def generate_cameras(normal_vector, distance=100.0, fov=50.0):
'''
Set camera positions and orientations
'''
from yt.utilities.orientation import Orientation
print "\nGenerating cameras"
north = np.array([0., 1., 0.])
orient = Orientation(normal_vector=normal_vector, north_vector=north)
R = np.linalg.inv(orient.inv_mat)
camera_set = OrderedDict([
['face', ([0., 0., 1.], [0., -1., 0], True)], #up is north=+y
['edge', ([0., 1., 0.], [0., 0., -1.], True)], #up is along z
['45', ([0., 0.7071, 0.7071], [0., 0., -1.], True)],
['Z-axis', ([0., 0., 1.], [0., -1., 0], False)], #up is north=+y
['Y-axis', ([0., 1., 0.], [0., 0., -1.], False)], #up is along z
])
segments = 10
np.random.seed(0)
ts = np.random.random(segments) * np.pi * 2
ps = np.random.random(segments) * np.pi - np.pi / 2.0
for i, (theta, phi) in enumerate(zip(ts, ps)):
pos = [np.cos(theta), 0., np.sin(phi)]
vc = [np.cos(np.pi / 2. - theta), 0., np.sin(np.pi / 2. - phi)]
camera_set['Random_%03i' % (i)] = (pos, vc, False)
i = 0
cameras = OrderedDict()
for name, (normal, north, do_rot) in camera_set.iteritems():
orient = Orientation(normal_vector=normal, north_vector=north)
if do_rot:
drot = R.copy()
else:
drot = np.identity(3)
sunrise_pos = np.dot(orient.normal_vector, drot)
sunrise_up = L.copy()
if np.all(np.abs(sunrise_up - sunrise_pos) < 1e-3):
sunrise_up[0] *= 0.5
sunrise_direction = -1.0 * sunrise_pos
sunrise_afov = 2.0 * np.arctan((fov / 2.0) / distance)
norm = lambda x: x / np.sqrt(np.sum(x * x))
if np.all(np.abs(norm(sunrise_up) - norm(sunrise_pos)) < 1e-3):
sunrise_up[0] *= 0.5
sunrise_up = norm(sunrise_up)
line = (distance * sunrise_pos, distance * sunrise_direction,
sunrise_up, sunrise_afov, fov, distance)
cameras[name] = line
i += 1
print "Successfully generated cameras\n"
return cameras
def __init__(self,
normal,
center,
north_vector=None,
ds=None,
field_parameters=None,
data_source=None):
validate_3d_array(normal)
validate_center(center)
if north_vector is not None:
validate_3d_array(north_vector)
validate_object(ds, Dataset)
validate_object(field_parameters, dict)
validate_object(data_source, YTSelectionContainer)
YTSelectionContainer2D.__init__(self, 4, ds, field_parameters,
data_source)
self._set_center(center)
self.set_field_parameter('center', center)
# Let's set up our plane equation
# ax + by + cz + d = 0
self.orienter = Orientation(normal, north_vector=north_vector)
self._norm_vec = self.orienter.normal_vector
self._d = -1.0 * np.dot(self._norm_vec, self.center)
self._x_vec = self.orienter.unit_vectors[0]
self._y_vec = self.orienter.unit_vectors[1]
# First we try all three, see which has the best result:
self._rot_mat = np.array([self._x_vec, self._y_vec, self._norm_vec])
self._inv_mat = np.linalg.pinv(self._rot_mat)
self.set_field_parameter('cp_x_vec', self._x_vec)
self.set_field_parameter('cp_y_vec', self._y_vec)
self.set_field_parameter('cp_z_vec', self._norm_vec)
def create_vlos(normal, no_shifting):
if no_shifting:
def _v_los(field, data):
return data.ds.arr(data["index", "zeros"], "cm/s")
elif isinstance(normal, string_types):
def _v_los(field, data):
return -data["gas", "velocity_%s" % normal]
else:
orient = Orientation(normal)
los_vec = orient.unit_vectors[2]
def _v_los(field, data):
vz = data["gas", "velocity_x"]*los_vec[0] + \
data["gas", "velocity_y"]*los_vec[1] + \
data["gas", "velocity_z"]*los_vec[2]
return -vz
return _v_los
def __init__(self,
normal,
center,
north_vector=None,
ds=None,
field_parameters=None):
YTSelectionContainer2D.__init__(self, 4, ds, field_parameters)
self._set_center(center)
self.set_field_parameter('center', center)
# Let's set up our plane equation
# ax + by + cz + d = 0
self.orienter = Orientation(normal, north_vector=north_vector)
self._norm_vec = self.orienter.normal_vector
self._d = -1.0 * np.dot(self._norm_vec, self.center)
self._x_vec = self.orienter.unit_vectors[0]
self._y_vec = self.orienter.unit_vectors[1]
# First we try all three, see which has the best result:
vecs = np.identity(3)
self._rot_mat = np.array([self._x_vec, self._y_vec, self._norm_vec])
self._inv_mat = np.linalg.pinv(self._rot_mat)
self.set_field_parameter('cp_x_vec', self._x_vec)
self.set_field_parameter('cp_y_vec', self._y_vec)
self.set_field_parameter('cp_z_vec', self._norm_vec)
def __init__(self,
ds,
normal,
field,
width=(1.0, "unitary"),
dims=(100, 100, 100),
velocity_bounds=None):
r""" Initialize a PPVCube object.
Parameters
----------
ds : dataset
The dataset.
normal : array_like
The normal vector along with to make the projections.
field : string
The field to project.
width : float or tuple, optional
The width of the projection in length units. Specify a float
for code_length units or a tuple (value, units).
dims : tuple, optional
A 3-tuple of dimensions (nx,ny,nv) for the cube.
velocity_bounds : tuple, optional
A 3-tuple of (vmin, vmax, units) for the velocity bounds to
integrate over. If None, the largest velocity of the
dataset will be used, e.g. velocity_bounds = (-v.max(), v.max())
Examples
--------
>>> i = 60*np.pi/180.
>>> L = [0.0,np.sin(i),np.cos(i)]
>>> cube = PPVCube(ds, L, "density", width=(10.,"kpc"),
... velocity_bounds=(-5.,4.,"km/s"))
"""
self.ds = ds
self.field = field
self.width = width
self.nx = dims[0]
self.ny = dims[1]
self.nv = dims[2]
normal = np.array(normal)
normal /= np.sqrt(np.dot(normal, normal))
vecs = np.identity(3)
t = np.cross(normal, vecs).sum(axis=1)
ax = t.argmax()
north = np.cross(normal, vecs[ax, :]).ravel()
orient = Orientation(normal, north_vector=north)
dd = ds.all_data()
fd = dd._determine_fields(field)[0]
self.field_units = ds._get_field_info(fd).units
if velocity_bounds is None:
vmin, vmax = dd.quantities.extrema("velocity_magnitude")
self.v_bnd = -vmax, vmax
else:
self.v_bnd = (ds.quan(velocity_bounds[0], velocity_bounds[2]),
ds.quan(velocity_bounds[1], velocity_bounds[2]))
self.vbins = np.linspace(self.v_bnd[0], self.v_bnd[1], num=self.nv + 1)
self.vmid = 0.5 * (self.vbins[1:] + self.vbins[:-1])
self.dv = (self.v_bnd[1] - self.v_bnd[0]) / self.nv
_vlos = create_vlos(orient.unit_vectors[2])
ds.field_info.add_field(("gas", "v_los"), function=_vlos, units="cm/s")
self.data = ds.arr(np.zeros((self.nx, self.ny, self.nv)),
self.field_units)
pbar = get_pbar("Generating cube.", self.nv)
for i in xrange(self.nv):
_intensity = self._create_intensity(i)
ds.add_field(("gas", "intensity"),
function=_intensity,
units=self.field_units)
prj = off_axis_projection(ds, ds.domain_center, normal, width,
(self.nx, self.ny), "intensity")
self.data[:, :, i] = prj[:, :]
ds.field_info.pop(("gas", "intensity"))
pbar.update(i)
pbar.finish()
def project_photons(self,
normal,
sky_center,
absorb_model=None,
nH=None,
no_shifting=False,
north_vector=None,
sigma_pos=None,
kernel="top_hat",
prng=None,
**kwargs):
r"""
Projects photons onto an image plane given a line of sight.
Returns a new :class:`~pyxsim.event_list.EventList`.
Parameters
----------
normal : character or array-like
Normal vector to the plane of projection. If "x", "y", or "z", will
assume to be along that axis (and will probably be faster). Otherwise,
should be an off-axis normal vector, e.g [1.0, 2.0, -3.0]
sky_center : array-like
Center RA, Dec of the events in degrees.
absorb_model : string or :class:`~pyxsim.spectral_models.AbsorptionModel`
A model for foreground galactic absorption, to simulate the
absorption of events before being detected. This cannot be applied
here if you already did this step previously in the creation of the
:class:`~pyxsim.photon_list.PhotonList` instance. Known options for
strings are "wabs" and "tbabs".
nH : float, optional
The foreground column density in units of 10^22 cm^{-2}. Only used
if absorption is applied.
no_shifting : boolean, optional
If set, the photon energies will not be Doppler shifted.
north_vector : a sequence of floats
A vector defining the "up" direction. This option sets the
orientation of the plane of projection. If not set, an arbitrary
grid-aligned north_vector is chosen. Ignored in the case where a
particular axis (e.g., "x", "y", or "z") is explicitly specified.
sigma_pos : float, optional
Apply a gaussian smoothing operation to the sky positions of the
events. This may be useful when the binned events appear blocky due
to their uniform distribution within simulation cells. However, this
will move the events away from their originating position on the
sky, and so may distort surface brightness profiles and/or spectra.
Should probably only be used for visualization purposes. Supply a
float here to smooth with a standard deviation with this fraction
of the cell size. Default: None
kernel : string, optional
The kernel used when smoothing positions of X-rays originating from
SPH particles, "gaussian" or "top_hat". Default: "top_hat".
prng : integer or :class:`~numpy.random.RandomState` object
A pseudo-random number generator. Typically will only be specified
if you have a reason to generate the same set of random numbers,
such as for a test. Default is to use the :mod:`numpy.random`
module.
Examples
--------
>>> L = np.array([0.1,-0.2,0.3])
>>> events = my_photons.project_photons(L, [30., 45.])
"""
prng = parse_prng(prng)
scale_shift = -1.0 / clight.to("km/s")
if "smooth_positions" in kwargs:
issue_deprecation_warning(
"'smooth_positions' has been renamed to "
"'sigma_pos' and the former is deprecated!")
sigma_pos = kwargs["smooth_positions"]
if "redshift_new" in kwargs or "area_new" in kwargs or \
"exp_time_new" in kwargs or "dist_new" in kwargs:
issue_deprecation_warning(
"Changing the redshift, distance, area, or "
"exposure time has been deprecated in "
"project_photons!")
if sigma_pos is not None and self.parameters[
"data_type"] == "particles":
raise RuntimeError(
"The 'smooth_positions' argument should not be used with "
"particle-based datasets!")
if isinstance(absorb_model, string_types):
if absorb_model not in absorb_models:
raise KeyError("%s is not a known absorption model!" %
absorb_model)
absorb_model = absorb_models[absorb_model]
if absorb_model is not None:
if nH is None:
raise RuntimeError(
"You specified an absorption model, but didn't "
"specify a value for nH!")
absorb_model = absorb_model(nH)
sky_center = YTArray(sky_center, "degree")
n_ph = self.photons["num_photons"]
if not isinstance(normal, string_types):
L = np.array(normal)
orient = Orientation(L, north_vector=north_vector)
x_hat = orient.unit_vectors[0]
y_hat = orient.unit_vectors[1]
z_hat = orient.unit_vectors[2]
else:
x_hat = np.zeros(3)
y_hat = np.zeros(3)
z_hat = np.zeros(3)
parameters = {}
D_A = self.parameters["fid_d_a"]
events = {}
eobs = self.photons["energy"].v
if not no_shifting:
if comm.rank == 0:
mylog.info("Doppler-shifting photon energies.")
if isinstance(normal, string_types):
shift = self.photons["vel"][:,
"xyz".index(normal)] * scale_shift
else:
shift = np.dot(self.photons["vel"], z_hat) * scale_shift
doppler_shift(shift, n_ph, eobs)
if absorb_model is None:
det = np.ones(eobs.size, dtype='bool')
num_det = eobs.size
else:
if comm.rank == 0:
mylog.info("Foreground galactic absorption: using "
"the %s model and nH = %g." %
(absorb_model._name, nH))
det = absorb_model.absorb_photons(eobs, prng=prng)
num_det = det.sum()
events["eobs"] = YTArray(eobs[det], "keV")
num_events = comm.mpi_allreduce(num_det)
if comm.rank == 0:
mylog.info("%d events have been detected." % num_events)
if num_det > 0:
if comm.rank == 0:
mylog.info("Assigning positions to events.")
if isinstance(normal, string_types):
norm = "xyz".index(normal)
else:
norm = normal
xsky, ysky = scatter_events(norm, prng, kernel,
self.parameters["data_type"], num_det,
det, self.photons["num_photons"],
self.photons["pos"].d,
self.photons["dx"].d, x_hat, y_hat)
if self.parameters[
"data_type"] == "cells" and sigma_pos is not None:
if comm.rank == 0:
mylog.info("Optionally smoothing sky positions.")
sigma = sigma_pos * np.repeat(self.photons["dx"].d, n_ph)[det]
xsky += sigma * prng.normal(loc=0.0, scale=1.0, size=num_det)
ysky += sigma * prng.normal(loc=0.0, scale=1.0, size=num_det)
d_a = D_A.to("kpc").v
xsky /= d_a
ysky /= d_a
if comm.rank == 0:
mylog.info("Converting pixel to sky coordinates.")
pixel_to_cel(xsky, ysky, sky_center)
else:
xsky = []
ysky = []
events["xsky"] = YTArray(xsky, "degree")
events["ysky"] = YTArray(ysky, "degree")
parameters["exp_time"] = self.parameters["fid_exp_time"]
parameters["area"] = self.parameters["fid_area"]
parameters["sky_center"] = sky_center
return EventList(events, parameters)
def project_photons(self,
normal,
area_new=None,
exp_time_new=None,
redshift_new=None,
dist_new=None,
absorb_model=None,
sky_center=None,
no_shifting=False,
north_vector=None,
prng=None):
r"""
Projects photons onto an image plane given a line of sight.
Returns a new :class:`~pyxsim.event_list.EventList`.
Parameters
----------
normal : character or array-like
Normal vector to the plane of projection. If "x", "y", or "z", will
assume to be along that axis (and will probably be faster). Otherwise,
should be an off-axis normal vector, e.g [1.0, 2.0, -3.0]
area_new : float, (value, unit) tuple, or :class:`~yt.units.yt_array.YTQuantity`, optional
New value for the (constant) collecting area of the detector. If
units are not specified, is assumed to be in cm**2.
exp_time_new : float, (value, unit) tuple, or :class:`~yt.units.yt_array.YTQuantity`, optional
The new value for the exposure time. If units are not specified
it is assumed to be in seconds.
redshift_new : float, optional
The new value for the cosmological redshift.
dist_new : float, (value, unit) tuple, or :class:`~yt.units.yt_array.YTQuantity`, optional
The new value for the angular diameter distance, used for nearby sources.
This may be optionally supplied instead of it being determined from the
cosmology. If units are not specified, it is assumed to be in Mpc. To use this, the
redshift must be zero.
absorb_model : :class:`~pyxsim.spectral_models.AbsorptionModel`
A model for foreground galactic absorption.
sky_center : array-like, optional
Center RA, Dec of the events in degrees.
no_shifting : boolean, optional
If set, the photon energies will not be Doppler shifted.
north_vector : a sequence of floats
A vector defining the "up" direction. This option sets the orientation of
the plane of projection. If not set, an arbitrary grid-aligned north_vector
is chosen. Ignored in the case where a particular axis (e.g., "x", "y", or
"z") is explicitly specified.
prng : :class:`~numpy.random.RandomState` object or :mod:`~numpy.random`, optional
A pseudo-random number generator. Typically will only be specified
if you have a reason to generate the same set of random numbers, such as for a
test. Default is the :mod:`numpy.random` module.
Examples
--------
>>> L = np.array([0.1,-0.2,0.3])
>>> events = my_photons.project_photons(L, area_new=10000.,
... redshift_new=0.05)
"""
if prng is None:
prng = np.random
if redshift_new is not None and dist_new is not None:
mylog.error("You may specify a new redshift or distance, " +
"but not both!")
if sky_center is None:
sky_center = YTArray([30., 45.], "degree")
else:
sky_center = YTArray(sky_center, "degree")
dx = self.photons["dx"].d
if isinstance(normal, string_types):
# if on-axis, just use the maximum width of the plane perpendicular
# to that axis
w = self.parameters["Width"].copy()
w["xyz".index(normal)] = 0.0
ax_idx = np.argmax(w)
else:
# if off-axis, just use the largest width to make sure we get everything
ax_idx = np.argmax(self.parameters["Width"])
nx = self.parameters["Dimension"][ax_idx]
dx_min = (self.parameters["Width"] /
self.parameters["Dimension"])[ax_idx]
if not isinstance(normal, string_types):
L = np.array(normal)
orient = Orientation(L, north_vector=north_vector)
x_hat = orient.unit_vectors[0]
y_hat = orient.unit_vectors[1]
z_hat = orient.unit_vectors[2]
n_ph = self.photons["NumberOfPhotons"]
n_ph_tot = n_ph.sum()
parameters = {}
zobs0 = self.parameters["FiducialRedshift"]
D_A0 = self.parameters["FiducialAngularDiameterDistance"]
scale_factor = 1.0
if (exp_time_new is None and area_new is None and redshift_new is None
and dist_new is None):
my_n_obs = n_ph_tot
zobs = zobs0
D_A = D_A0
else:
if exp_time_new is None:
Tratio = 1.
else:
exp_time_new = parse_value(exp_time_new, "s")
Tratio = exp_time_new / self.parameters["FiducialExposureTime"]
if area_new is None:
Aratio = 1.
else:
area_new = parse_value(area_new, "cm**2")
Aratio = area_new / self.parameters["FiducialArea"]
if redshift_new is None and dist_new is None:
Dratio = 1.
zobs = zobs0
D_A = D_A0
else:
if dist_new is not None:
if redshift_new is not None and redshift_new > 0.0:
mylog.warning(
"Redshift must be zero for nearby sources. Resetting redshift to 0.0."
)
zobs = 0.0
D_A = parse_value(dist_new, "Mpc")
else:
zobs = redshift_new
D_A = self.cosmo.angular_diameter_distance(
0.0, zobs).in_units("Mpc")
scale_factor = (1. + zobs0) / (1. + zobs)
Dratio = D_A0*D_A0*(1.+zobs0)**3 / \
(D_A*D_A*(1.+zobs)**3)
fak = Aratio * Tratio * Dratio
if fak > 1:
raise ValueError(
"This combination of requested parameters results in "
"%g%% more photons collected than are " % (100. *
(fak - 1.)) +
"available in the sample. Please reduce the collecting "
"area, exposure time, or increase the distance/redshift "
"of the object. Alternatively, generate a larger sample "
"of photons.")
my_n_obs = np.int64(n_ph_tot * fak)
Nn = 4294967294
if my_n_obs == n_ph_tot:
if my_n_obs <= Nn:
idxs = np.arange(my_n_obs, dtype='uint32')
else:
idxs = np.arange(my_n_obs, dtype='uint64')
else:
if n_ph_tot <= Nn:
idxs = np.arange(n_ph_tot, dtype='uint32')
prng.shuffle(idxs)
idxs = idxs[:my_n_obs]
else:
Nc = np.int32(n_ph_tot / Nn)
idxs = np.zeros(my_n_obs, dtype=np.uint64)
Nup = np.uint32(my_n_obs / Nc)
for i in range(Nc + 1):
if (i + 1) * Nc < n_ph_tot:
idtm = np.arange(i * Nc, (i + 1) * Nc, dtype='uint64')
Nupt = Nup
else:
idtm = np.arange(i * Nc, n_ph_tot, dtype='uint64')
Nupt = my_n_obs - i * Nup
prng.shuffle(idtm)
idxs[i * Nup, i * Nup + Nupt] = idtm[:Nupt]
del (idtm)
# idxs = prng.permutation(n_ph_tot)[:my_n_obs].astype("int64")
obs_cells = np.searchsorted(self.p_bins, idxs, side='right') - 1
delta = dx[obs_cells]
if isinstance(normal, string_types):
if self.parameters["DataType"] == "cells":
xsky = prng.uniform(low=-0.5, high=0.5, size=my_n_obs)
ysky = prng.uniform(low=-0.5, high=0.5, size=my_n_obs)
elif self.parameters["DataType"] == "particles":
xsky = prng.normal(loc=0.0, scale=1.0, size=my_n_obs)
ysky = prng.normal(loc=0.0, scale=1.0, size=my_n_obs)
xsky *= delta
ysky *= delta
xsky += self.photons[axes_lookup[normal][0]].d[obs_cells]
ysky += self.photons[axes_lookup[normal][1]].d[obs_cells]
if not no_shifting:
vz = self.photons["v%s" % normal]
else:
if self.parameters["DataType"] == "cells":
x = prng.uniform(low=-0.5, high=0.5, size=my_n_obs)
y = prng.uniform(low=-0.5, high=0.5, size=my_n_obs)
z = prng.uniform(low=-0.5, high=0.5, size=my_n_obs)
elif self.parameters["DataType"] == "particles":
x = prng.normal(loc=0.0, scale=1.0, size=my_n_obs)
y = prng.normal(loc=0.0, scale=1.0, size=my_n_obs)
z = prng.normal(loc=0.0, scale=1.0, size=my_n_obs)
if not no_shifting:
vz = self.photons["vx"]*z_hat[0] + \
self.photons["vy"]*z_hat[1] + \
self.photons["vz"]*z_hat[2]
x *= delta
y *= delta
z *= delta
x += self.photons["x"].d[obs_cells]
y += self.photons["y"].d[obs_cells]
z += self.photons["z"].d[obs_cells]
xsky = x * x_hat[0] + y * x_hat[1] + z * x_hat[2]
ysky = x * y_hat[0] + y * y_hat[1] + z * y_hat[2]
del (delta)
if no_shifting:
eobs = self.photons["Energy"][idxs]
else:
# shift = -vz.in_cgs()/clight
# shift = np.sqrt((1.-shift)/(1.+shift))
# eobs = self.photons["Energy"][idxs]*shift[obs_cells]
shift = -vz[obs_cells].in_cgs() / clight
shift = np.sqrt((1. - shift) / (1. + shift))
eobs = self.photons["Energy"][idxs]
eobs *= shift
del (shift)
eobs *= scale_factor
if absorb_model is None:
detected = np.ones(eobs.shape, dtype='bool')
else:
detected = absorb_model.absorb_photons(eobs, prng=prng)
events = {}
dtheta = YTQuantity(np.rad2deg(dx_min / D_A), "degree")
events["xpix"] = xsky[detected] / dx_min.v + 0.5 * (nx + 1)
events["ypix"] = ysky[detected] / dx_min.v + 0.5 * (nx + 1)
events["eobs"] = eobs[detected]
events = comm.par_combine_object(events, datatype="dict", op="cat")
num_events = len(events["xpix"])
if comm.rank == 0:
mylog.info("Total number of observed photons: %d" % num_events)
if exp_time_new is None:
parameters["ExposureTime"] = self.parameters[
"FiducialExposureTime"]
else:
parameters["ExposureTime"] = exp_time_new
if area_new is None:
parameters["Area"] = self.parameters["FiducialArea"]
else:
parameters["Area"] = area_new
parameters["Redshift"] = zobs
parameters["AngularDiameterDistance"] = D_A.in_units("Mpc")
parameters["sky_center"] = sky_center
parameters["pix_center"] = np.array([0.5 * (nx + 1)] * 2)
parameters["dtheta"] = dtheta
return EventList(events, parameters)
