def _scan_extension_fast(self, name, **kwargs):
state = SourceMapState(self.like, [name])
self.free_norm(name)
optimizer = kwargs.get('optimizer', {})
width = kwargs.get('width')
spatial_model = kwargs.get('spatial_model')
skydir = kwargs.pop('skydir', self.roi[name].skydir)
psf_scale_fn = kwargs.pop('psf_scale_fn', None)
reoptimize = kwargs.pop('reoptimize', True)
src = self.roi.copy_source(name)
spatial_pars = {'ra': skydir.ra.deg, 'dec': skydir.dec.deg}
loglike = []
for i, w in enumerate(width):
spatial_pars['SpatialWidth'] = max(w, 0.00316)
self.set_source_morphology(name,
spatial_model=spatial_model,
spatial_pars=spatial_pars,
use_pylike=False,
psf_scale_fn=psf_scale_fn)
if reoptimize:
fit_output = self._fit(loglevel=logging.DEBUG, **optimizer)
loglike += [fit_output['loglike']]
else:
loglike += [-self.like()]
state.restore()
return np.array(loglike)
def _scan_extension_fast(self, name, **kwargs):
state = SourceMapState(self.like, [name])
self.free_norm(name)
optimizer = kwargs.get('optimizer', {})
width = kwargs.get('width')
spatial_model = kwargs.get('spatial_model')
skydir = kwargs.pop('skydir', self.roi[name].skydir)
psf_scale_fn = kwargs.pop('psf_scale_fn', None)
reoptimize = kwargs.pop('reoptimize', True)
src = self.roi.copy_source(name)
spatial_pars = {'ra': skydir.ra.deg, 'dec': skydir.dec.deg}
loglike = []
for i, w in enumerate(width):
spatial_pars['SpatialWidth'] = max(w, 0.00316)
self.set_source_morphology(name,
spatial_model=spatial_model,
spatial_pars=spatial_pars,
use_pylike=False,
psf_scale_fn=psf_scale_fn)
if reoptimize:
fit_output = self._fit(loglevel=logging.DEBUG, **optimizer)
loglike += [fit_output['loglike']]
else:
loglike += [-self.like()]
state.restore()
return np.array(loglike)
def _scan_extension_fast_ebin(self, name, **kwargs):
state = SourceMapState(self.like, [name])
self.free_norm(name)
optimizer = kwargs.get('optimizer', {})
width = kwargs.get('width')
spatial_model = kwargs.get('spatial_model')
skydir = kwargs.pop('skydir', self.roi[name].skydir)
psf_scale_fn = kwargs.pop('psf_scale_fn', None)
reoptimize = kwargs.pop('reoptimize', True)
src = self.roi.copy_source(name)
spatial_pars = {'ra': skydir.ra.deg, 'dec': skydir.dec.deg}
loglike = np.ones((self.enumbins, len(width)))
for i, w in enumerate(width):
spatial_pars['SpatialWidth'] = max(w, 0.00316)
self.set_source_morphology(name,
spatial_model=spatial_model,
spatial_pars=spatial_pars,
use_pylike=False,
psf_scale_fn=psf_scale_fn)
for j, (logemin, logemax) in enumerate(
zip(self.log_energies[:-1], self.log_energies[1:])):
self.set_energy_range(logemin, logemax)
if reoptimize:
fit_output = self._fit(loglevel=logging.DEBUG, **optimizer)
loglike[j, i] = fit_output['loglike']
else:
loglike[j, i] = -self.like()
self.set_energy_range(self.log_energies[0], self.log_energies[-1])
state.restore()
return loglike
def _scan_extension_fast_ebin(self, name, **kwargs):
state = SourceMapState(self.like, [name])
self.free_norm(name)
optimizer = kwargs.get('optimizer', {})
width = kwargs.get('width')
spatial_model = kwargs.get('spatial_model')
skydir = kwargs.pop('skydir', self.roi[name].skydir)
psf_scale_fn = kwargs.pop('psf_scale_fn', None)
reoptimize = kwargs.pop('reoptimize', True)
src = self.roi.copy_source(name)
spatial_pars = {'ra': skydir.ra.deg, 'dec': skydir.dec.deg}
loglike = np.ones((self.enumbins, len(width)))
for i, w in enumerate(width):
spatial_pars['SpatialWidth'] = max(w, 0.00316)
self.set_source_morphology(name,
spatial_model=spatial_model,
spatial_pars=spatial_pars,
use_pylike=False,
psf_scale_fn=psf_scale_fn)
for j, (logemin, logemax) in enumerate(zip(self.log_energies[:-1],
self.log_energies[1:])):
self.set_energy_range(logemin, logemax)
if reoptimize:
fit_output = self._fit(loglevel=logging.DEBUG, **optimizer)
loglike[j, i] = fit_output['loglike']
else:
loglike[j, i] = -self.like()
self.set_energy_range(self.log_energies[0], self.log_energies[-1])
state.restore()
return loglike
def _fit_position_opt(self, name, use_cache=True):
state = SourceMapState(self.like, [name])
src = self.roi.copy_source(name)
if use_cache:
self._create_srcmap_cache(src.name, src)
loglike = []
skydir = src.skydir
skywcs = self.geom.wcs
src_pix = skydir.to_pixel(skywcs)
c = skydir.transform_to('icrs')
src.set_radec(c.ra.deg, c.dec.deg)
self._update_srcmap(src.name, src)
print(src_pix, self.like())
import time
def fit_fn(params):
t0 = time.time()
c = SkyCoord.from_pixel(params[0], params[1], self.geom.wcs)
c = c.transform_to('icrs')
src.set_radec(c.ra.deg, c.dec.deg)
t1 = time.time()
self._update_srcmap(src.name, src)
t2 = time.time()
val = self.like()
t3 = time.time()
print(params, val)
# print(t1-t0,t2-t1,t3-t2)
return val
#lnl0 = fit_fn(src_pix[0],src_pix[1])
#lnl1 = fit_fn(src_pix[0]+0.1,src_pix[1])
# print(lnl0,lnl1)
import scipy
#src_pix[1] += 3.0
p0 = [src_pix[0], src_pix[1]]
#p0 = np.array([14.665692574327048, 16.004594098101926])
#delta = np.array([0.3,-0.4])
#p0 = [14.665692574327048, 16.004594098101926]
o = scipy.optimize.minimize(fit_fn, p0,
bounds=[(0.0, 39.0),
(0.0, 39.0)],
# method='L-BFGS-B',
method='SLSQP',
tol=1e-6)
print('fit 2')
o = scipy.optimize.minimize(fit_fn, o.x,
bounds=[(0.0, 39.0),
(0.0, 39.0)],
# method='L-BFGS-B',
method='SLSQP',
tol=1e-6)
print(o)
print(fit_fn(p0))
print(fit_fn(o.x))
print(fit_fn(o.x + np.array([0.02, 0.02])))
print(fit_fn(o.x + np.array([0.02, -0.02])))
print(fit_fn(o.x + np.array([-0.02, 0.02])))
print(fit_fn(o.x + np.array([-0.02, -0.02])))
state.restore()
return o
def _fit_position_opt(self, name, use_cache=True):
state = SourceMapState(self.like, [name])
src = self.roi.copy_source(name)
if use_cache:
self._create_srcmap_cache(src.name, src)
loglike = []
skydir = src.skydir
skywcs = self._skywcs
src_pix = skydir.to_pixel(skywcs)
c = skydir.transform_to('icrs')
src.set_radec(c.ra.deg, c.dec.deg)
self._update_srcmap(src.name, src)
print(src_pix, self.like())
import time
def fit_fn(params):
t0 = time.time()
c = SkyCoord.from_pixel(params[0], params[1], self._skywcs)
c = c.transform_to('icrs')
src.set_radec(c.ra.deg, c.dec.deg)
t1 = time.time()
self._update_srcmap(src.name, src)
t2 = time.time()
val = self.like()
t3 = time.time()
print(params, val)
# print(t1-t0,t2-t1,t3-t2)
return val
#lnl0 = fit_fn(src_pix[0],src_pix[1])
#lnl1 = fit_fn(src_pix[0]+0.1,src_pix[1])
# print(lnl0,lnl1)
import scipy
#src_pix[1] += 3.0
p0 = [src_pix[0], src_pix[1]]
#p0 = np.array([14.665692574327048, 16.004594098101926])
#delta = np.array([0.3,-0.4])
#p0 = [14.665692574327048, 16.004594098101926]
o = scipy.optimize.minimize(fit_fn, p0,
bounds=[(0.0, 39.0),
(0.0, 39.0)],
# method='L-BFGS-B',
method='SLSQP',
tol=1e-6)
print ('fit 2')
o = scipy.optimize.minimize(fit_fn, o.x,
bounds=[(0.0, 39.0),
(0.0, 39.0)],
# method='L-BFGS-B',
method='SLSQP',
tol=1e-6)
print(o)
print(fit_fn(p0))
print(fit_fn(o.x))
print(fit_fn(o.x + np.array([0.02, 0.02])))
print(fit_fn(o.x + np.array([0.02, -0.02])))
print(fit_fn(o.x + np.array([-0.02, 0.02])))
print(fit_fn(o.x + np.array([-0.02, -0.02])))
state.restore()
return o