like,like2
# Index of closest distance modulus
grid = ugali.analysis.scan.GridSearch(self.config,roi,mask,catalog,isochrone,kernel)
self.catalog = catalog
self.simulator = simulator
self.grid = grid
self.loglike = self.grid.loglike
# ADW: Should allow fit_distance to float in order to model search procedure
#fit_distance = float(distance_modulus)
distance_idx = np.fabs(grid.distance_modulus_array-params['distance_modulus']).argmin()
fit_distance = grid.distance_modulus_array[distance_idx]
grid.search(coords=(lon,lat),distance_modulus=fit_distance)
logger.info(str(self.loglike))
mle = grid.mle()
results[i]['kernel'] = simulator.kernel.name
results[i]['fit_kernel'] = grid.loglike.kernel.name
results[i]['ts'] = 2*grid.log_likelihood_sparse_array[distance_idx][pix]
results[i]['fit_ts'] = 2*np.max(grid.log_likelihood_sparse_array[:,pix])
results[i]['fit_mass'] = grid.stellar_mass_conversion*mle['richness']
results[i]['fit_distance'] = fit_distance #mle['distance_modulus']
err = grid.err()
richness_err = (err['richness'][1]-err['richness'][0])/2.
results[i]['fit_mass_err'] = grid.stellar_mass_conversion*richness_err
distance_modulus_err = (err['distance_modulus'][1]-err['distance_modulus'][0])/2.
results[i]['fit_distance_err'] = distance_modulus_err
for d in dtype:
logger.info('\t%s: %s'%(d[0], results[i][d[0]]))
if i%self.config['simulate']['save']==0 and outfile:
self.write(outfile,results)
if outfile: self.write(outfile,results)
return results