def __init__(self,
mjd_start,
sun_RA_start=0,
nside=32,
filters={
'u': 0,
'g': 1,
'r': 2,
'i': 3,
'z': 4,
'y': 5
},
period=365.25,
step_func=None):
self.period = period
self.nside = nside
if step_func is None:
step_func = Step_line()
self.step_func = step_func
self.mjd_start = mjd_start
self.sun_RA_start = sun_RA_start
self.npix = hp.nside2npix(nside)
self.filters = filters
self.ra, self.dec = _hpid2RaDec(self.nside, np.arange(self.npix))
# Set the phase of each healpixel. If RA to sun is zero, we are at phase np.pi/2.
self.phase = (-self.ra + self.sun_RA_start + np.pi / 2) % (2. * np.pi)
self.phase = self.phase * (self.period / 2. / np.pi)
# Empty footprints to start
self.out_dtype = list(zip(filters, [float] * len(filters)))
self.footprints = np.zeros((len(filters), self.npix), dtype=float)
self.estimate = np.zeros((len(filters), self.npix), dtype=float)
self.current_footprints = np.zeros((len(filters), self.npix),
dtype=float)
self.zero = self.step_func(0., self.phase)
self.mjd_current = None
def __init__(self, surveys, nside=default_nside, camera='LSST'):
"""
Parameters
----------
surveys : list of survey objects
A list of surveys to consider. If multiple surveys retrurn the same highest
reward value, the survey at the earliest position in the list will be selected.
nside : int
A HEALpix nside value.
camera : str ('LSST')
Which camera to use for computing overlapping HEALpixels for an observation.
Can be 'LSST' or 'comcam'
"""
if nside is None:
nside = set_default_nside()
# initialize a queue of observations to request
self.queue = []
self.surveys = surveys
self.nside = nside
hpid = np.arange(hp.nside2npix(nside))
self.ra_grid_rad, self.dec_grid_rad = _hpid2RaDec(nside, hpid)
self.conditions = None
# Should just make camera a class that takes a pointing and returns healpix indices
if camera == 'LSST':
self.pointing2hpindx = hp_in_lsst_fov(nside=nside)
elif camera == 'comcam':
self.pointing2hpindx = hp_in_comcam_fov(nside=nside)
else:
raise ValueError('camera %s not implamented' % camera)
def test_rad(self):
nside = 64
hpids = np.arange(hp.nside2npix(nside))
ra, dec = utils._hpid2RaDec(nside, hpids)
mjd = 59852.
obs = utils.ObservationMetaData(mjd=mjd)
alt1, az1, pa1 = utils._altAzPaFromRaDec(ra, dec, obs)
alt2, az2 = utils._approx_RaDec2AltAz(ra, dec, obs.site.latitude_rad,
obs.site.longitude_rad, mjd)
# Check that the fast is similar to the more precice transform
tol = np.radians(2)
tol_mean = np.radians(1.)
separations = utils._angularSeparation(az1, alt1, az2, alt2)
self.assertLess(np.max(separations), tol)
self.assertLess(np.mean(separations), tol_mean)
# Check that the fast can nearly round-trip
ra_back, dec_back = utils._approx_altAz2RaDec(alt2, az2, obs.site.latitude_rad,
obs.site.longitude_rad, mjd)
separations = utils._angularSeparation(ra, dec, ra_back, dec_back)
self.assertLess(np.max(separations), tol)
self.assertLess(np.mean(separations), tol_mean)
def hp2screen(inmap,
mjd,
height=500,
alt_limit=10.,
npts=600,
grid_x=None,
grid_y=None):
"""
Convert a healpy map to a flat screen at height h
"""
nside = hp.npix2nside(inmap.size)
unmasked = np.where(inmap != hp.UNSEEN)[0]
ra, dec = _hpid2RaDec(nside, unmasked)
alt, az = stupidFast_RaDec2AltAz(ra, dec, lat, lon, mjd)
good = np.where(np.degrees(alt) > alt_limit)
r = height / np.tan(alt[good])
x = r * np.cos(az[good])
y = r * np.sin(az[good])
z = inmap[unmasked][good]
if grid_x is None:
grid_x, grid_y = np.mgrid[x.min():x.max():600j, y.min():y.max():600j]
screen_grid = interpolate.griddata(np.vstack((x, y)).T,
z, (grid_x, grid_y),
method='nearest')
# Maybe compute the alt-az of the grid_x, grid_y so that it'll be fast to convert back to ra, dec?
# Pretty clear this should be a class rather than a bunch of functions.
return grid_x, grid_y, screen_grid
def ra_dec_hp_map(nside=64):
"""
Return all the RA,dec points for the centers of a healpix map
"""
ra, dec = _hpid2RaDec(nside, np.arange(hp.nside2npix(nside)))
return ra, dec
def __init__(self, nside=default_nside, alt_min=20.):
"""
Parameters
----------
alt_min : float
The minimum altitude one can point the telescope (degrees)
"""
if nside is None:
nside = utils.set_default_nside()
self.ra, self.dec = _hpid2RaDec(nside, np.arange(hp.nside2npix(nside)))
self.min_alt = np.radians(alt_min)
self.sin_dec = np.sin(self.dec)
self.cos_dec = np.cos(self.dec)
site = Site('LSST')
self.sin_lat = np.sin(site.latitude_rad)
self.cos_lat = np.cos(site.latitude_rad)
self.lon = site.longitude_rad
# Compute hour angle when field hits the alt_min
ha_alt_min = -np.arccos(
(np.sin(self.min_alt) - self.sin_dec * self.sin_lat) /
(self.cos_dec * self.cos_lat))
self.ha_alt_min = ha_alt_min
lmst_alt_min = ha_alt_min + self.ra
lmst_alt_min[np.where(lmst_alt_min < 0)] += 2. * np.pi
self.lmst_min = lmst_alt_min
self.nans = np.isnan(self.lmst_min)
def __init__(self,
nside=None,
min_alt=20.,
max_alt=82.,
shadow_minutes=40.,
penalty=np.nan,
site='LSST'):
super(Zenith_shadow_mask_basis_function, self).__init__(nside=nside)
self.update_on_newobs = False
self.penalty = penalty
self.min_alt = np.radians(min_alt)
self.max_alt = np.radians(max_alt)
self.ra, self.dec = _hpid2RaDec(nside, np.arange(hp.nside2npix(nside)))
self.shadow_minutes = np.radians(shadow_minutes / 60. * 360. / 24.)
# Compute the declination band where things could drift into zenith
self.decband = np.zeros(self.dec.size, dtype=float)
self.zenith_radius = np.radians(90. - max_alt) / 2.
site = Site(name=site)
self.lat_rad = site.latitude_rad
self.lon_rad = site.longitude_rad
self.decband[np.where(
(int_rounded(self.dec) < int_rounded(self.lat_rad +
self.zenith_radius))
& (int_rounded(self.dec) > int_rounded(self.lat_rad -
self.zenith_radius)))] = 1
self.result = np.empty(hp.nside2npix(self.nside), dtype=float)
self.result.fill(self.penalty)
def __init__(self, nside=default_nside, alt_max=86.5):
"""
Parameters
----------
alt_max : float
The maximum altitude one can point the telescope (degrees)
"""
if nside is None:
nside = utils.set_default_nside()
self.ra, self.dec = _hpid2RaDec(nside, np.arange(hp.nside2npix(nside)))
self.max_alt = np.radians(alt_max)
self.sin_dec = np.sin(self.dec)
self.cos_dec = np.cos(self.dec)
site = Site('LSST')
self.sin_lat = np.sin(site.latitude_rad)
self.cos_lat = np.cos(site.latitude_rad)
self.lon = site.longitude_rad
# compute the hour angle when a point hits the alt_max
cos_ha = (np.sin(self.max_alt) -
self.sin_dec * self.sin_lat) / (self.cos_dec * self.cos_lat)
self.lmst_max = np.arccos(cos_ha) + self.ra
self.nans = np.isnan(self.lmst_max)
def test_rad(self):
nside = 64
hpids = np.arange(hp.nside2npix(nside))
ra, dec = utils._hpid2RaDec(nside, hpids)
mjd = 59852.
obs = utils.ObservationMetaData(mjd=mjd)
alt1, az1, pa1 = utils._altAzPaFromRaDec(ra, dec, obs)
alt2, az2 = utils._approx_RaDec2AltAz(ra, dec, obs.site.latitude_rad,
obs.site.longitude_rad, mjd)
# Check that the fast is similar to the more precice transform
tol = np.radians(2)
tol_mean = np.radians(1.)
separations = utils._angularSeparation(az1, alt1, az2, alt2)
self.assertLess(np.max(separations), tol)
self.assertLess(np.mean(separations), tol_mean)
# Check that the fast can nearly round-trip
ra_back, dec_back = utils._approx_altAz2RaDec(alt2, az2,
obs.site.latitude_rad,
obs.site.longitude_rad,
mjd)
separations = utils._angularSeparation(ra, dec, ra_back, dec_back)
self.assertLess(np.max(separations), tol)
self.assertLess(np.mean(separations), tol_mean)
def __init__(self, nside=default_nside, max_airmass=2.5, polar_limit=-80.):
"""
Parameters
----------
max_airmass : float (2.5)
The maximum airmass to consider a point visible
polar_limit : float (-80.)
Consider anything below dec polar_limit to always be visible. (degrees)
"""
# most fields should have a min and max lmst where they are less than max_airmass
self.ra, self.dec = _hpid2RaDec(nside, np.arange(hp.nside2npix(nside)))
alt_limit = np.pi/2. - np.arccos(1./max_airmass)
site = Site('LSST')
lat = site.latitude_rad
self.lon = site.longitude_rad
sinalt = np.sin(alt_limit)
sindec = np.sin(self.dec)
sinlat = np.sin(lat)
cosdec = np.cos(self.dec)
coslat = np.cos(lat)
cosha = (sinalt - sindec*sinlat)/(cosdec*coslat)
# Here's the hour angle (plus or minus) for each healpixel
self.ha_limit = np.arccos(cosha)*12/np.pi
# Consider some regions circumpolar
self.ha_limit[np.where(self.dec < np.radians(polar_limit))] = 12.
self.polar_limit = polar_limit
self.feature = self.ra * 0.
def ra_dec_hp_map(nside=None):
"""
Return all the RA,dec points for the centers of a healpix map, in radians.
"""
if nside is None:
nside = set_default_nside()
ra, dec = _hpid2RaDec(nside, np.arange(hp.nside2npix(nside)))
return ra, dec
def bulge_pix(nside=32):
result = np.zeros(hp.nside2npix(nside))
ra, dec = _hpid2RaDec(nside, np.arange(hp.nside2npix(nside)))
coord = SkyCoord(ra=ra*u.rad, dec=dec*u.rad)
g_long, g_lat = coord.galactic.l.deg, coord.galactic.b.deg
bulge_pix = np.where((g_long > -20) & (g_long < 20.) & (g_lat > -10) & (g_lat < 10.))
result[bulge_pix] = 1
return result
def ss_footprints(nside=32, dist_to_eclip=25.):
fp = standard_goals(nside=nside)
et = ecliptic_target(nside=nside, dist_to_eclip=dist_to_eclip)
ra, dec = _hpid2RaDec(nside, np.arange(hp.nside2npix(nside)))
bridge = np.where((et > 0) & (dec < 0) & (fp['r'] != 1))
for key in fp:
fp[key][bridge] = np.max(fp[key])
return fp
def create_season_offset(nside, sun_RA_rad):
"""
Make an offset map so seasons roll properly
"""
hpindx = np.arange(hp.nside2npix(nside))
ra, dec = _hpid2RaDec(nside, hpindx)
offset = ra - sun_RA_rad + 2.*np.pi
offset = offset % (np.pi*2)
offset = offset * 365.25/(np.pi*2)
offset = -offset - 365.25
return offset
def _readMap(self):
filename = 'TRIstarDensity_%s_nside_%i.npz' % (self.filtername, self.nside)
starMap = np.load(os.path.join(self.mapDir, filename))
self.starMap = starMap['starDensity'].copy()
self.starMapBins = starMap['bins'].copy()
self.starmapNside = hp.npix2nside(np.size(self.starMap[:, 0]))
# note, the trilegal maps are in galactic coordinates, and nested healpix.
gal_l, gal_b = _hpid2RaDec(self.nside, np.arange(hp.nside2npix(self.nside)), nest=True)
# Convert that to RA,dec. Then do nearest neighbor lookup.
ra, dec = _equatorialFromGalactic(gal_l, gal_b)
self.tree = _buildTree(ra, dec)
def testRaDecsRad(self):
"""
Test that the Ra Dec conversions round-trip
"""
nside = 64
hpids = np.arange(hp.nside2npix(nside))
ra, dec = utils._hpid2RaDec(nside, hpids)
hpids_return = utils._raDec2Hpid(nside, ra, dec)
np.testing.assert_array_equal(hpids, hpids_return)
def __init__(self,
surveys,
nside=None,
camera='LSST',
rotator_limits=[85., 275.],
log=None):
"""
Parameters
----------
surveys : list (or list of lists) of lsst.sims.featureScheduler.survey objects
A list of surveys to consider. If multiple surveys return the same highest
reward value, the survey at the earliest position in the list will be selected.
Can also be a list of lists to make heirarchical priorities.
nside : int
A HEALpix nside value.
camera : str ('LSST')
Which camera to use for computing overlapping HEALpixels for an observation.
Can be 'LSST' or 'comcam'
rotator_limits : sequence of floats
"""
if nside is None:
nside = set_default_nside()
if log is None:
self.log = logging.getLogger(type(self).__name__)
else:
self.log = log.getChild(type(self).__name__)
# initialize a queue of observations to request
self.queue = []
# The indices of self.survey_lists that provided the last addition(s) to the queue
self.survey_index = [None, None]
# If we have a list of survey objects, convert to list-of-lists
if isinstance(surveys[0], list):
self.survey_lists = surveys
else:
self.survey_lists = [surveys]
self.nside = nside
hpid = np.arange(hp.nside2npix(nside))
self.ra_grid_rad, self.dec_grid_rad = _hpid2RaDec(nside, hpid)
# Should just make camera a class that takes a pointing and returns healpix indices
if camera == 'LSST':
self.pointing2hpindx = hp_in_lsst_fov(nside=nside)
elif camera == 'comcam':
self.pointing2hpindx = hp_in_comcam_fov(nside=nside)
else:
raise ValueError('camera %s not implamented' % camera)
# keep track of how many observations get flushed from the queue
self.flushed = 0
self.rotator_limits = np.sort(np.radians(rotator_limits))
def ecliptic_target(nside=32, dist_to_eclip=40., dec_max=30.):
"""Generate a target_map for the area around the ecliptic
"""
ra, dec = _hpid2RaDec(nside, np.arange(hp.nside2npix(nside)))
result = np.zeros(ra.size)
coord = SkyCoord(ra=ra * u.rad, dec=dec * u.rad)
eclip_lat = coord.barycentrictrueecliptic.lat.radian
good = np.where((np.abs(eclip_lat) < np.radians(dist_to_eclip))
& (dec < np.radians(dec_max)))
result[good] += 1
return result
def __init__(self,
nside=default_nside,
condition_features=None,
minAlt=20.,
maxAlt=82.,
azWidth=15.,
survey_features=None,
lat=-30.2444,
zenith_pad=15.,
zenith_min_alt=40.):
"""
Parameters
----------
minAlt : float (20.)
The minimum altitude to consider (degrees)
maxAlt : float (82.)
The maximum altitude to leave unmasked (degrees)
azWidth : float (15.)
The full-width azimuth to leave unmasked (degrees)
"""
if nside is None:
nside = utils.set_default_nside()
self.lat = np.radians(lat)
if survey_features is None:
self.survey_features = {}
if condition_features is None:
self.condition_features = {}
self.condition_features['altaz'] = features.AltAzFeature(
nside=nside)
self.minAlt = np.radians(minAlt)
self.maxAlt = np.radians(maxAlt)
# Convert to half-width for convienence
self.azWidth = np.radians(azWidth / 2.)
self.nside = nside
self.zenith_map = np.empty(hp.nside2npix(self.nside), dtype=float)
self.zenith_map.fill(hp.UNSEEN)
hpids = np.arange(self.zenith_map.size)
ra, dec = _hpid2RaDec(nside, hpids)
close_dec = np.where(
np.abs(dec - np.radians(lat)) < np.radians(zenith_pad))
self.zenith_min_alt = np.radians(zenith_min_alt)
self.zenith_map[close_dec] = 1
def __call__(self, ra, dec, rotSkyPos):
"""
Parameters
----------
ra : float
RA in radians
dec : float
Dec in radians
rotSkyPos : float
The rotation angle of the camera in radians
Returns
-------
indx : numpy array
The healpixels that are within the FoV
"""
x, y, z = _xyz_from_ra_dec(np.max(ra), np.max(dec))
# Healpixels within the inner circle
indices = self.tree.query_ball_point((x, y, z), self.inner_radius)
# Healpixels withing the outer circle
indices_all = np.array(self.tree.query_ball_point((x, y, z), self.outter_radius))
indices_to_check = indices_all[np.in1d(indices_all, indices, invert=True)]
cos_rot = np.cos(rotSkyPos)
sin_rot = np.sin(rotSkyPos)
x_rotated = self.corners_x*cos_rot - self.corners_y*sin_rot
y_rotated = self.corners_x*sin_rot + self.corners_y*cos_rot
# Draw the square that we want to check if points are in.
bbPath = mplPath.Path(np.array([[x_rotated[0], y_rotated[0]],
[x_rotated[1], y_rotated[1]],
[x_rotated[2], y_rotated[2]],
[x_rotated[3], y_rotated[3]],
[x_rotated[0], y_rotated[0]]]))
ra_to_check, dec_to_check = _hpid2RaDec(self.nside, indices_to_check)
# Project the indices to check to the tangent plane, see if they fall inside the polygon
x, y = gnomonic_project_toxy(ra_to_check, dec_to_check, ra, dec)
for i, xcheck in enumerate(x):
# I wonder if I can do this all at once rather than a loop?
if bbPath.contains_point((x[i], y[i])):
indices.append(indices_to_check[i])
return np.array(indices)
def big_sky(
nside=32,
weights={
'u': [0.31, 0.15, False],
'g': [0.44, 0.15],
'r': [1., 0.3],
'i': [1., 0.3],
'z': [0.9, 0.3],
'y': [0.9, 0.3, False]
}):
"""
Based on the Olsen et al Cadence White Paper
"""
# wfd covers -72.25 < dec < 12.4. Avoid galactic plane |b| > 15 deg
wfd_north = np.radians(12.4)
wfd_south = np.radians(-72.25)
full_north = np.radians(30.)
g_lat_limit = np.radians(8.)
ra, dec = _hpid2RaDec(nside, np.arange(hp.nside2npix(nside)))
total_map = np.zeros(ra.size)
coord = SkyCoord(ra=ra * u.rad, dec=dec * u.rad)
g_long, g_lat = coord.galactic.l.radian, coord.galactic.b.radian
# let's make a first pass here
total_map[np.where(dec < full_north)] = 1e-6
total_map[np.where((dec > wfd_south) & (dec < wfd_north)
& (np.abs(g_lat) > g_lat_limit))] = 1.
# Now let's break it down by filter
result = {}
for key in weights:
result[key] = total_map + 0.
result[key][np.where(result[key] == 1)] = weights[key][0]
result[key][np.where(result[key] == 1e-6)] = weights[key][1]
if len(weights[key]) == 3:
result[key][np.where(dec > wfd_north)] = 0.
return result
def hp_kd_tree(nside=set_default_nside(), leafsize=100):
"""
Generate a KD-tree of healpixel locations
Parameters
----------
nside : int
A valid healpix nside
leafsize : int (100)
Leafsize of the kdtree
Returns
-------
tree : scipy kdtree
"""
hpid = np.arange(hp.nside2npix(nside))
ra, dec = _hpid2RaDec(nside, hpid)
x, y, z = treexyz(ra, dec)
tree = kdtree(list(zip(x, y, z)), leafsize=leafsize, balanced_tree=False, compact_nodes=False)
return tree
def generate_events(nside=32,
mjd0=59853.5,
radius=15.,
survey_length=365.25 * 10,
rate=10.,
expires=3.,
seed=42):
"""
Parameters
----------
"""
np.random.seed(seed=seed)
ra, dec = _hpid2RaDec(nside, np.arange(hp.nside2npix(nside)))
radius = np.radians(radius)
# Use a ceil here so we get at least 1 event even if doing a short run.
n_events = np.int(np.ceil(survey_length / 365.25 * rate))
names = ['mjd_start', 'ra', 'dec', 'expires']
types = [float] * 4
event_table = np.zeros(n_events, dtype=list(zip(names, types)))
event_table['mjd_start'] = np.sort(
np.random.random(n_events)) * survey_length + mjd0
event_table['expires'] = event_table['mjd_start'] + expires
# Make sure latitude points spread correctly
# http://mathworld.wolfram.com/SpherePointPicking.html
event_table['ra'] = np.random.rand(n_events) * np.pi * 2
event_table['dec'] = np.arccos(2. * np.random.rand(n_events) -
1.) - np.pi / 2.
events = []
for i, event_time in enumerate(event_table['mjd_start']):
dist = _angularSeparation(ra, dec, event_table['ra'][i],
event_table['dec'][i])
good = np.where(dist <= radius)
footprint = np.zeros(ra.size, dtype=float)
footprint[good] = 1
events.append(TargetoO(i, footprint, event_time, expires))
events = Sim_targetoO_server(events)
return events, event_table
def __call__(self):
"""
Just point at the highest reward healpix
"""
if not self.reward_checked:
self.reward = self.calc_reward_function()
# Just find the best one
highest_reward = self.reward[np.where(~self.reward.mask)].max()
best = [np.min(np.where(self.reward == highest_reward)[0])]
# Could move this up to be a lookup rather than call every time.
ra, dec = _hpid2RaDec(default_nside, best)
observations = []
for i, indx in enumerate(best):
obs = empty_observation()
obs['RA'] = ra[i]
obs['dec'] = dec[i]
obs['filter'] = self.filtername
obs['nexp'] = 2.
obs['exptime'] = 30.
observations.append(obs)
return observations
def generate_events_simple(nside=32, mjd0=59853.5, radius=15.):
"""
Generate 3 simple ToO events
"""
ra, dec = _hpid2RaDec(nside, np.arange(hp.nside2npix(nside)))
radius = np.radians(radius)
nights = [1, 3, 6]
expires = 3
event_ra = np.radians([0, 10, 350])
event_dec = np.radians(-40.)
events = []
for i, nignt in enumerate(nights):
dist = _angularSeparation(ra, dec, event_ra[i], event_dec)
good = np.where(dist <= radius)
footprint = np.zeros(ra.size, dtype=float)
footprint[good] = 1
events.append(TargetoO(i, footprint, mjd0 + nights[i], expires))
events = Sim_targetoO_server(events)
return events
def hp_kd_tree(nside=None, leafsize=100, scale=1e5):
"""
Generate a KD-tree of healpixel locations
Parameters
----------
nside : int
A valid healpix nside
leafsize : int (100)
Leafsize of the kdtree
Returns
-------
tree : scipy kdtree
"""
if nside is None:
nside = set_default_nside()
hpid = np.arange(hp.nside2npix(nside))
ra, dec = _hpid2RaDec(nside, hpid)
return _buildTree(ra, dec, leafsize, scale=scale)
def hp2screen(inmap, mjd, height=500, alt_limit=10., npts=600, grid_x=None, grid_y=None):
"""
Convert a healpy map to a flat screen at height h
"""
nside = hp.npix2nside(inmap.size)
unmasked = np.where(inmap != hp.UNSEEN)[0]
ra, dec = _hpid2RaDec(nside, unmasked)
alt, az = stupidFast_RaDec2AltAz(ra, dec, lat, lon, mjd)
good = np.where(np.degrees(alt) > alt_limit)
r = height/np.tan(alt[good])
x = r*np.cos(az[good])
y = r*np.sin(az[good])
z = inmap[unmasked][good]
if grid_x is None:
grid_x, grid_y = np.mgrid[x.min():x.max():600j, y.min():y.max():600j]
screen_grid = interpolate.griddata(np.vstack((x, y)).T, z, (grid_x, grid_y), method='nearest')
# Maybe compute the alt-az of the grid_x, grid_y so that it'll be fast to convert back to ra, dec?
# Pretty clear this should be a class rather than a bunch of functions.
return grid_x, grid_y, screen_grid
def bulge_footprint(nside=32, bulge_frac=1., ll_frac=1., i_heavy=False):
sg = big_sky(nside=nside)
wfd_north = np.radians(12.4)
wfd_south = np.radians(-72.25)
ra, dec = _hpid2RaDec(nside, np.arange(hp.nside2npix(nside)))
wfd_pix = np.where(sg['r'] == 1)
# Zero out the bulge as it is now
bulge_pix = np.where((sg['r'] == 0.15) & (dec > wfd_south))
for key in sg:
sg[key][bulge_pix] = 0
coord = SkyCoord(ra=ra * u.rad, dec=dec * u.rad)
g_long, g_lat = coord.galactic.l.deg, coord.galactic.b.deg
bulge_pix = np.where((g_long > -20) & (g_long < 20.) & (g_lat > -10)
& (g_lat < 10.))
lo_lat_pix = np.where((np.abs(g_lat) < 10.) & (dec < wfd_north))
# scale things as desired
for key in sg:
sg[key][lo_lat_pix] = ll_frac * np.max(sg[key][wfd_pix])
sg[key][bulge_pix] = bulge_frac * np.max(sg[key][wfd_pix])
if i_heavy:
i_val = np.max(sg['i'][bulge_pix])
total_val = 0
for key in sg:
total_val += np.max(sg[key][bulge_pix])
not_i = total_val - i_val
scale_down = (not_i - i_val) / (not_i)
for key in sg:
if key == 'i':
sg[key][bulge_pix] = 2. * sg[key][bulge_pix]
else:
sg[key][bulge_pix] = scale_down * sg[key][bulge_pix]
return sg
def __init__(self, nside=default_nside, radius=1.75, radius_search=1.75):
"""
"""
self.nside = nside
# XXX -- should I shrink the radius slightly to get rid of overlap? That would be clever!
self.p2hp_search = pointings2hp(nside=nside, radius=radius_search)
self.p2hp = pointings2hp(nside=nside, radius=radius)
# Load up a list of pointings, chop them down to a small block
# XXX--Should write code to generate a new tellelation so we know where it came from,
# not just a random .dat file that's been floating around!
fields = read_fields()
good = np.where((fields['RA'] > np.radians(360. - 15.))
| (fields['RA'] < np.radians(15.)))
fields = fields[good]
good = np.where(np.abs(fields['dec']) < np.radians(15.))
fields = fields[good]
self.ra = fields['RA']
self.dec = fields['dec']
# Healpixel ra and dec
self.hp_ra, self.hp_dec = _hpid2RaDec(nside,
np.arange(hp.nside2npix(nside)))
def generate_taget_maps(nside=128):
"""
Generate a suite of target depths for each filter.
"""
npix = hp.nside2npix(nside)
ra, dec = utils._hpid2RaDec(nside, np.arange(npix))
coord = SkyCoord(ra=ra*u.rad, dec=dec*u.rad)
g_long, g_lat = coord.galactic.l.radian, coord.galactic.b.radian
eclip_long, eclip_lat = coord.barycentrictrueecliptic.lon.radian, coord.barycentrictrueecliptic.lat.radian
# Set borders of different regions
main_dec = 0
south_dec = np.radians(-60.)
nes_dec = 0.
nes_eclip = np.radians(15.)
gal_lat = np.radians(10.)
gal_long_end1 = np.radians(70.)
gal_long_end2 = np.radians(270.)
wfd_region = np.where((dec <= main_dec) &
((np.abs(g_lat) >= gal_lat) | (g_long > gal_long_end1)) &
(dec >= south_dec))
south_pole_region = np.where(dec < south_dec)
nes_region = np.where((dec > nes_dec) & (eclip_lat < nes_eclip))
# Need to add in the galactic plane taper.
galatic_plane_region = np.where((np.abs(g_lat) < gal_lat) &
((g_long < gal_long_end1) | (g_long > gal_long_end2)))
wfd_depth = {'u': 26.1, 'g': 27.4, 'r': 27.5, 'i': 26.8, 'z': 26.1, 'y': 24.9}
nes_depth = {'g': 26.9, 'r': 27.0, 'i': 26.3, 'z': 23.6}
south_pole_depth = {'u': 25.6, 'g': 26.9, 'r': 27.0, 'i': 26.3, 'z': 23.6, 'y': 24.4}
galactic_plane_depth = {'u': 25.6, 'g': 26.9, 'r': 27.0, 'i': 26.3, 'z': 23.6, 'y': 24.4}
all_filters = []
all_filters.extend(wfd_depth.keys())
all_filters.extend(nes_depth.keys())
all_filters.extend(south_pole_depth.keys())
all_filters.extend(galactic_plane_depth.keys())
all_filters = list(set(all_filters))
all_depth = {}
for filtername in all_filters:
all_depth[filtername] = np.empty(npix, dtype=float)
all_depth[filtername].fill(hp.UNSEEN)
# Should probably throw an error if any of the regions overlap.
for filtername in wfd_depth:
all_depth[filtername][wfd_region] = wfd_depth[filtername]
for filtername in south_pole_depth:
all_depth[filtername][south_pole_region] = south_pole_depth[filtername]
for filtername in nes_depth:
all_depth[filtername][nes_region] = nes_depth[filtername]
for filtername in galactic_plane_depth:
all_depth[filtername][galatic_plane_region] = galactic_plane_depth[filtername]
return all_depth
def gen_cloud_maps():
sm = sb.SkyModel(mags=True)
hpindx = np.arange(hp.nside2npix(32))
ra, dec = utils._hpid2RaDec(32, hpindx)
do_query = False
dbAddress = 'sqlite:///meddata.sqlite'
engine = sqla.create_engine(dbAddress)
connection = engine.raw_connection()
cursor = connection.cursor()
# Check all the mjd values in the db
if do_query:
query = 'select distinct(mjd) from medskybrightness;'
cursor.execute(query)
data = cursor.fetchall()
mjd_clean = [ack[0] for ack in data]
mjd_clean = np.array(mjd_clean)
# Let's go 57200 as a start
mjd_start = 57200
mjd_length = 40.
query = 'select distinct(mjd) from medskybrightness where mjd > %f and mjd < %f;' % (mjd_start,
mjd_start+mjd_length)
cursor.execute(query)
data = cursor.fetchall()
mjd_clean = np.array([ack[0] for ack in data])
cloud_maps = []
mjd_final = []
moon_limit = np.radians(30.)
am_limit = 2.5
maxI = mjd_clean[1:].size
for i, mjd in enumerate(mjd_clean[1:]):
if (mjd_clean[i]-mjd_clean[i-1]) < (2./60./24.):
sf = single_frame(mjd_clean[i-1])
sf2 = single_frame(mjd_clean[i])
if (sf.max() != hp.UNSEEN) & (sf2.max() != hp.UNSEEN):
diff = (sf-sf2)/sf
mask = np.where((sf == hp.UNSEEN) | (sf2 == hp.UNSEEN))
diff[mask] = hp.UNSEEN
if diff.max() != hp.UNSEEN:
cf, cloud_mask = cloudyness(diff, skyRMS_max=0.05)
sm.setRaDecMjd(ra, dec, mjd)
skymap = sm.returnMags()['r']
mask = np.where((skymap == hp.UNSEEN) | np.isnan(skymap))[0]
cloud_mask[mask] = 2
moon_dist = utils.haversine(ra, dec, sm.moonRA, sm.moonDec)
mask = np.where((moon_dist < moon_limit) | (sm.airmass < 1.) | (sm.airmass > am_limit))
cloud_mask[mask] = 2
cloud_maps.append(cloud_mask)
# put in some masking near the moon
mjd_final.append(mjd)
progress = i/float(maxI)*100
text = "\r progress = %.1f%%" % progress
sys.stdout.write(text)
sys.stdout.flush()
cloud_maps = np.array(cloud_maps)
map_key = {'good': 0, 'masked': 2, 'bright_in_current': -1, 'bright_in_last': 1}
np.savez('month_o_clouds.npz', cloud_maps=cloud_maps, mjds=mjd_final, map_key=map_key)
def __init__(self,
basis_functions,
basis_weights,
extra_features=None,
extra_basis_functions=None,
filtername='r',
filter2='g',
slew_approx=7.5,
filter_change_approx=140.,
read_approx=2.,
nexp=2,
min_exptime=15.,
max_exptime=60.,
ideal_pair_time=22.,
min_pair_time=15.,
search_radius=30.,
alt_max=85.,
az_range=90.,
smoothing_kernel=None,
nside=default_nside,
dither=True,
seed=42,
ignore_obs='ack',
tag_fields=False,
tag_map=None,
tag_names=None,
sun_alt_limit=-19.,
survey_note='blob',
sitename='LSST'):
"""
Parameters
----------
min_exp_time : float (15.)
The minimum exposure time to use (seconds). This is the exposure time when
conditions are dark time on the meridian (or better).
max_exp_time : float (60.)
The maximum exposure time to use (seconds)
filtername : str ('r')
The filter to observe in.
filter2 : str ('g')
The filter to pair with the first observation. If set to None, no pair
will be observed.
slew_approx : float (7.5)
The approximate slewtime between neerby fields (seconds). Used to calculate
how many observations can be taken in the desired time block.
filter_change_approx : float (140.)
The approximate time it takes to change filters (seconds).
ideal_pair_time : float (22.)
The ideal time gap wanted between observations to the same pointing (minutes)
min_pair_time : float (15.)
The minimum acceptable pair time (minutes)
search_radius : float (30.)
The radius around the reward peak to look for additional potential pointings (degrees)
alt_max : float (85.)
The maximum altitude to include (degrees).
az_range : float (90.)
The range of azimuths to consider around the peak reward value (degrees).
sitename : str ('LSST')
The name of the site to lookup latitude and longitude.
"""
if nside is None:
nside = set_default_nside()
if extra_features is None:
extra_features = {}
extra_features['night'] = features.Current_night()
extra_features['mounted_filters'] = features.Mounted_filters()
extra_features['mjd'] = features.Current_mjd()
extra_features[
'night_boundaries'] = features.CurrentNightBoundaries()
extra_features['sun_moon_alt'] = features.Sun_moon_alts()
extra_features['lmst'] = features.Current_lmst(
) # Pretty sure in hours
extra_features['current_filter'] = features.Current_filter()
extra_features['altaz'] = features.AltAzFeature()
if extra_basis_functions is None:
extra_basis_functions = {}
extra_basis_functions['filter1_m5diff'] = M5_diff_basis_function(
filtername=filtername, nside=nside)
if filter2 is not None:
extra_basis_functions[
'filter2_m5diff'] = M5_diff_basis_function(
filtername=filter2, nside=nside)
super(Vary_expt_survey,
self).__init__(basis_functions=basis_functions,
basis_weights=basis_weights,
extra_features=extra_features,
extra_basis_functions=extra_basis_functions,
filtername=filtername,
block_size=0,
smoothing_kernel=smoothing_kernel,
dither=dither,
seed=seed,
ignore_obs=ignore_obs,
tag_fields=tag_fields,
tag_map=tag_map,
tag_names=tag_names,
nside=nside)
self.nexp = nexp
self.min_exptime = min_exptime
self.max_exptime = max_exptime
self.slew_approx = slew_approx
self.read_approx = read_approx
self.hpids = np.arange(hp.nside2npix(self.nside))
# If we are taking pairs in same filter, no need to add filter change time.
if filtername == filter2:
filter_change_approx = 0
# Compute the minimum time needed to observe a blob (or observe, then repeat.)
if filter2 is not None:
self.time_needed = (min_pair_time * 60. * 2. + read_approx +
filter_change_approx) / 24. / 3600. # Days
else:
self.time_needed = (min_pair_time * 60. +
read_approx) / 24. / 3600. # Days
self.filter_set = set(filtername)
if filter2 is None:
self.filter2_set = self.filter_set
else:
self.filter2_set = set(filter2)
self.sun_alt_limit = np.radians(sun_alt_limit)
self.ra, self.dec = _hpid2RaDec(self.nside, self.hpids)
# Look up latitude and longitude for alt,az conversions later
# XXX: TODO: lat and lon should be in the Observatory feature. But that feature
# needs documentation on what's in it!
site = Site(name=sitename)
self.lat = site.latitude_rad
self.lon = site.longitude_rad
self.survey_note = survey_note
self.counter = 1 # start at 1, because 0 is default in empty observation
self.filter2 = filter2
self.search_radius = np.radians(search_radius)
self.az_range = np.radians(az_range)
self.alt_max = np.radians(alt_max)
self.min_pair_time = min_pair_time
self.ideal_pair_time = ideal_pair_time
def gen_cloud_maps():
sm = sb.SkyModel(mags=True)
hpindx = np.arange(hp.nside2npix(32))
ra, dec = utils._hpid2RaDec(32, hpindx)
do_query = False
dbAddress = 'sqlite:///meddata.sqlite'
engine = sqla.create_engine(dbAddress)
connection = engine.raw_connection()
cursor = connection.cursor()
# Check all the mjd values in the db
if do_query:
query = 'select distinct(mjd) from medskybrightness;'
cursor.execute(query)
data = cursor.fetchall()
mjd_clean = [ack[0] for ack in data]
mjd_clean = np.array(mjd_clean)
# Let's go 57200 as a start
mjd_start = 57200
mjd_length = 40.
query = 'select distinct(mjd) from medskybrightness where mjd > %f and mjd < %f;' % (
mjd_start, mjd_start + mjd_length)
cursor.execute(query)
data = cursor.fetchall()
mjd_clean = np.array([ack[0] for ack in data])
cloud_maps = []
mjd_final = []
moon_limit = np.radians(30.)
am_limit = 2.5
maxI = mjd_clean[1:].size
for i, mjd in enumerate(mjd_clean[1:]):
if (mjd_clean[i] - mjd_clean[i - 1]) < (2. / 60. / 24.):
sf = single_frame(mjd_clean[i - 1])
sf2 = single_frame(mjd_clean[i])
if (sf.max() != hp.UNSEEN) & (sf2.max() != hp.UNSEEN):
diff = (sf - sf2) / sf
mask = np.where((sf == hp.UNSEEN) | (sf2 == hp.UNSEEN))
diff[mask] = hp.UNSEEN
if diff.max() != hp.UNSEEN:
cf, cloud_mask = cloudyness(diff, skyRMS_max=0.05)
sm.setRaDecMjd(ra, dec, mjd)
skymap = sm.returnMags()['r']
mask = np.where((skymap == hp.UNSEEN)
| np.isnan(skymap))[0]
cloud_mask[mask] = 2
moon_dist = utils.haversine(ra, dec, sm.moonRA, sm.moonDec)
mask = np.where((moon_dist < moon_limit)
| (sm.airmass < 1.)
| (sm.airmass > am_limit))
cloud_mask[mask] = 2
cloud_maps.append(cloud_mask)
# put in some masking near the moon
mjd_final.append(mjd)
progress = i / float(maxI) * 100
text = "\r progress = %.1f%%" % progress
sys.stdout.write(text)
sys.stdout.flush()
cloud_maps = np.array(cloud_maps)
map_key = {
'good': 0,
'masked': 2,
'bright_in_current': -1,
'bright_in_last': 1
}
np.savez('month_o_clouds.npz',
cloud_maps=cloud_maps,
mjds=mjd_final,
map_key=map_key)
def __init__(self,
basis_functions,
basis_weights,
filtername1='r',
filtername2='g',
slew_approx=7.5,
filter_change_approx=140.,
read_approx=2.,
exptime=30.,
nexp=2,
nexp_dict=None,
ideal_pair_time=22.,
min_pair_time=15.,
search_radius=30.,
alt_max=85.,
az_range=90.,
flush_time=30.,
smoothing_kernel=None,
nside=None,
dither=True,
seed=42,
ignore_obs=None,
survey_note='blob',
detailers=None,
camera='LSST',
twilight_scale=True,
in_twilight=False,
check_scheduled=True,
min_area=None,
grow_blob=True,
area_required=None):
if nside is None:
nside = set_default_nside()
super(Blob_survey, self).__init__(basis_functions=basis_functions,
basis_weights=basis_weights,
filtername=None,
block_size=0,
smoothing_kernel=smoothing_kernel,
dither=dither,
seed=seed,
ignore_obs=ignore_obs,
nside=nside,
detailers=detailers,
camera=camera,
area_required=area_required)
self.flush_time = flush_time / 60. / 24. # convert to days
self.nexp = nexp
self.nexp_dict = nexp_dict
self.exptime = exptime
self.slew_approx = slew_approx
self.read_approx = read_approx
self.hpids = np.arange(hp.nside2npix(self.nside))
self.twilight_scale = twilight_scale
self.in_twilight = in_twilight
self.grow_blob = grow_blob
if self.twilight_scale & self.in_twilight:
warnings.warn(
'Both twilight_scale and in_twilight are set to True. That is probably wrong.'
)
self.min_area = min_area
self.check_scheduled = check_scheduled
# If we are taking pairs in same filter, no need to add filter change time.
if filtername1 == filtername2:
filter_change_approx = 0
# Compute the minimum time needed to observe a blob (or observe, then repeat.)
if filtername2 is not None:
self.time_needed = (min_pair_time * 60. * 2. + exptime +
read_approx +
filter_change_approx) / 24. / 3600. # Days
else:
self.time_needed = (min_pair_time * 60. + exptime +
read_approx) / 24. / 3600. # Days
self.filter_set = set(filtername1)
if filtername2 is None:
self.filter2_set = self.filter_set
else:
self.filter2_set = set(filtername2)
self.ra, self.dec = _hpid2RaDec(self.nside, self.hpids)
self.survey_note = survey_note
self.counter = 1 # start at 1, because 0 is default in empty observation
self.filtername1 = filtername1
self.filtername2 = filtername2
self.search_radius = np.radians(search_radius)
self.az_range = np.radians(az_range)
self.alt_max = np.radians(alt_max)
self.min_pair_time = min_pair_time
self.ideal_pair_time = ideal_pair_time
self.pixarea = hp.nside2pixarea(self.nside, degrees=True)
# If we are only using one filter, this could be useful
if (self.filtername2 is None) | (self.filtername1 == self.filtername2):
self.filtername = self.filtername1
def __init__(self, nside=128, radius=1.75):
hpindex = np.arange(hp.nside2npix(nside))
ra, dec = _hpid2RaDec(nside, hpindex)
self._setRad(radius=radius)
self._buildTree(ra, dec)