def time_delay(theta,
phi,
ifo1,
ifo2,
coord="E",
tgeocent=None,
degrees=False):
"""
returns the time delay between two ifo's (in seconds)
interprets direction as (theta, phi) or (dec, ra) depending on \"coord\"
"""
if ifo1 not in detectors.keys():
raise ValueError("ifo1=%s not understood" % ifo1)
if ifo2 not in detectors.keys():
raise ValueError("ifo2=%s not understood" % ifo2)
dr = detectors[ifo1].dr - detectors[ifo2].dr
if degrees:
theta *= deg2rad
phi *= deg2rad
if coord == "C":
theta, phi = __celest2earth(theta, phi, tgeocent)
theta = theta % np.pi
phi = phi % (2 * np.pi)
if isinstance(theta, float):
return np.sum(hp.ang2vec(theta, phi) * dr)
else:
return np.sum(hp.ang2vec(theta, phi) * dr, axis=1)
def test_pointingTree(fname, opsimversion, tableNames, angleUnit):
"""test that `PointingTree` is set up correctly to handle
pointings provided by `OpSimOutput.summary` by checking that
distance calculations for the 10 nearest ones match
"""
fname = os.path.join(oss.example_data, fname)
ops_out = OpSimOutput.fromOpSimDB(fname,
opsimversion=opsimversion,
tableNames=tableNames)
pointings = ops_out.summary.iloc[:100]
# Find distances to a specific location ra=54.0 deg, dec=-27.5 deg
radeg = 54.0
decdeg = -27.5
ra = np.radians(radeg)
dec = np.radians(decdeg)
ptree = PointingTree(pointings, raCol='_ra', decCol='_dec')
# Find the 10 nearest pointings
dists, idxs = ptree.tree.query([(dec, ra)], k=10)
# Now calculate the distances more directly
# order the 10 pointings as above
obs = pointings.iloc[idxs[0]]
vec = hp.ang2vec(radeg, decdeg, lonlat=True)
vecs = hp.ang2vec(np.degrees(obs._ra), np.degrees(obs._dec), lonlat=True)
assert_allclose(np.dot(vecs, vec), np.cos(dists[0]))
def mask_unseen(hpmap, az, el, doconvert = False, nest = False):
'''
Mask unseen directions in the scan. Use as input the house keeping coordinates.
Parameters:
hpmap. Healpix map (npix)
az and el. Azimuth and Elevation coordinates read from azimuth.fits and elevation.fits
doconvert. Consider the data as real azimuth and elevation
(not implemented in that way in demodulation yet).
Return:
hpmap[masked]
'''
if doconvert:
hkcoords = np.meshgrid(az, el)
radec = qubic.hor2equ(hkcoords[0].ravel(), hkcoords[1].ravel(), 0)
phi = radec[0]
theta = radec[1]
#Rotation from Az,El housekiping to Az, El = 0,0
newcoords = np.dot(sbfit.rotmatY(qubic.hor2equ(0,+50,0)[1]),
hp.ang2vec(np.pi/2-np.deg2rad(theta), np.deg2rad(phi)).T).T
else:
hkcoords = np.meshgrid(az, el-50)
phi = hkcoords[0].ravel()
theta = hkcoords[1].ravel()
newcoords = hp.ang2vec(np.pi/2-np.deg2rad(theta), np.deg2rad(phi))
nside = hp.get_nside(hpmap)
coordspix = hp.vec2pix(nside, newcoords[...,0], newcoords[...,1], newcoords[...,2], nest = nest)
mask = np.zeros((12 * nside **2 ), dtype = bool)
mask[coordspix] = 1
hpmap[~mask] = hp.UNSEEN
#hp.mollview(hpmap, nest = True)
show()
return hpmap
def test_synopsimPTree(fname, opsimversion, tableNames, angleUnit):
"""
test that the pointings for a particular point at ra = 54.0 deg and
dec = -27.5 deg are found through the pointing tree and through a
calculation through the vectors and dot products match
"""
fname = os.path.join(oss.example_data, fname)
synopsim = SynOpSim.fromOpSimDB(fname,
opsimversion=opsimversion,
tableNames=tableNames,
usePointingTree=True,
angleUnit=angleUnit)
# Find distances to a specific location ra=54.0 deg, dec=-27.5 deg
radeg = 54.0
decdeg = -27.5
pts = synopsim.pointingTree.pointingsEnclosing(ra=radeg,
dec=decdeg,
circRadius=0.,
pointingRadius=1.75)
res = np.sort(pts[0])
# calculate differently
X = synopsim.pointings[['_dec', '_ra']].apply(np.degrees)
dec = X.values[:, 0]
ra = X.values[:, 1]
vecs = hp.ang2vec(ra, dec, lonlat=True)
vec = hp.ang2vec(radeg, decdeg, lonlat=True)
X.loc[:, 'dist'] = np.dot(vecs, vec)
costhresh = np.cos(np.radians(1.75))
assert_array_equal(np.sort(X.query('dist > @costhresh').index), res)
def get_points(self, idx, get_childs=False):
x = []
y = []
z = []
c = []
if len(idx.split("cen_")) > 1:
unit_vector = hp.ang2vec(self.centers[idx].phi,
self.centers[idx].theta,
lonlat=True)
curr_vector = (unit_vector * self.centers[idx].r)[0]
#print(curr_vector)
x.append(curr_vector[0])
y.append(curr_vector[1])
z.append(curr_vector[2])
c.append("b")
if not get_childs:
return x, y, z, c
for key in self.centers[idx].childs:
x_, y_, z_, c_ = self.get_points(key, get_childs)
x.append(x_[0])
y.append(y_[0])
z.append(z_[0])
c.append(c_[0])
if len(idx.split("rim_")) > 1:
unit_vector = hp.ang2vec(self.rims[idx].phi,
self.rims[idx].theta,
lonlat=True)
curr_vector = (unit_vector * self.rims[idx].r)[0]
x.append(curr_vector[0])
y.append(curr_vector[1])
z.append(curr_vector[2])
c.append("red")
if not get_childs:
return x, y, z, c
for key in self.rims[idx].childs:
x_, y_, z_, c_ = self.get_points(key, get_childs)
x.append(x_[0])
y.append(y_[0])
z.append(z_[0])
c.append(c_[0])
if len(idx.split("cenClump_")) > 1:
unit_vector = hp.ang2vec(self.clumps_center[idx].phi,
self.clumps_center[idx].theta,
lonlat=True)
curr_vector = (unit_vector * self.clumps_center[idx].r)[0]
x.append(curr_vector[0])
y.append(curr_vector[1])
z.append(curr_vector[2])
c.append("green")
if len(idx.split("flatClump_")) > 1:
unit_vector = hp.ang2vec(self.clumps_flat[idx].phi,
self.clumps_flat[idx].theta,
lonlat=True)
curr_vector = (unit_vector * self.clumps_flat[idx].r)[0]
x.append(curr_vector[0])
y.append(curr_vector[1])
z.append(curr_vector[2])
c.append("blue")
return x, y, z, c
def getCirclePixels(ra_pointing,
dec_pointing,
radius,
nside,
alpha=0.4,
color='k',
edgecolor='k',
rotation=None):
theta = 0.5 * np.pi - np.deg2rad(dec_pointing)
phi = np.deg2rad(ra_pointing)
xyz = hp.ang2vec(theta, phi)
ipix = hp.query_disc(nside, xyz, np.deg2rad(radius))
radecs = get_ellipse_coords(a=radius / np.cos(np.deg2rad(dec_pointing)),
b=radius,
x=ra_pointing,
y=dec_pointing,
angle=0.0,
npts=25)
idx = np.where(radecs[:, 1] > 90.0)[0]
radecs[idx, 1] = 180.0 - radecs[idx, 1]
idx = np.where(radecs[:, 1] < -90.0)[0]
radecs[idx, 1] = -180.0 - radecs[idx, 1]
idx = np.where(radecs[:, 0] > 360.0)[0]
radecs[idx, 0] = 720.0 - radecs[idx, 0]
idx = np.where(radecs[:, 0] < 0.0)[0]
radecs[idx, 0] = 360.0 + radecs[idx, 0]
radecs = np.array(radecs)
idx1 = np.where(radecs[:, 0] >= 180.0)[0]
idx2 = np.where(radecs[:, 0] < 180.0)[0]
idx3 = np.where(radecs[:, 0] > 300.0)[0]
idx4 = np.where(radecs[:, 0] < 60.0)[0]
if (len(idx1) > 0
and len(idx2) > 0) and not (len(idx3) > 0 and len(idx4) > 0):
alpha = 0.0
xyz = hp.ang2vec(radecs[:, 0], radecs[:, 1], lonlat=True)
proj = hp.projector.MollweideProj(rot=rotation, coord=None)
x, y = proj.vec2xy(xyz[:, 0], xyz[:, 1], xyz[:, 2])
xy = np.zeros(radecs.shape)
xy[:, 0] = x
xy[:, 1] = y
#path = matplotlib.path.Path(xyz[:,1:3])
path = matplotlib.path.Path(xy)
patch = matplotlib.patches.PathPatch(path,
alpha=alpha,
color=color,
fill=True,
zorder=3,
edgecolor=edgecolor)
area = np.pi * radius**2
return ipix, radecs, patch, area
def pointingsEnclosing(self, ra, dec, circRadius=0., pointingRadius=1.75,
usePointingTree=None, transform=None, subset='all'):
"""
Helper method returning a generator of pointings overlapping with
circles of radius `circRadius around a sequence of positions given in
terms of `ra` and `dec`. The calculation uses a `Tree` to make the
calculations more efficient if `usePointingTree` is True, or uses direct
calculations if this variable is set to `False`. A user may choose to
obtain a subset of the `pointing` by supplying a subset in the form a
list via the parameter `subset`.
Parameters
----------
ra : `np.ndarray` or a float, unit of degrees
a float or an array of floats representing the ra values
dec : `np.ndarray` or a float, unit of degrees
a float or an array of floats representing the dec values
circRadius: float, unit of degrees, defaults to 0.
a circle around each of the
pointingRadius : degrees, defaults to 1.75
radius of the field of view
usePointingTree: {None|True|False}, defaults to `None`
if None, usePointingTree = self.usePointingTree
else the variable takes the {True|False} values assigned
transform: function, Not implemented
subset: (list of strings| 'all')
if 'all', df is returned. Otherwise, return
df[subset], with the same index.
Returns
-------
A generator with the pointings that overlap with ra, dec
.. note: 1. the values in the generator may be accessed by next(generator)
2. subset=[] returns only the index
"""
print('check using ptree', usePointingTree)
if transform is not None:
raise NotImplementedError('transforms are not implemented yet')
if usePointingTree is None:
usePointingTree = self.usePointingTree
if usePointingTree:
hidxs = self.pointingTree.pointingsEnclosing(ra, dec, circRadius,
pointingRadius)
for hidx in hidxs:
yield self.df_subset_columns(self.pointings.loc[hidx], subset)
else:
print(self.pointings['_ra'].max())
x = self.pointings[['_ra', '_dec']].copy().apply(np.degrees)
pvecs = hp.ang2vec(x._ra, x._dec, lonlat=True)
vecs = hp.ang2vec(ra, dec, lonlat=True)
prad = np.radians(pointingRadius + circRadius)
for vec in vecs:
x.loc[:, 'dist'] = np.arccos(np.dot(pvecs, vec))
idx = x.query('dist < @prad').index
yield self.df_subset_columns(self.pointings.loc[idx], subset)
def makeGauss(nside, ra, dec, amp,
varparam=np.sqrt(2 * np.pi)): #bounds=None):#radius=6447797.0):
print('RA length:' + str(len(ra)))
print('dec length:' + str(len(dec)))
print('amp length:' + str(len(amp)))
raMin = min(ra) - 2.0
raMax = max(ra) + 2.0
decMin = min(dec) - 2.0
decMax = max(dec) + 2.0
#m = np.arange(hp.nside2npix(nside))
vertices = hp.ang2vec([raMin, raMin, raMax, raMax],
[decMin, decMax, decMax, decMin],
lonlat=True)
m = hp.query_polygon(nside, vertices)
m_ra, m_dec = hp.pix2ang(nside, m, lonlat=True)
m_amp = np.zeros_like(m, dtype='float64')
#ra = ra*1.0
#dec = dec*1.0
numpix = m.size
#if True: #bounds != None:
# raMin = min(ra)-2.0
# raMax = max(ra)+2.0
# decMin = min(dec)-2.0
# decMax = min(dec)+2.0
# Compute distance of center of data to each pixel
# centDist = haversine(np.full_like(m,raCent),np.full_like(m,decCent),m_ra,m_dec)
# Get indices of pixels to delete
# toDelete = [ k for (i,j,k) in zip(m_ra,m_dec,m) if i < raMin or i > raMax or j < decMin or j > decMax ]
# #m = np.delete(m, toDelete)
# m_ra = np.delete(m_ra, toDelete)
# m_dec = np.delete(m_dec, toDelete)
# m_amp = np.delete(m_amp, toDelete)
# numdeleted = str(len(toDelete))
# print(numdeleted + ' uninteresting pixels deleted of '+str(numpix))
### TO IMPLEMENT: Check ra, dec, amp are of same size
#if isinstance(ra, np.ndarray):
for rai, deci, ampi in itertools.izip(ra, dec, amp):
# Used to convert from ra/dec to pixel to vec; should not be necessary
#nearpix = hp.pix2vec(nside,hp.ang2pix(nside,rai,deci,lonlat=True))
nearpix = hp.ang2vec(rai, deci, lonlat=True)
# Calculate Gaussian on pixels within following radius (in degrees)
radius = 3.0 / 60.0 # Hard-coded; remove eventually
pxls = hp.query_disc(nside, nearpix, radius * np.pi / 180.0)
indices = np.searchsorted(m, pxls)
near_ra, near_dec = hp.pix2ang(nside, pxls, lonlat=True)
#print('Using nearest '+str(len(near_ra))+' pixels to source.')
#near_ra = m_ra[pxls]
#near_dec = m_dec[pxls]
m_dist = np.zeros_like(near_ra)
var = varparam * 1.0
m_dist = haversine(np.full_like(near_ra, rai),
np.full_like(near_ra, deci), near_ra, near_dec)
m_amp[indices] += computeGauss(ampi, m_dist, var)
print('makeGauss output length: ' + str(m_ra.size))
return m_ra, m_dec, m_dist, m_amp
def make_ring_mask(inner, outer, ring_b, ring_l, nside):
""" Masks outside inner < r < outer, of a ring centred at (ring_b,ring_l)
"""
mask_none = np.arange(hp.nside2npix(nside))
return np.logical_not(
(np.cos(np.radians(inner)) >= np.dot(
hp.ang2vec(np.radians(90 - ring_b), np.radians(ring_l)),
hp.pix2vec(nside, mask_none))) *
(np.dot(hp.ang2vec(np.radians(90 - ring_b), np.radians(ring_l)),
hp.pix2vec(nside, mask_none)) >= np.cos(np.radians(outer))))
def integrate_intensity_map(Imap,nside,latmin=-2,latmax=2. ,nsteps_long=500,rad_units=False,planck_map=False):
"""
Compute the integral of the intensity map along latitude and longitude; to compare observed
intensity map and the model one.
To check consistency of the model we compute the integral as in eqs.(6) and (7) of
`Puglisi+ 2017 <http://arxiv.org/abs/1701.07856>`_.
*Parameters*
- `Imap`:{array}
intensity map
- `nside`: {int}
:mod:`healpy` gridding parameter
- `latmin`, `latmax`:{double}
minimum and maximum latitudes in `degree` where to perform the integral (default :math:`\pm 2\, deg`)
if you have the angles in radiants set `rad_units` to `True`.
- `nsteps_long`:{int}
number of longitudinal bins, (default 500)
- `planck_map`:{bool}
if set to `True`, it sets to zero all the :mod:`healpy.UNSEEN` masked pixels of the map,
(useful when dealing with observational maps).
**Returns**
- `I_l` :{array}
latitude integration within the set interval :math:`[b_{min}, b_{max}]`
- `I_tot`:{double}
integration of `I_l` in :math:`\ell \in [0,2 \pi]`.
"""
if planck_map:
arr=np.ma.masked_equal(Imap,hp.UNSEEN)
Imap[arr.mask]=0.
if not rad_units:
latmin=np.pi/2.+(np.deg2rad(latmin))
latmax=np.pi/2.+(np.deg2rad(latmax))
nbins_long=nsteps_long-1
long_edges=np.linspace(0.,2*np.pi,num=nsteps_long)
long_centr=[.5*(long_edges[i]+ long_edges[i+1]) for i in range(nbins_long)]
listpix=[]
for i in range(nbins_long):
v=[ hp.ang2vec(latmax, long_edges[i]),
hp.ang2vec(latmax, long_edges[i+1]),
hp.ang2vec(latmin, long_edges[i+1]),
hp.ang2vec(latmin, long_edges[i])]
listpix.append(hp.query_polygon(nside,v))
delta_b=pixelsize(nside,arcmin=False)
delta_l=2*np.pi/nbins_long
I_l=[sum(Imap[l])*delta_b for l in listpix ]
Itot= sum(I_l)*delta_l
return Itot,I_l
def angsep2(lon_1, lat_1, lon_2, lat_2):
"""
Angular separation (deg) between two sky coordinates.
"""
import healpy
v10, v11, v12 = healpy.ang2vec(np.radians(90. - lat_1), np.radians(lon_1)).transpose()
v20, v21, v22 = healpy.ang2vec(np.radians(90. - lat_2), np.radians(lon_2)).transpose()
val = (v10 * v20) + (v11 * v21) + (v12 * v22)
val = np.clip(val, -1., 1.)
return np.degrees(np.arccos(val))
def angsep2(lon_1, lat_1, lon_2, lat_2):
"""
Angular separation (deg) between two sky coordinates.
"""
import healpy
v10, v11, v12 = healpy.ang2vec(np.radians(90. - lat_1), np.radians(lon_1)).transpose()
v20, v21, v22 = healpy.ang2vec(np.radians(90. - lat_2), np.radians(lon_2)).transpose()
val = (v10 * v20) + (v11 * v21) + (v12 * v22)
val = np.clip(val, -1., 1.)
return np.degrees(np.arccos(val))
def antipode(x, y, coord="C", degrees=False):
"""
if coord=C : x->ra, y->dec
if coord=E : x->phi, y->theta
returns the anitpode position
"""
if coord == "C":
Y, X = hp.vec2ang(-hp.ang2vec(0.5 * np.pi - y, x))
Y = 0.5 * np.pi - Y
elif coord == "E":
Y, X = hp.vec2ang(-hp.ang2vec(y, x))
return X, Y
def gamma_from_pol(gal, pol, fixed_basis=False):
'''
It should be possible to distribute the inner product among all time
observations, but the matrix algebra is escaping me right now. In any case,
for a one time operation this doesn't seem too slow yet.
'''
# gal and pol are galactic lonlat vectors
dir_A_gal = hp.ang2vec(gal[:, 0] % np.pi,
gal[:, 1] % (2 * np.pi),
lonlat=False)
dir_A_pol = hp.ang2vec(pol[:, 0] % np.pi,
pol[:, 1] % (2 * np.pi),
lonlat=False)
dir_Z = np.array([0, 0, 1])
sin_theta_A = np.sqrt(dir_A_gal[:, 0]**2 + dir_A_gal[:, 1]**2)
dir_A_west_x = dir_A_gal[:, 1] / sin_theta_A
dir_A_west_y = -dir_A_gal[:, 0] / sin_theta_A
dir_A_west_z = dir_A_gal[:, 1] * 0
dir_A_west = np.array([dir_A_west_x, dir_A_west_y, dir_A_west_z]).T
dir_A_north = (dir_Z - dir_A_gal[2] * dir_A_gal) / sin_theta_A[:,
np.newaxis]
'''
if sin_theta_A == 0:
dir_A_west = np.array([1,0,0])
dir_A_north = np.array([0,1,0])
assert dir_A_north.dot(dir_A_west) == approx(0), 'Vectors not orthogonal'
assert dir_A_north.dot(dir_A_north) == approx(1), 'North-vector not normalized'
assert dir_A_west.dot(dir_A_west) == approx(1), 'North-vector not normalized'
'''
sin_gamma_A = dir_A_pol[:,
0] * dir_A_west[:,
0] + dir_A_pol[:,
1] * dir_A_west[:,
1] + dir_A_pol[:,
2] * dir_A_west[:,
2]
cos_gamma_A = dir_A_pol[:,
0] * dir_A_north[:,
0] + dir_A_pol[:,
1] * dir_A_north[:,
1] + dir_A_pol[:,
2] * dir_A_north[:,
2]
cos_2_gamma_A = 2 * cos_gamma_A**2 - 1
sin_2_gamma_A = 2 * sin_gamma_A * cos_gamma_A
return sin_2_gamma_A, cos_2_gamma_A
def orthoslant_project(shell, center, radius, degrees=False):
"""
Take a conical section of a sphere and bin into a cube for power spectrum estimation.
"""
Npix, Nfreq = shell.shape
Nside = hp.npix2nside(Npix)
if degrees:
radius *= np.pi/180.
if len(center) == 2:
# This is an (az, alt) coordinate.
if degrees:
center = hp.ang2vec(center[0], center[1], lonlat=True)
else:
center = hp.ang2vec(np.pi - center[1], center[0])
# Define xy grid by the pixel area and selection radius.
radpix = hp.nside2resol(Nside)
extent = 2*np.floor(radius/radpix).astype(int)
orthogrid = np.zeros((extent,extent, Nfreq))
# Get vectors, rotate so center is overhead, project vectors to get xy bins
hpx_inds = hp.query_disc(Nside, center, 2*np.sqrt(2)*radius) #TODO This can be replaced with a "query_polygon"
vecs = hp.pix2vec(Nside, hpx_inds)
dp, dt = hp.rotator.vec2dir(center, lonlat=True)
rot = hp.rotator.Rotator(rot=(dp, dt, 0))
vx, vy, vz = hp.rotator.rotateVector(rot.mat, vecs)
## Project onto the yz plane, having rotated so the x axis is at zenith
xinds = np.floor(vy/radpix).astype(int) + extent/2
yinds = np.floor(vz/radpix).astype(int) + extent/2 #Center on the grid
boxinds = np.where((xinds < extent)&(xinds>-1)&(yinds < extent)&(yinds>-1))
xinds = xinds[boxinds]
yinds = yinds[boxinds] #Radial selection was bigger than the grid
hpx_inds = hpx_inds[boxinds]
weights = np.ones((orthogrid.shape[0], orthogrid.shape[1]))
#np.add.at(weights, (xinds,yinds), 1.0)
#nz = np.where(weights > 0)
#weights[nz] = 1/weights[nz]
for fi in np.arange(Nfreq):
np.add.at(orthogrid[:,:,fi],(xinds, yinds), shell[hpx_inds,fi])
orthogrid[:,:,fi] *= weights
return orthogrid
def plot_exposures(pointings,
Aeff_fact,
index=1,
lat=0.,
lon=np.radians(260.),
Earth=True,
antiEarth=False,
NSIDE=32,
doplot=True):
npointings = len(pointings)
sc = Spacecraft(pointings, lat=lat, lon=lon)
exposure_positions_hp = np.arange(hp.nside2npix(NSIDE))
exposure_positions_pix = hp.pix2ang(NSIDE,
exposure_positions_hp,
lonlat=True)
exposure_positions = np.vstack(exposure_positions_pix)
exposures = np.array([[
detector.exposure(position[0], position[1], alt=-90., index=index)
for position in exposure_positions.T
] for detector in sc.detectors])
exps = exposures.sum(axis=0) * Aeff_fact
fs = exps #-min(gbm_exps))/max(gbm_exps)
if Earth:
vec = hp.ang2vec(180, 0, lonlat=True)
i = hp.query_disc(NSIDE, vec, 67 * np.pi / 180.)
fs[i] = 0
exposures[:, i] = 0
if antiEarth:
vec = hp.ang2vec(np.degrees(lon) - 260. + 180., 0, lonlat=True)
i = hp.query_disc(NSIDE, vec, 67 * np.pi / 180.)
fs[i] = 0
exposures[:, i] = 0
if doplot:
plot.figure(figsize=(20, npointings))
s = np.argsort(pointings.keys())
for j in range(npointings):
i = s[j]
hp.mollview(exposures[i]/max(exposures[i])*Aeff_fact,title='Detector ',\
sub = [np.round(npointings/3.+0.5),3,int(str(j+1))])
#+pointings.keys()[i],\
hp.mollview(fs, title='Sum of All Detectors')
# plot.savefig(biadir+'exposure_maps_'+str(ang)+'.png')
return sc, fs, exposure_positions, pointings, exposures
def addring(nside, ra, dec, anga, angb):
"""
take a map, and add 1 to elements between anga and angb from ra, dec
"""
theta = np.deg2rad(90-dec)
phi = np.deg2rad(ra)
temp_map = np.zeros(hp.nside2npix(nside))
assert angb > anga # else error
# Everything from 0 to angb = 1
pixlist = hp.query_disc(nside, hp.ang2vec(theta, phi), np.deg2rad(angb))
temp_map[pixlist] += 1
# now delete everything from 0 to anga
pixlist = hp.query_disc(nside, hp.ang2vec(theta, phi), np.deg2rad(anga))
temp_map[pixlist] -= 1
return temp_map
def _get_response_B(theta, phi, spectral_irradiance, nu, horn, primary_beam):
"""
Return the complex electric amplitude and phase [W^(1/2)] from sources
of specified spectral irradiance [W/m^2/Hz] going through each horn.
Parameters
----------
theta : array-like
The source zenith angle [rad].
phi : array-like
The source azimuthal angle [rad].
spectral_irradiance : array-like
The source spectral power per unit surface [W/m^2/Hz].
nu : float
The frequency for which the response is computed [Hz].
horn : PackedArray
The horn layout.
primary_beam : Beam
The primary beam.
Returns
-------
out : complex array of shape (#horns, #sources)
The phase and amplitudes from the sources to the horns.
"""
shape = np.broadcast(theta, phi, spectral_irradiance).shape
theta, phi, spectral_irradiance = [np.ravel(_) for _ in theta, phi, spectral_irradiance]
uvec = hp.ang2vec(theta, phi)
source_E = np.sqrt(spectral_irradiance * primary_beam(theta, phi) * np.pi * horn.radius ** 2)
const = 2j * np.pi * nu / c
product = np.dot(horn[horn.open].center, uvec.T)
out = ne.evaluate("source_E * exp(const * product)")
return out.reshape((-1,) + shape)
def gc2radec(self, gc_phi, gc_theta, min_lon=None):
"""
convert coordinates in galactic coordinate system to equatorial
ra, dec.
Parameters
----------
gc_phi : `np.ndarray`, float
angular coordinate in Galactic coordinate system, where the MW
disk is at theta=0.
gc_theta : `np.ndarray`, float
azimuthal coordinate in Galactic coordinate system.
min_lon : degrees, defaults to None
min longitude beyond which the galaxy is cut off. If None,
`self.min_lon` is used.
max_lon : None, Not implemented yet
"""
vec = hp.ang2vec(gc_theta, gc_phi)
vec_radec = np.asarray(list(np.dot(self.rmat_inv, v) for v in vec))
theta, phi = hp.vec2ang(vec_radec)
ra, dec = convertToCelestialCoordinates(theta, phi)
if min_lon is None:
min_lon = self.min_lon
mask = dec > min_lon
return ra[mask], dec[mask]
def _get_mask(self):
if self.config.get('mask_file', None) is not None:
mask = hp.read_map(self.config['mask_file'])
mask = hp.ud_grade(rotate_mask(mask, self.rot),
nside_out=self.nside)
mask[mask > 0.5] = 1.
mask[mask <= 0.5] = 0.
else:
mask = np.ones(self.npix)
r = hp.Rotator(coord=['C', 'G'])
RApix, DEpix = hp.pix2ang(self.nside,
np.arange(self.npix),
lonlat=True)
lpix, bpix = r(RApix, DEpix, lonlat=True)
# angular conditions
mask[(DEpix < self.config.get('DEC_min_deg', -40)) |
(np.fabs(bpix) < self.config.get('GLAT_max_deg', 5))] = 0
if self.file_sourcemask is not None:
# holes catalog
RAmask, DEmask, radmask = np.loadtxt(self.file_sourcemask,
unpack=True)
vecmask = hp.ang2vec(RAmask, DEmask, lonlat=True)
for vec, radius in zip(vecmask, radmask):
ipix_hole = hp.query_disc(self.nside,
vec,
np.radians(radius),
inclusive=True)
mask[ipix_hole] = 0
mask = rotate_mask(mask, self.rot, binarize=True)
return mask
def write_k2_moc(campaign=0, norder_moc=NORDER_MOC, output_fn=None):
if output_fn is None:
fieldinfo = getFieldInfo(campaign)
if "preliminary" in fieldinfo:
output_fn = "../moc/k2-footprint-c{:02d}-proposed.moc".format(campaign)
else:
output_fn = "../moc/k2-footprint-c{:02d}.moc".format(campaign)
# Obtain the footprint corners in polar coordinates
log.info("Preparing footprint polygons for C{}".format(campaign))
polygons = []
for _, channel in FOOTPRINT["c{}".format(campaign)]["channels"].items():
polygon = [np.pi/2. - np.radians(channel["corners_dec"]),
np.radians(channel["corners_ra"])]
polygons.append(polygon)
# Obtain the healpix diamonds that cover the polygons entirely
# and add these to a `MOC` object
log.info("Converting polygons into healpix format")
moc = mocpy.MOC(moc_order=norder_moc)
for p in polygons:
pix_list = hp.query_polygon(2**norder_moc,
hp.ang2vec(p[0], p[1]),
inclusive=True, nest=True)
for pix in pix_list:
moc.add_pix(norder_moc, pix)
# Finally, write the resulting MOC file to disk
log.info("Writing {}".format(output_fn))
moc.plot() # IMPORTANT! moc.write is corrupt if plot is not called first
moc.write(output_fn)
def pencil(self, theta, phi, angrad, **kwargs):
"""Query a pencil beam from this `SkyMap`
Returns the subset of this `SkyMap` that covers an angular disc on
the sky (i.e., a pencil beam)
Parameters
----------
theta : `float`
zenith angle (radians) at the center of the disc
phi : `float`
azimuth angle (radians) at the center of the disc
angrad : `float` or `~astropy.units.Quantity`
angular radius (radians) subtended by the disc
**kwargs : `dict`, optional
additional keyword arguments to `~healpy.query_disc`
Returns
-------
out : `SkyMap`
the subset of `SkyMap` subtended by this pencil beam
"""
if isinstance(angrad, units.Quantity):
angrad = angrad.to("rad").value
direction = healpy.ang2vec(theta, phi)
indices = healpy.query_disc(self.nside,
direction,
angrad,
nest=self.nest,
**kwargs)
return self[indices]
def angToVec(lon, lat, convention, unit):
"""
Compute an array of unit vectors corresponding to the array of longitude
and latitude in either celestial conventions (ra, dec) or usual spherical
coordinate system conventions.
Parameters
----------
lon : scalar or `numpy.ndarray` of floats
co-longitude which may be ra, or phi depending on the convention
lat : scalar or `numpy.ndarray` of floats
co-latitude which may be dec, or theta depending on the convention
convention : {'celestial'|'spherical'}
whether the input angular positions are in spherical or celestial
coordinates.
unit : string
{'degrees'| 'radians'}
"""
if unit.lower() not in ('degrees', 'radians'):
raise ValueError('illegal value for units, '
'must be either degrees or radians')
if convention.lower() not in ('celestial', 'spherical'):
raise ValueError('illegal value for convention, '
'must be either spherical or celestial')
lon = np.ravel(lon)
lat = np.ravel(lat)
if convention.lower() == 'celestial':
theta, phi = convertToSphericalCoordinates(lon, lat, unit=unit.lower())
if convention.lower() == 'spherical' and unit.lower() == 'degrees':
theta = np.radians(lat)
phi = np.radians(lon)
return hp.ang2vec(theta, phi)
def grow_hp(inmap, hpids, radius=1.75, replace_val=np.nan):
"""
grow a healpix mask
Parameters
----------
inmap : np.array
A HEALpix map
hpids : array
The healpixel values to grow around
radius : float (1.75)
The radius to grow around each point (degrees)
replace_val : float (np.nan)
The value to plug into the grown areas
"""
nside = hp.npix2nside(np.size(inmap))
theta, phi = hp.pix2ang(nside=nside, ipix=hpids)
vec = hp.ang2vec(theta, phi)
ipix_disc = [
hp.query_disc(nside=nside, vec=vector, radius=np.radians(radius))
for vector in vec
]
ipix_disc = np.unique(np.concatenate(ipix_disc))
outmap = inmap + 0
outmap[ipix_disc] = replace_val
return outmap
def find_field_visits(visits, ra, decl, nside=512, field_radius_deg=1.75):
"""Return visits centered near a pointing
Parameters
----------
visits : `pandas.DataFrame`
The visits in which to look for fields
ra : `float`
The RA around which to search.
decl : `float`
The declination around which to search
nside : `int`
The nside for the healpix search
field_radious_deg : `float`
The radious round which to search, in degrees
Returns
-------
field_visits : `pandas.DataFrame`
The visits on the field.
"""
field_hpxs = healpy.query_disc(
nside,
healpy.ang2vec(ra, decl, lonlat=True),
np.radians(field_radius_deg),
)
visit_hpxs = healpy.ang2pix(nside,
visits["fieldRA"].values,
visits["fieldDec"].values,
lonlat=True)
field_visits = visits.loc[np.isin(visit_hpxs, field_hpxs)]
return field_visits
def pixels_with_clusters(clusters_files, big_pixels, nside, min_rad=0.62):
from astropy.coordinates import SkyCoord
from astropy import units as u
import pandas as pd
import healpy as hp
import numpy as np
from DS_healpix_fragmentation import radec2pix
df = [pd.read_csv(clusters_file) for clusters_file in clusters_files]
df = pd.concat(df, ignore_index=True)
pix2 = radec2pix(df['RA'], df['DEC'], 2)
df = df[np.in1d(pix2, big_pixels)]
df.index = np.arange(df.shape[0])
small_pixels = set()
sc = SkyCoord(ra=np.array(df['RA']) * u.degree,
dec=np.array(df['DEC']) * u.degree,
frame='icrs')
vecs = hp.ang2vec(theta=sc.galactic.l.degree,
phi=sc.galactic.b.degree,
lonlat=True)
for i in range(df.shape[0]):
small_pixels = small_pixels.union(
hp.query_disc(nside, vecs[i], np.radians(min_rad), nest=True))
small_pixels = np.array(list(small_pixels))
return small_pixels, df
def query_disc(self, lon, lat, radius, fact=4): """ Find pixels that overlap with discs centered at (lon,lat) Inputs ------ lon : float, ndarray longitude (degr) lat : float, ndarray latitude (degr) radius : float, ndarray radius of disk (degrees) fact : float supersampling factor to find overlapping pixels (see healpy.query_disc doc) Returns ------ ndarray : pixel indices """ phi = np.array(lon) * self.deg2rad theta = (90 - np.array(lat)) * self.deg2rad vec = healpy.ang2vec(theta, phi) pix = healpy.query_disc(self.nside, vec, radius * self.deg2rad, inclusive=True, fact=fact, nest=self.nest) return pix
def calculate_efficiency(hpx, d0, s0, s1, nfields, fieldop):
'''This function determines the score of a given set of tiling parameters for the optimization.
The figure of merit is the sum value of the for the <nfields> higher fields of the tiling'''
nside = hp.npix2nside(len(hpx))
keptfields = build_fields(hpx, d0, s0, s1, nfields, fieldop)
# totaldots = np.sum(keptfields["prob"])
total = 0
prob_integral = []
for indec in range(0, len(keptfields)):
# cornerra, cornerdec = getcorners(keptfields["ra"][indec],keptfields["dec"][indec],field=4.2)
# xyz = hp.ang2vec(checktheta(dectotheta(cornerdec)),checkphi(ratophi(cornerdec)))
# hp.query_polygon(nside,xyz)
xyz = hp.ang2vec(dectotheta(keptfields["dec"][indec]),
ratophi(keptfields["ra"][indec]))
ipix_disc = hp.query_disc(nside, xyz, np.deg2rad(
fieldop)) #here radius seems to be a diameter instead * (sq2+1)/4)
totdisc = hpx[ipix_disc].sum()
prob_integral.append(totdisc)
total += totdisc
# efficiency = total / nfields
#print (total)
return total, prob_integral
def handling_exception(params, constraints):
NSIDEmax = params['NSIDE max']
vec = hp.ang2vec(params["ang"][0], params["ang"][1], lonlat=True)
pixels = hp.query_disc(NSIDEmax,
vec,
np.radians(params['r'] / 3600.),
inclusive=True)
subjobs = mastcasjobs.MastCasJobs(context="PanSTARRS_DR2")
for pixel in pixels:
ang, r = qr.parameters(NSIDEmax, pixel)
subquery = sub_query_string(ang[0], ang[1], r)
accept = True
while accept:
try:
subtab = subjobs.quick(subquery,
task_name="python cone search")
accept = False
except Exception:
from time import sleep
sleep(60)
pass
subtab = qr.fixcolnames(ascii.read(subtab))
subtab = qr.query_constraints(subtab, constraints)
if pixel == pixels[0]:
table = subtab
else:
table = vstack([table, subtab])
return table
def gen_map_disc(radec_cen, rad, nside):
'''Generates a Healpix map with the only non-zero values in the
pixels inside the input disc.
Parameters
----------
radec_cen : array-like with shape (2,)
The center ra,dec of the disc in degrees
rad : float
The radius of the disc in degrees
nside : int
The nside of the output Healpix map
Returns
-------
hpx_map : array-like
A Healpix map with non-zero values inside the disc
'''
theta = np.pi/2 - np.radians(radec_cen[1])
phi = np.radians(radec_cen[0])
vec = H.ang2vec(theta, phi)
ipix = H.query_disc(nside, vec, np.radians(rad))
hpx_map = np.zeros(H.nside2npix(nside))
hpx_map[ipix] = 1.0
return hpx_map
def _get_box_pix(self, nside, ra_cent, dec_cent, width, height):
"""
Get healpix pixels overlapping a box.
Parameters
----------
nside : `int`
ra_cent : `float`
dec_cent : `float`
width : `float`
height : `float`
Returns
-------
pixels : `np.ndarray`
"""
wid = width / np.cos(np.deg2rad(dec_cent))
vertices = hp.ang2vec(np.array([ra_cent - wid/2.,
ra_cent - wid/2.,
ra_cent + wid/2.,
ra_cent + wid/2.]),
np.array([dec_cent - height/2.,
dec_cent + height/2.,
dec_cent + height/2.,
dec_cent - height/2.]),
lonlat=True)
return hp.query_polygon(nside, vertices, nest=True)
def beamfunc(pixels, fwhm_arcmin, amplitude, ecc, incl_rad, theta_beam, phi_beam):
pixel_theta, pixel_phi = pixels
sigma_u = np.deg2rad(fwhm_arcmin / 60.0) / (2 * np.sqrt(2 * np.log(2)))
sigma_v = sigma_u * (1 - ecc)
vec = healpy.ang2vec(pixel_theta, pixel_phi)
Ry = [
[np.cos(theta_beam), 0, np.sin(theta_beam)],
[0, 1, 0],
[-np.sin(theta_beam), 0, np.cos(theta_beam)],
]
Rz = [
[np.cos(phi_beam), -np.sin(phi_beam), 0],
[np.sin(phi_beam), np.cos(phi_beam), 0],
[0, 0, 1],
]
RotMat = np.dot(Rz, Ry)
vecrot = np.dot(vec, RotMat)
beam = np.zeros(len(vecrot))
x = np.transpose(vecrot)[0]
y = np.transpose(vecrot)[1]
z = np.transpose(vecrot)[2]
u = np.cos(incl_rad) * x + np.sin(incl_rad) * y
v = -np.sin(incl_rad) * x + np.cos(incl_rad) * y
beam = amplitude * np.exp(
-((u / (np.sqrt(2) * sigma_u)) ** 2 + (v / (np.sqrt(2) * sigma_v)) ** 2)
)
beam[z < 0] = 0.0
return beam
def gen_map_disc(radec_cen, rad, nside):
'''Generates a Healpix map with the only non-zero values in the
pixels inside the input disc.
Parameters
----------
radec_cen : array-like with shape (2,)
The center ra,dec of the disc in degrees
rad : float
The radius of the disc in degrees
nside : int
The nside of the output Healpix map
Returns
-------
hpx_map : array-like
A Healpix map with non-zero values inside the disc
'''
theta = np.pi / 2 - np.radians(radec_cen[1])
phi = np.radians(radec_cen[0])
vec = H.ang2vec(theta, phi)
ipix = H.query_disc(nside, vec, np.radians(rad))
hpx_map = np.zeros(H.nside2npix(nside))
hpx_map[ipix] = 1.0
return hpx_map
def corner_xyz(self):
if len(self.__corner_xyz) == 0:
# coord_vertices = self.corner_coords
# xyz_vertices = []
# for c in coord_vertices:
# theta = 0.5 * np.pi - np.deg2rad(c.dec.degree)
# phi = np.deg2rad(c.ra.degree)
# xyz_vertices.append(hp.ang2vec(theta, phi))
# self.__corner_xyz = np.asarray(xyz_vertices)
coord_vertices = self.corner_coords
xyz_vertices = []
for c in coord_vertices:
theta = 0.5 * np.pi - np.deg2rad(c[1])
phi = np.deg2rad(c[0])
xyz_vertices.append(hp.ang2vec(theta, phi))
self.__corner_xyz = np.asarray(xyz_vertices)
return self.__corner_xyz
def compute_circle(N, t, p, r, r_in):
# Compute query (outer circle) and query_in (inner circle)
query = hp.query_disc(nside=N,
vec=hp.ang2vec(np.pi / 2 - t, p),
radius=r,
inclusive=False,
nest=False)
query_in = hp.query_disc(nside=N,
vec=hp.ang2vec(np.pi / 2 - t, p),
radius=r_in,
inclusive=False,
nest=False)
# Inverse intersection of query (outer circle) and query_in (inner circle)
inner = minus(query, query_in)
return inner
def get_pixels(centr_ra, centr_decl, fov_radius):
"""
Get a list of HEALPIX zones that contain a given image.
"""
vector = healpy.ang2vec(radians(90.0 - centr_decl),
radians(centr_ra))
pixels = healpy.query_disc(32, vector, radians(fov_radius),
inclusive=True, nest=True)
return str(pixels.tolist())[1:-1]
def gaussian_on_a_sphere(mean_th, mean_phi, sigma):
"""
This function returns a 2D normal pdf on a discretized healpy grid.
To chose the function values correctly in spherical coordinates,
the true angular distances to the mean are used.
Pixels farther away from the mean than clip * sigma are clipped
because the normal distribution falls quickly to zero. The error
made by this can be easily estimated and a discussion can be found
in [arXiv:1005.1929](https://arxiv.org/abs/1005.1929v2).
Parameters
----------
mean_th : float
Position of the mean in healpy coordinate `theta`. `theta` is in
[0, pi] going from north to south pole.
mean_phi : float
Position of the mean in healpy coordinate `phi`. `phi` is in [0, 2pi]
and is equivalent to the azimuth angle.
sigma : float
Standard deviation of the 2D normal distribution. Only symmetric
normal pdfs are used here.
Returns
-------
kernel : array
Values of the 2D normal distribution at the selected pixels. If clip
False, kernel is a valid healpy map with resolution NSIDE.
keep_idx : array
Pixel indices that are kept from the full healpy map after clipping.
If clip is False this is a sorted integer array with values
[0, 1, ..., NPIX-1] with NPIX tha number of pixels in the full healpy
map with resolution NSIDE.
"""
# Clip unneccessary pixels, just keep clip*sigma (radians)
# around the mean. Using inlusive=True to make sure at least
# one pixel gets returned.
# Always returns a list, so no manual np.array() required.
keep_idx = hp.query_disc(
self._nside, hp.ang2vec(mean_th, mean_phi),
self._clip * sigma, inclusive=True)
# Create only the needed the pixel healpy coordinates
th, phi = hp.pix2ang(self._nside, keep_idx)
sinTh = np.sin(th)
# For each pixel get the distance to (mean_th, mean_phi) direction
dist = angdist(mean_phi, mean_th, phi, sinTh)
# Get the 2D gaussian values at those distances -> kernel function
# Because the kernel is radial symmetric we use a simplified
# version -> 1D gaussian, properly normed
sigma2 = 2 * sigma**2
kernel = np.exp(-dist**2 / sigma2) / (np.pi * sigma2)
return kernel, keep_idx
def sky2vec(cls,sky):
"""
Convert sky positions in to 3d-vectors
:param sky: [(ra,dec), ...]
:return: [(x,y,z), ...]
"""
theta_phi = cls.sky2ang(sky)
theta,phi = map(np.array,zip(*theta_phi))
vec=hp.ang2vec(theta,phi)
return vec
def compute_healpix_vec(self): """ convert galactic latitude and longitude positions in :mod:`healpy` mapping quantinties: position vectors. """ rtod=180./np.pi b_h=np.pi/2. - self.lat l_h=self.long vec=hp.ang2vec(b_h,l_h) return vec
def from_point_sources(self,d,t):
# if p.circular_pol == True:
# I,Q,U,V = sky_list
# else:
# I,Q,U = sky_list
print "t is " + str(t)
total_angle = float(self.nhours * 15) # degrees
offset_angle = float(self.hour_offset * 15) # degrees
zl_ra = ((float(t) / float(self.ntime)) * np.radians(total_angle) + np.radians(offset_angle)) % (2.*np.pi)# radians
RotAxis = np.array([0.,0.,1.])
RotAngle = -zl_ra # basicly Hour Angle with t=0
# R_t = irf.rotation_matrix(RotAxis, RotAngle)
# s0 = hp.ang2vec(src_cza,src_ra)
# sf = np.einsum('...ab,...b->...b', R_t, s0)
self.src_ra_f = self.src_ra + RotAngle
sf = hp.ang2vec(self.src_cza, self.src_ra_f)
# s0 = hp.ang2vec(src_cza,src_ra)
# hpR_t = hp.Rotator(rot=[RotAngle])
# sf = np.array(hpR_t(s0[:,0],s0[:,1],s0[:,2])).T
if len(sf.shape) > 1:
inds_f = hp.vec2pix(self.nside, sf[:,0],sf[:,1],sf[:,2])
else:
inds_f = hp.vec2pix(self.nside, sf[0],sf[1],sf[2])
It = self.I #XXX ??
ijones_t = self.ijones[:,inds_f,:,:]
ijonesH_t = self.ijonesH[:,inds_f,:,:]
Kt = self.K[:,inds_f]
if self.unpolarized == True:
Ut = np.zeros((self.nfreq,It.shape[-1]))
sky_t = np.array([
[It, Ut],
[Ut, It]]).transpose(2,3,0,1) # sky_t.shape = (p.nfreq, p.npix, 2, 2)
else:
Ut = self.U
Qt = self.Q
sky_t = np.array([
[It + Qt, Ut],
[Ut, It - Qt]]).transpose(2,3,0,1) # sky_t.shape = (p.nfreq, p.npix, 2, 2)
irnf.instrRIME_integral(ijones_t, sky_t, ijonesH_t, Kt, self.Vis[d,t,:,:,:].squeeze())
def probability_inside_circle(self, ra, dec, radius):
"""Return the probability inside a circle."""
prob = hp.read_map(self.skymap, verbose=False)
theta = 0.5 * np.pi - np.deg2rad(dec)
phi = np.deg2rad(ra)
radius = np.deg2rad(radius)
xyz = hp.ang2vec(theta, phi)
ipix_disc = hp.query_disc(self.nside, xyz, radius)
probability_inside_disc = prob[ipix_disc].sum()
return "%.1e" % probability_inside_disc
def probability_inside_box(self, infile, ra_vertices, dec_vertices):
"""Return the probability inside a polygon."""
prob = hp.read_map(self.skymap, verbose=False)
theta = 0.5 * np.pi - np.deg2rad(dec_vertices)
phi = np.deg2rad(ra_vertices)
xyz = hp.ang2vec(theta, phi)
ipix_poly = hp.query_polygon(self.nside, xyz)
probability_inside_polygon = prob[ipix_poly].sum()
return probability_inside_polygon
def hp_interpolator(map_, el, az, n_pix=4):
NSide = hp.pixelfunc.get_nside(map_)
direction = np.array([np.pi/2.-el.to_rad(),az.to_rad()])
steplength = hp.pixelfunc.max_pixrad(NSide)
for i, r in enumerate(np.arange(steplength, np.pi, steplength)):
pixels = np.array(hp.query_disc(NSide,hp.ang2vec(direction[0], direction[1]), r))
filled = np.where(map_[pixels] > -1.)[0]
l = len(filled)
if l >= n_pix:
# print(i, l)
filled_pixel = pixels[filled]
filled_pixel_directions = hp.pix2vec(NSide, filled_pixel)
angular_distance = hp.rotator.angdist(direction, filled_pixel_directions)
if angular_distance.min() == 0.: # do we really want this?
return map_[filled_pixel[angular_distance.argmin()]]
return np.average(map_[filled_pixel], weights=np.power(1./angular_distance, 2))
def addVec(df, raCol='ditheredRA', decCol='ditheredDec'):
"""
Add a column of vectors to the dataFrame df
Parameters
----------
df : `pd.DataFrame`
dataframe with two columns raCol and decCol having ra,
dec in radians.
raCol : string, optional, defaults to 'ditheredRA'
column name for the column which has ra values
decCol : string, optional, defaults to 'ditheredDec'
column name for the column which has dec values
"""
thetas = - df[decCol] + np.pi /2.
phis = df[raCol]
df['vec'] = list(hp.ang2vec(thetas, phis))
def wmap5_parameters():
"""WMAP5 solar system dipole parameters,
from: http://arxiv.org/abs/0803.0732"""
# 369.0 +- .9 Km/s
SOLSYSSPEED = 369e3
#l.critical('USING 371')
#SOLSYSSPEED = 371e3
## direction in galactic coordinates
##(d, l, b) = (3.355 +- 0.008 mK,263.99 +- 0.14,48.26deg +- 0.03)
SOLSYSDIR_GAL_THETA = np.deg2rad( 90 - 48.26 )
SOLSYSDIR_GAL_PHI = np.deg2rad( 263.99 )
SOLSYSSPEED_GAL_U = healpy.ang2vec(SOLSYSDIR_GAL_THETA,SOLSYSDIR_GAL_PHI)
#SOLSYSSPEED_GAL_V = SOLSYSSPEED * SOLSYSSPEED_GAL_U
SOLSYSSPEED_ECL_U = gal2ecl(SOLSYSSPEED_GAL_U)
SOLSYSDIR_ECL_THETA, SOLSYSDIR_ECL_PHI = healpy.vec2ang(SOLSYSSPEED_ECL_U)
########## /WMAP5
return SOLSYSSPEED, SOLSYSDIR_ECL_THETA, SOLSYSDIR_ECL_PHI
def randsphere(num, system='eq', **kw):
"""
Generate random points on the sphere
parameters
----------
num: integer
The number of randoms to generate
system: string
'eq' for equatorial ra,dec in degrees
ra in [0,360]
dec in [-90,90]
'ang' for angular theta,phi in radians
theta in [0,pi]
phi in [0,2*pi]
'vec' for x,y,z on the unit sphere
**kw:
ra_range=, dec_range= to limit range of randoms
in ra,dec system
theta_range=, phi_range= to limit range of randoms
in theta,phi system
output
------
ra,dec: tuple of arrays
the random points
"""
if system=='eq':
return randsphere_eq(num, **kw)
elif system=='ang':
return randsphere_ang(num, **kw)
elif system=='vec':
import healpy
theta,phi=randsphere_ang(num)
return healpy.ang2vec(theta,phi)
else:
raise ValueError("bad system: '%s'" % sytem)
def test_query_disc(self):
if not DO_HEALPY:
# not possible on Windows
return
with NumpyRNGContext(12345):
nside = 1024
hpx = Healpix(nside)
disc_size = np.radians(0.5 / 60.)
thetas = np.random.uniform(0, np.pi, 10)
phis = np.random.uniform(0, 2 * np.pi, 10)
for theta, phi in zip(thetas, phis):
_vec = hp.ang2vec(theta, phi)
hp_disc = hp.query_disc(nside, _vec, disc_size)
assert_equal(
hp_disc,
hpx.query_disc(theta, phi, disc_size)
)
def create_healpix_tsz_template(rm, n_theta_core=5.,
weight_min=0.1, beta=0.7):
fill_free_electron_parameters(rm)
# only add those above some weight threshold
wh=np.where(rm['weight']>weight_min)[0]
print 'USING %i CLUSTERS WITH WEIGHT>%0.2f'%(len(wh), weight_min)
for k in rm.keys(): rm[k]=rm[k][wh]
npix = hp.nside2npix(nside)
template = np.zeros(npix)
ncl = len(rm['ra'])
vec_hpix = hp.ang2vec(rm['th_gal'], rm['phi_gal'])
ind_hpix = hp.ang2pix(nside, rm['th_gal'], rm['phi_gal'], nest=False)
for i in range(ncl):
this_vec = vec_hpix[i,:]
this_theta_c = rm['theta_c'][i] #radians
this_ind = ind_hpix[i]
ind_nearby = hp.query_disc(nside, this_vec, n_theta_core*this_theta_c, nest=False)
vec_nearby = hp.pix2vec(nside, ind_nearby, nest=False)
theta_nearby = hp.rotator.angdist(this_vec, vec_nearby)
values_nearby = (1.+(theta_nearby/this_theta_c)**2.)**((1.-3.*beta)/2.)
#values_nearby *= rm['t0'][i]
template[ind_nearby] += values_nearby
return template
def gen_map_polygon(vertices, nside):
'''Generates a Healpix map with the only non-zero values in the pixels
inside the input polygon
Parameters
----------
vertices : array-like with shape (n,2) or (2,n)
The lon,lat vertices of the polygon in degrees. n >= 3
nside : int
The nside of the output Healpix map
Returns
-------
hpx_map : array-like
A Healpix map with non-zero values inside the polygon
'''
vertices = np.array(vertices)
if vertices.shape[1] != 2:
vertices = np.transpose(vertices)
if vertices.shape[1] != 2:
raise ValueError("Need a n x 2 or 2 x n input vertices array")
thetas = np.pi/2 - np.radians(vertices[:, 1])
phis = np.radians(vertices[:, 0])
vecs = H.ang2vec(thetas, phis)
ipix = H.query_polygon(nside, vecs)
hpx_map = np.zeros(H.nside2npix(nside))
hpx_map[ipix] = 1.0
return hpx_map
def angToVec(lon, lat):
theta = lat2theta(lat)
phi = lon2phi(lon)
vec = hp.ang2vec(theta, phi)
return vec
#onestep_fit_flux( write_png=False, inc_disc=False, inc_haze=True, inc_bubbles=True, raw_sync=False, # mfile='ebvh_mask_top-half_disc_40_ns0128.fits', model_list=['12132h'], plot_moll=False ) sys.exit() dir = '/global/scratch2/sd/dpietrob/DM-haze/data/maps/ns0128/' nside = 128 snside = '0128' # ==================================================================== # Make bubbles template as two discs for the haze/gamma-rays counterpart # Not used: back to Gaussian for consistency and much difference after masking if False: dirn = hp.ang2vec( (90.-22.5)/180.*np.pi, 0. ) dn = make_disc( nside, 20., direction=dirn) dirs = hp.ang2vec( (90.+22.5)/180.*np.pi, 0. ) ds = make_disc( nside, 20., direction=dirs) bubbles = (ds + dn ) * 0.01 bubbles = hp.smoothing( bubbles, fwhm=10./180.*np.pi) hp.mollview( bubbles ) hp.write_map( dir+'fermi_bubbles_ns'+snside+'.fits', bubbles) pl.show() # ==================================================================== # Residual maps for the best fit: Tab3col1 first and 11113 if False: from myclass import * from mcmc import *
############################################################################# nside = 256 lmax = 3*nside-1 #### Mask maxang = 5. center = [0.,0.] veccenter = hp.ang2vec(pi/2-np.radians(center[1]), np.radians(center[0])) vecpix = hp.pix2vec(nside,np.arange(12*nside**2)) cosang = np.dot(veccenter,vecpix) maskok = np.degrees(np.arccos(cosang)) < maxang maskmap=np.zeros(12*nside**2) maskmap[maskok]=1 wl = hp.anafast(maskmap,regression=False) wl = wl[0:lmax+1] #### Do a Monte-Carlo nbmc = 1000 allcls = np.zeros((nbmc, lmax+1)) for i in np.arange(nbmc): print(i)
opts.input.name, distances=True)
npix = len(prob)
nside = hp.npix2nside(npix)
progress.update(-1, 'Preparing projection')
if opts.align_to is None:
prob2, mu2, sigma2 = prob, mu, sigma
else:
(prob2, mu2, sigma2, _), _ = fits.read_sky_map(
opts.align_to.name, distances=True)
R = np.ascontiguousarray(principal_axes(prob2, mu2, sigma2))
if opts.chain:
chain = np.recfromtxt(opts.chain, names=True)
chain = np.dot(R.T, (hp.ang2vec(
0.5 * np.pi - chain['dec'], chain['ra'])
* np.atleast_2d(chain['dist']).T).T)
fig = plt.figure(frameon=False)
n = 1 if opts.projection else 2
gs = gridspec.GridSpec(
n, n, left=0.01, right=0.99, bottom=0.01, top=0.99,
wspace=0.05, hspace=0.05)
imgwidth = int(opts.dpi * opts.figure_width / n)
s = np.linspace(-opts.max_distance, opts.max_distance, imgwidth)
xx, yy = np.meshgrid(s, s)
dtheta = 0.5 * np.pi / nside / 4
# Color palette for markers
colors = seaborn.color_palette(n_colors=len(opts.radecdist) + 1)
maprms_dust = FitsArray('small2_maprms_dust.fits')
maprms_all = FitsArray('small2_maprms_all.fits')
racenter = 0
deccenter = -57.0
deltatheta = 15.
deltatheta2 = 20.
deltatheta3 = 25.
racirc, deccirc = circ_around_radec([racenter, deccenter], deltatheta)
racirc2, deccirc2 = circ_around_radec([racenter, deccenter], deltatheta2)
racirc3, deccirc3 = circ_around_radec([racenter, deccenter], deltatheta3)
# Initial dust map
clf()
l,b = equ2gal(racenter, deccenter)
vv = hp.ang2vec(np.radians(90-b), np.radians(l))
pixels = hp.query_disc(hp.npix2nside(len(mapdust)), vv, np.radians(15))
mm = np.min(mapdust[pixels])
#hp.gnomview(mapdust-mm, coord='GE', title='Planck Dust Map', rot=[racenter,deccenter], fig=1, reso=25,min=0,max=1e-4)
hp.mollview(mapdust-mm, coord='GE', title='Planck Dust Map', rot=[racenter,deccenter], fig=2, min=0,max=1e-4)
hp.projplot(pi/2-np.radians(deccenter), np.radians(racenter), 'kx')
hp.projplot(pi/2-np.radians(deccirc), np.radians(racirc), 'k')
hp.projplot(pi/2-np.radians(deccirc2), np.radians(racirc2), 'k')
hp.projplot(pi/2-np.radians(deccirc3), np.radians(racirc3), 'k')
hp.projplot(pi/2-np.radians(-45), np.radians(4*15+40/60+12/3600), 'ko')
hp.projplot(pi/2-np.radians(-30/60), np.radians(11*15+453/60+0/3600), 'ko')
hp.projplot(pi/2 - np.radians(-32-48/60), np.radians(23*15+1/60+48/3600), 'ko')
bicepalpha = np.ravel([np.zeros(10)-60, np.linspace(-60, 60, 10), np.zeros(10)+60, np.linspace(60, -60, 10)])
bicepdelta = np.ravel([np.linspace(-70, -45, 10),np.zeros(10)-45, np.linspace(-45, -70, 10),np.zeros(10)-70])
hp.projplot(pi/2-np.radians(bicepdelta), np.radians(bicepalpha), 'k--')
def compute_SOLSYSSPEED_V(norm, theta, phi):
return norm * healpy.ang2vec(theta,phi)
width = 10
poles_margin = 0.1
halfwidth = width/2
for theta in range(halfwidth, 180, width):
for phi in range(halfwidth, 360, width):
filename = "plancktest/submaps/%03d_%03d_%03d" % (freq, theta, phi)
print filename
vertex_angles = np.radians(
[[theta-halfwidth, phi-halfwidth],
[theta+halfwidth, phi-halfwidth],
[theta+halfwidth, phi+halfwidth],
[theta-halfwidth, phi+halfwidth]])
if theta - halfwidth == 0:
vertex_angles[0,0] = np.radians(poles_margin)
vertex_angles[3,0] = np.radians(poles_margin)
if theta + halfwidth == 180:
vertex_angles[1,0] = np.radians(180-poles_margin)
vertex_angles[2,0] = np.radians(180-poles_margin)
pix=hp.query_polygon(nside, hp.ang2vec(vertex_angles[:,0], vertex_angles[:,1]), np.radians(13))
submap = np.zeros(len(pix), dtype=[("pix",np.int64), ("temp", np.float64)])
submap["pix"] = pix
submap["temp"] = m[pix]
np.save(filename, submap)
from lalinference import plot
from wcsaxes import WCS
import numpy as np
from matplotlib import pyplot as plt
import healpy as hp
from matplotlib import collections
from matplotlib import patches
from matplotlib import ticker
ra0 = 130
dec0 = -10
phi0 = np.deg2rad(ra0)
theta0 = 0.5 * np.pi - np.deg2rad(dec0)
# ra0 = 137.8
# dec0 = -39.9
xyz0 = hp.ang2vec(0.5*np.pi - np.deg2rad(dec0), np.deg2rad(ra0))
prob, meta = fits.read_sky_map('../cbc/emfollowup/papers/first2years/data/2015/compare/18951/bayestar.fits.gz', nest=True)
def get_vertices(m):
m = np.copy(m)
top_npix = len(m)
top_nside = hp.npix2nside(top_npix)
top_order = int(np.log2(top_nside))
for order in range(top_order + 1):
nside = 1 << order
npix = hp.nside2npix(nside)
stride = 1 << (2 * (top_order - order))
keep = (hp.pix2vec(nside, np.arange(npix), nest=True) * np.expand_dims(xyz0, 1)).sum(0) >= np.cos(np.deg2rad(30))
if order < top_order:
mm = m.reshape((-1, stride))
def query_disc(self, coord1, coord2, radius, system='eq', mine=False, **kw):
"""
get pixels that are contained within or intersect the disc
hpix.query_disc(ra,dec,radius_degrees)
hpix.query_disc(theta,phi,radius_radians,system='ang')
parameters
----------
coord1: scalar
If system=='eq' this is ra degrees
If system=='ang' this is theta radians
coord2: scalar
If system=='eq' this is dec degrees
If system=='ang' this is phi radians
radius: scalar
radius of disc
If system=='eq' this is in degrees
If system=='ang' this is in radians
system: string
'eq' for equatorial ra,dec in degrees
'ang' for angular theta,phi in radians
default 'eq'
keywords for healpy.query_disc
inclusive: bool, optional
If False, return the exact set of pixels whose pixel centers lie
within the disk; if True, return all pixels that overlap with the
disk, and maybe a few more. Default: False
fact : int, optional
Only used when inclusive=True. The overlapping test will be done at
the resolution fact*nside. For NESTED ordering, fact must be a
power of 2, else it can be any positive integer. Default: 4.
nest: bool, optional
if True, assume NESTED pixel ordering, otherwise, RING pixel
ordering
buff: int array, optional
if provided, this numpy array is used to contain the return values
and must be at least long enough to do so
returns
-------
pixnums: array
Array of pixel numbers that are contained or intersect the disc
examples
--------
from hpix_util import HealPix
hpix = HealPix("ring", 4096)
ra=200.0
dec=0.0
radius=1.0
pixnums=hpix.query_disc(ra, dec, radius)
"""
import healpy
if mine:
if system=='eq':
sysnum=coords.SYSTEM_EQ
elif system=='ang':
sysnum=SYSTEM_ANG
else:
raise ValueError("system should be 'eq' or 'ang'")
inclusive=kw.get('inclusive',False)
incnum = 1 if inclusive else 0
pixnums=self._query_disc(coord1,coord2,radius,sysnum,incnum)
else:
if system=='eq':
vec=coords.eq2vec(coord1, coord2)
rad_send=numpy.deg2rad(radius)
elif system=='ang':
vec=healpy.ang2vec(coord1, coord2)
rad_send=radius
else:
raise ValueError("system should be 'eq' or 'ang'")
kw['nest'] = self.nested
pixnums=healpy.query_disc(self.nside, vec, rad_send, **kw)
return pixnums
print 'split :',split
vis_root=np.empty([BL_NUM/split,4],dtype=np.complex128)
UV_root=np.empty([BL_NUM,3],dtype=np.float)
#==========
###############calculating UVW####################
UV_bl=Vis.BL_radec
mpi_UV_bl_index=np.array_split(np.arange(len(UV_bl)),size)
UV = []
for i in mpi_UV_bl_index[rank]:
UV.append(Vis.Get_UV_radec(source[0], source[1], bl=Vis.BL_radec[i])[0])
UV = np.array(UV,dtype=np.float)
##############################mpi util##########################################
theta, phi = hp.pix2ang(map_nside, np.arange(map_nside**2 * 12))
thphi=np.c_[theta,phi]
sigma=hp.ang2vec(theta,phi)
BL_bl_split=np.array_split(BL_bl,split)
for i in np.arange(len(BL_bl_split)):
mpi_bl=np.array_split(BL_bl_split[i],size)
vis=np.dot(np.exp(-2*np.pi/lam*1j*np.dot(mpi_bl[rank],sigma.T)),map)
vis_result=np.c_[mpi_bl[rank],vis/(map_nside**2*12.)]
comm.Gather(vis_result,vis_root,root=0)
if rank==0:
# print mpi_bl[rank]
# print i,mpi_bl[rank]
# print vis_result.shape
print '%d...'%(i+1)
if i==0:
vis_all_root=np.vstack(vis_root)
else:
def main():
# parse command-line arguments
parser = argparse.ArgumentParser(
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('--verbose', action='store_true',
help='more verbose output')
parser.add_argument('--order', type=int, default=5,
help='order of healpix hierachical tesselation')
parser.add_argument('--input', type=str, default=None,
help='input file name')
args = parser.parse_args()
nside = 2**args.order
npix = hp.nside2npix(nside)
m = np.arange(npix)
print 'HEALPix map info:'
print '\tn pixels:', npix
print '\tresolution:', hp.nside2resol(nside)
print '\tpixel area:', hp.nside2pixarea(nside)
print '\tmax distance from center:', hp.max_pixrad(nside)
print 'xi binning info:'
transerve_comoving_scale = 3820.43 # Mpc, at z = 2.1
maxang = 200/transerve_comoving_scale
print '\tmax ang scale:', maxang
print 'test stuff:'
test_theta = 1.58853
test_phi = 0.154
testvec = hp.ang2vec(test_theta, test_phi)
testmax = 0.03
print '\ttheta, phi:', test_theta, test_phi
print '\t3d vec:', testvec
print '\tpixel index:', hp.ang2pix(nside, test_theta, test_phi)
print '\tclosest neighbors:', hp.pixelfunc.get_all_neighbours(nside, test_theta, test_phi)
print '\tneighbors within max ang scale:', hp.query_disc(nside, testvec, testmax, inclusive=True)
# include_pix = hp.query_disc(nside, hp.ang2vec(5./6.*np.pi/2, 7./6.*np.pi), 2*maxang, inclusive=True)
# print '\tn pixels for test:', len(include_pix)
qso_pos = np.loadtxt(args.input)
print 'Shape of input data: ', qso_pos.shape
m = {}
qso_pos[:,0] = np.deg2rad(qso_pos[:,0])
qso_pos[:,1] = np.deg2rad(90-qso_pos[:,1])
class Quasar():
def __init__(self, theta, phi, z):
self.theta = theta
self.phi = phi
self.z = z
# fake this for now
self.wave = np.log10((1+z)*(1100+np.arange(10)))
quasars = []
for i, qso in enumerate(qso_pos):
ra, dec, z = qso
phi = ra
theta = dec
q = Quasar(theta, phi, z)
map_index = hp.ang2pix(nside, theta, phi)
# if map_index not in include_pix:
# continue
#q.map_index = map_index
if map_index in m:
m[map_index].append(i)
else:
m[map_index] = [i]
q.i = i
quasars.append(q)
print 'Number of pixels with at least one quasar: %d' % (len(m.keys()))
print 'Average number of quasars per pixel: %f' % (float(len(quasars))/len(m.keys()))
nearby_pixels_sum = 0
nearby_quasars_sum = 0
for q_index, q in enumerate(quasars):
nearby_pixels = hp.query_disc(nside, hp.ang2vec(q.theta, q.phi), maxang, inclusive=True)
nearby_pixels_sum += len(nearby_pixels)
for nearby_pixel in nearby_pixels:
if nearby_pixel in m:
nearby_quasars = m[nearby_pixel]
nearby_quasars_sum += len(nearby_quasars)
p_phi = qso_pos[nearby_quasars,0]
p_theta = qso_pos[nearby_quasars,1]
angdist = np.sin(q.theta)*np.sin(p_theta) + np.cos(q.theta)*np.cos(p_theta)*(np.cos(q.phi)*np.cos(p_phi) + np.sin(q.phi)*np.sin(p_phi))
for p_i, p_index in enumerate(nearby_quasars):
theta, phi, z = qso_pos[p_index]
p = Quasar(theta, phi, z)
loglamdist = abs(np.subtract.outer(q.wave, p.wave))
p_angdist = angdist[p_i]
#losdist = abs(np.log(q_wave/p.wave))
print 'Average number of pixels within 200 Mpc of a quasar: %f' % (float(nearby_pixels_sum)/len(quasars))
print 'Average number of quasars within nearby pixels of a quasar: %f' % (float(nearby_quasars_sum)/len(quasars))
mask = np.ones(npix).astype(np.bool)
mask[m.keys()] = False
mm = hp.ma(np.zeros(npix))
mm.mask = mask
mm[~mm.mask] = [len(m[i]) for i in m.keys()]
hp.mollview(mm.filled(), rot=(270,0,0), flip='geo')
plt.show()
