def _checkNeighbors(self, qs, root, x, y, tolerance):
i = qs.id(root, x, y)
neighbors = qs.getNeighbors(i)
centers = [qs.getCenter(n) for n in neighbors]
c = qs.getCenter(i)
for v in centers:
self.assertTrue(geom.cartesianAngularSep(c, v) < tolerance)
def _checkNeighbors(self, qs, root, x, y, tolerance):
i = qs.id(root, x, y)
neighbors = qs.getNeighbors(i)
centers = [qs.getCenter(n) for n in neighbors]
c = qs.getCenter(i)
for v in centers:
self.assertTrue(geom.cartesianAngularSep(c, v) < tolerance)
def getGeometry(self, pixelId, fiducial=False):
"""Returns a spherical convex polygon corresponding to the fiducial
or padded boundaries of the sky-pixel with the specified id.
"""
root, ix, iy = self.coords(pixelId)
xl = 2.0 * float(ix) / float(self.resolution) - 1.0
xr = 2.0 * float(ix + 1) / float(self.resolution) - 1.0
yb = 2.0 * float(iy) / float(self.resolution) - 1.0
yt = 2.0 * float(iy + 1) / float(self.resolution) - 1.0
c, x, y = self.center[root], self.x[root], self.y[root]
v = list(
map(geom.normalize,
[(c[0] + xl * x[0] + yb * y[0], c[1] + xl * x[1] + yb * y[1],
c[2] + xl * x[2] + yb * y[2]),
(c[0] + xr * x[0] + yb * y[0], c[1] + xr * x[1] + yb * y[1],
c[2] + xr * x[2] + yb * y[2]),
(c[0] + xr * x[0] + yt * y[0], c[1] + xr * x[1] + yt * y[1],
c[2] + xr * x[2] + yt * y[2]),
(c[0] + xl * x[0] + yt * y[0], c[1] + xl * x[1] + yt * y[1],
c[2] + xl * x[2] + yt * y[2])]))
if not fiducial and self.padding > 0.0:
# Determine angles by which edge planes must be rotated outwards
sp = math.sin(self.padding)
theta = [
0.5 * geom.cartesianAngularSep(x[0], x[1])
for x in ((v[0], v[3]), (v[1], v[0]), (v[2], v[1]), (v[3],
v[2]))
]
sina = [sp / math.cos(math.radians(x)) for x in theta]
cosa = [math.sqrt(1.0 - x * x) for x in sina]
# find plane equations of fiducial pixel boundaries
xlp = self.xplane[root][ix][0]
ybp = self.yplane[root][iy][0]
xrp = self.xplane[root][ix + 1][0]
ytp = self.yplane[root][iy + 1][0]
# rotate edge planes outwards
xlp = self.xrot[root](xlp, -sina[0], cosa[0])
ybp = self.yrot[root](ybp, -sina[1], cosa[1])
xrp = self.xrot[root](xrp, sina[2], cosa[2])
ytp = self.yrot[root](ytp, sina[3], cosa[3])
# intersect rotated planes to find vertices of padded sky-pixel
v = list(
map(geom.normalize, [
geom.cross(xlp, ybp),
geom.cross(xrp, ybp),
geom.cross(xrp, ytp),
geom.cross(xlp, ytp)
]))
return geom.SphericalConvexPolygon(v)
def getGeometry(self, pixelId, fiducial=False):
"""Returns a spherical convex polygon corresponding to the fiducial
or padded boundaries of the sky-pixel with the specified id.
"""
root, ix, iy = self.coords(pixelId)
xl = 2.0 * float(ix) / float(self.resolution) - 1.0
xr = 2.0 * float(ix + 1) / float(self.resolution) - 1.0
yb = 2.0 * float(iy) / float(self.resolution) - 1.0
yt = 2.0 * float(iy + 1) / float(self.resolution) - 1.0
c, x, y = self.center[root], self.x[root], self.y[root]
v = map(geom.normalize, [(c[0] + xl * x[0] + yb * y[0],
c[1] + xl * x[1] + yb * y[1],
c[2] + xl * x[2] + yb * y[2]),
(c[0] + xr * x[0] + yb * y[0],
c[1] + xr * x[1] + yb * y[1],
c[2] + xr * x[2] + yb * y[2]),
(c[0] + xr * x[0] + yt * y[0],
c[1] + xr * x[1] + yt * y[1],
c[2] + xr * x[2] + yt * y[2]),
(c[0] + xl * x[0] + yt * y[0],
c[1] + xl * x[1] + yt * y[1],
c[2] + xl * x[2] + yt * y[2])])
if not fiducial and self.padding > 0.0:
# Determine angles by which edge planes must be rotated outwards
sp = math.sin(self.padding)
theta = map(lambda x: 0.5 * geom.cartesianAngularSep(x[0], x[1]),
((v[0],v[3]), (v[1],v[0]), (v[2],v[1]), (v[3],v[2])))
sina = map(lambda x: sp / math.cos(math.radians(x)), theta)
cosa = map(lambda x: math.sqrt(1.0 - x * x), sina)
# find plane equations of fiducial pixel boundaries
xlp = self.xplane[root][ix][0]
ybp = self.yplane[root][iy][0]
xrp = self.xplane[root][ix + 1][0]
ytp = self.yplane[root][iy + 1][0]
# rotate edge planes outwards
xlp = self.xrot[root](xlp, -sina[0], cosa[0])
ybp = self.yrot[root](ybp, -sina[1], cosa[1])
xrp = self.xrot[root](xrp, sina[2], cosa[2])
ytp = self.yrot[root](ytp, sina[3], cosa[3])
# intersect rotated planes to find vertices of padded sky-pixel
v = map(geom.normalize, [geom.cross(xlp, ybp),
geom.cross(xrp, ybp),
geom.cross(xrp, ytp),
geom.cross(xlp, ytp)])
return geom.SphericalConvexPolygon(v)
def __init__(self, resolution, paddingRad):
"""Creates a new quad-sphere sky pixelisation.
resolution: the number of pixels along the x and y axes of each root
pixel.
paddingRad: the angular separation (rad) by which fiducial sky-pixels
are to be padded.
"""
if not isinstance(resolution, numbers.Integral):
raise TypeError(
'Quad-sphere resolution parameter must be an integer')
if resolution < 3 or resolution > 16384:
raise RuntimeError(
'Quad-sphere resolution must be in range [3, 16384]')
if not isinstance(paddingRad, float):
raise TypeError('Quad-sphere pixel padding angle must be a float')
if paddingRad < 0.0 or paddingRad >= math.pi * 0.25:
raise RuntimeError(
'Quad-sphere pixel padding radius must be in range [0, 45) deg'
)
R = resolution
self.resolution = R
self.padding = paddingRad
x, y, z = (1.0, 0.0, 0.0), (0.0, 1.0, 0.0), (0.0, 0.0, 1.0)
nx, ny, nz = (-1.0, 0.0, 0.0), (0.0, -1.0, 0.0), (0.0, 0.0, -1.0)
# center of each root pixel
self.center = [z, x, y, nx, ny, nz]
# x basis vector of each root pixel
self.x = [ny, y, nx, ny, x, ny]
# y basis vector of each root pixel
self.y = [x, z, z, z, z, nx]
# functions for rotating vectors in direction of increasing x (per root pixel)
self.xrot = [_rotX, _rotZ, _rotZ, _rotZ, _rotZ, _rotNX]
# functions for rotating vectors in direction of increasing y (per root pixel)
self.yrot = [_rotY, _rotNY, _rotX, _rotY, _rotNX, _rotY]
# Compute fiducial and splitting x/y planes for each root pixel
self.xplane = []
self.yplane = []
sp = math.sin(self.padding)
for root in range(6):
xplanes = []
yplanes = []
c, x, y = self.center[root], self.x[root], self.y[root]
for i in range(R + 1):
xfp = self._fiducialXPlane(root, i)
yfp = self._fiducialYPlane(root, i)
f = 2.0 * float(i) / float(R) - 1.0
thetaX = math.radians(0.5 * geom.cartesianAngularSep(
(c[0] + f * x[0] + y[0], c[1] + f * x[1] + y[1],
c[2] + f * x[2] + y[2]),
(c[0] + f * x[0] - y[0], c[1] + f * x[1] - y[1],
c[2] + f * x[2] - y[2])))
thetaY = math.radians(0.5 * geom.cartesianAngularSep(
(c[0] + x[0] + f * y[0], c[1] + x[1] + f * y[1],
c[2] + x[2] + f * y[2]),
(c[0] - x[0] + f * y[0], c[1] - x[1] + f * y[1],
c[2] - x[2] + f * y[2])))
sinrx = sp / math.cos(thetaX)
sinry = sp / math.cos(thetaY)
cosrx = math.sqrt(1.0 - sinrx * sinrx)
cosry = math.sqrt(1.0 - sinry * sinry)
if i == 0:
xlp = None
ybp = None
else:
xlp = self.xrot[root](xfp, sinrx, cosrx)
ybp = self.yrot[root](yfp, sinry, cosry)
xlp = (-xlp[0], -xlp[1], -xlp[2])
ybp = (-ybp[0], -ybp[1], -ybp[2])
if i == R:
xrp = None
ytp = None
else:
xrp = self.xrot[root](xfp, -sinrx, cosrx)
ytp = self.yrot[root](yfp, -sinry, cosry)
xplanes.append((xfp, xlp, xrp))
yplanes.append((yfp, ybp, ytp))
self.xplane.append(xplanes)
self.yplane.append(yplanes)
# Corner pixel neighbors.
self.cornerNeighbors = (
# root pixel 0
(
# x, y = 0, 0
(self.id(0, 1, 0), self.id(0, 1, 1), self.id(0, 0, 1),
self.id(2, R - 1, R - 1), self.id(2, R - 2, R - 1),
self.id(3, 0, R - 1), self.id(3, 1, R - 1)),
# x, y = 0, R-1
(self.id(0, 1, R - 1), self.id(0, 1,
R - 2), self.id(0, 0, R - 2),
self.id(1, R - 1, R - 1), self.id(1, R - 2, R - 1),
self.id(2, 0, R - 1), self.id(2, 1, R - 1)),
# x, y = R-1, 0
(self.id(0, R - 2, 0), self.id(0, R - 2,
1), self.id(0, R - 1, 1),
self.id(3, R - 2, R - 1), self.id(3, R - 1, R - 1),
self.id(4, 0, R - 1), self.id(4, 1, R - 1)),
# x, y = R-1, R-1
(self.id(0, R - 2, R - 1), self.id(0, R - 2, R - 2),
self.id(0, R - 1,
R - 2), self.id(1, 0, R - 1), self.id(1, 1, R - 1),
self.id(4, R - 2, R - 1), self.id(4, R - 1, R - 1))),
# root pixel 1
(
# x, y = 0, 0
(self.id(1, 1, 0), self.id(1, 1, 1), self.id(1, 0, 1),
self.id(4, R - 1, 0), self.id(4, R - 1, 1),
self.id(5, R - 2, 0), self.id(5, R - 1, 0)),
# x, y = 0, R-1
(self.id(1, 1, R - 1), self.id(1, 1,
R - 2), self.id(1, 0, R - 2),
self.id(4, R - 1, R - 2), self.id(4, R - 1, R - 1),
self.id(0, R - 2, R - 1), self.id(0, R - 1, R - 1)),
# x, y = R-1, 0
(self.id(1, R - 2, 0), self.id(1, R - 2, 1),
self.id(1, R - 1, 1), self.id(2, 0, 0), self.id(2, 0, 1),
self.id(5, 0, 0), self.id(5, 1, 0)),
# x, y = R-1, R-1
(self.id(1, R - 2, R - 1), self.id(1, R - 2, R - 2),
self.id(1, R - 1, R - 2), self.id(2, 0, R - 2),
self.id(2, 0, R - 1), self.id(0, 0,
R - 1), self.id(0, 1, R - 1))),
# root pixel 2
(
# x, y = 0, 0
(self.id(2, 1, 0), self.id(2, 1, 1), self.id(2, 0, 1),
self.id(1, R - 1,
0), self.id(1, R - 1,
1), self.id(5, 0, 0), self.id(5, 0, 1)),
# x, y = 0, R-1
(self.id(2, 1, R - 1), self.id(2, 1,
R - 2), self.id(2, 0, R - 2),
self.id(1, R - 1, R - 2), self.id(1, R - 1, R - 1),
self.id(0, 0, R - 2), self.id(0, 0, R - 1)),
# x, y = R-1, 0
(self.id(2, R - 2, 0), self.id(2, R - 2, 1),
self.id(2, R - 1, 1), self.id(3, 0, 0), self.id(3, 0, 1),
self.id(5, 0, R - 2), self.id(5, 0, R - 1)),
# x, y = R-1, R-1
(self.id(2, R - 2, R - 1), self.id(2, R - 2, R - 2),
self.id(2, R - 1, R - 2), self.id(3, 0, R - 2),
self.id(3, 0, R - 1), self.id(0, 0, 0), self.id(0, 0, 1))),
# root pixel 3
(
# x, y = 0, 0
(self.id(3, 1, 0), self.id(3, 1, 1), self.id(3, 0, 1),
self.id(2, R - 1, 0), self.id(2, R - 1, 1),
self.id(5, 0, R - 1), self.id(5, 1, R - 1)),
# x, y = 0, R-1
(self.id(3, 1, R - 1), self.id(3, 1, R - 2),
self.id(3, 0, R - 2), self.id(2, R - 1, R - 2),
self.id(2, R - 1, R - 1), self.id(0, 0, 0), self.id(0, 1, 0)),
# x, y = R-1, 0
(self.id(3, R - 2, 0), self.id(3, R - 2, 1),
self.id(3, R - 1, 1), self.id(4, 0, 0), self.id(4, 0, 1),
self.id(5, R - 2, R - 1), self.id(5, R - 1, R - 1)),
# x, y = R-1, R-1
(self.id(3, R - 2, R - 1), self.id(3, R - 2, R - 2),
self.id(3, R - 1, R - 2), self.id(4, 0, R - 2),
self.id(4, 0, R - 1), self.id(0, R - 2,
0), self.id(0, R - 1, 0))),
# root pixel 4
(
# x, y = 0, 0
(self.id(4, 1, 0), self.id(4, 1, 1), self.id(4, 0, 1),
self.id(3, R - 1, 0), self.id(3, R - 1, 1),
self.id(5, R - 1, R - 2), self.id(5, R - 1, R - 1)),
# x, y = 0, R-1
(self.id(4, 1, R - 1), self.id(4, 1, R - 2),
self.id(4, 0, R - 2), self.id(3, R - 1, R - 2),
self.id(3, R - 1, R - 1), self.id(0, R - 1,
0), self.id(0, R - 1, 1)),
# x, y = R-1, 0
(self.id(4, R - 2, 0), self.id(4, R - 2, 1),
self.id(4, R - 1, 1), self.id(1, 0, 0), self.id(1, 0, 1),
self.id(5, R - 1, 0), self.id(5, R - 1, 1)),
# x, y = R-1, R-1
(self.id(4, R - 2, R - 1), self.id(4, R - 2, R - 2),
self.id(4, R - 1,
R - 2), self.id(1, 0, R - 2), self.id(1, 0, R - 1),
self.id(0, R - 1, R - 2), self.id(0, R - 1, R - 1))),
# root pixel 5
(
# x, y = 0, 0
(self.id(5, 1, 0), self.id(5, 1, 1), self.id(5, 0, 1),
self.id(1, R - 2,
0), self.id(1, R - 1,
0), self.id(2, 0, 0), self.id(2, 1, 0)),
# x, y = 0, R-1
(self.id(5, 1, R - 1), self.id(5, 1, R - 2),
self.id(5, 0, R - 2), self.id(2, R - 2, 0),
self.id(2, R - 1, 0), self.id(3, 0, 0), self.id(3, 1, 0)),
# x, y = R-1, 0
(self.id(5, R - 2, 0), self.id(5, R - 2, 1),
self.id(5, R - 1, 1), self.id(1, 0, 0), self.id(1, 1, 0),
self.id(4, R - 2, 0), self.id(4, R - 1, 0)),
# x, y = R-1, R-1
(self.id(5, R - 2, R - 1), self.id(5, R - 2, R - 2),
self.id(5, R - 1, R - 2), self.id(3, R - 2, 0),
self.id(3, R - 1, 0), self.id(4, 0, 0), self.id(4, 1, 0))),
)
def getAllSipWcs(cursor, qsp, kind):
"""Constructs a Wcs object from each entry in the Science_Ccd_Exposure
table. SIP distortion parameters are read in from
Science_Ccd_Exposure_Metadata. Returns a 3-tuple with the following
contents:
- a mapping from sky-tiles to a list of (WCS, filter) tuples for
overlapping science CCDs
- a list of all (WCS, filter) tuples read in
- a bounding circle (lsst.geom.SphericalCircle) for all science CCDs.
"""
wcsMap = {}
wcsList = []
offset = 0
centers = []
radius = 0.0
n = 0
blocksize = 1000
# An alternative would be to create WCSes from FITS metadata or FITS
# files themselves, but these aren't always available.
while True:
cursor.execute("""SELECT scienceCcdExposureId, filterId,
raDeSys, equinox, ctype1, ctype2,
crpix1, crpix2, crval1, crval2,
cd1_1, cd1_2, cd2_1, cd2_2
FROM Science_Ccd_Exposure LIMIT %d, %d
""" % (offset, blocksize))
rows = cursor.fetchall()
if len(rows) == 0:
break
# For each CCD in the results, build up a metadata PropertySet
# including SIP distortion terms and use it to obtain a WCS.
for row in rows:
ps = dafBase.PropertySet()
ps.setString("RADESYS", row[2])
ps.setDouble("EQUINOX", row[3])
ps.setString("CTYPE1", row[4])
ps.setString("CTYPE2", row[5])
ps.setString("CUNIT1", "deg")
ps.setString("CUNIT2", "deg")
ps.setDouble("CRPIX1", row[6])
ps.setDouble("CRPIX2", row[7])
ps.setDouble("CRVAL1", row[8])
ps.setDouble("CRVAL2", row[9])
ps.setDouble("CD1_1", row[10])
ps.setDouble("CD1_2", row[11])
ps.setDouble("CD2_1", row[12])
ps.setDouble("CD2_2", row[13])
if row[4].endswith("-SIP") and row[5].endswith("-SIP"):
cursor.execute("""
SELECT metadataKey, intValue
FROM Science_Ccd_Exposure_Metadata
WHERE scienceCcdExposureId = %d AND
intValue IS NOT NULL AND
metadataKey RLIKE '^[AB]P?_ORDER$'
""" % row[0])
for k, v in cursor.fetchall():
ps.setInt(k, v)
cursor.execute("""
SELECT metadataKey, doubleValue
FROM Science_Ccd_Exposure_Metadata
WHERE scienceCcdExposureId = %d AND
doubleValue IS NOT NULL AND
metadataKey RLIKE '^[AB]P?_[0-9]+_[0-9]+$'
""" % row[0])
for k, v in cursor.fetchall():
ps.setDouble(k, v)
wcs = afwImage.makeWcs(ps)
wcsList.append([wcs, row[1]])
if kind == 'imsim':
poly = qs.imageToPolygon(wcs, 4072.0, 4000.0, 0.0)
else:
poly = qs.imageToPolygon(wcs, 2048.0, 4610.0, 0.0)
bc = poly.getBoundingCircle()
radius = max(radius, bc.getRadius())
centers.append(geom.cartesianUnitVector(bc.getCenter()))
pix = qsp.intersect(poly)
for skyTile in pix:
if skyTile in wcsMap:
wcsMap[skyTile].append([wcs, row[1]])
else:
wcsMap[skyTile] = [[wcs, row[1]]]
n = offset + len(rows)
print "... %d" % n
offset += blocksize
x, y, z = 0.0, 0.0, 0.0
for v in centers:
x += v[0]
y += v[1]
z += v[2]
cen = geom.normalize((x, y, z))
r = max(geom.cartesianAngularSep(cen, v) for v in centers) + 2.0 * radius
return wcsMap, wcsList, geom.SphericalCircle(cen, r)
def __init__(self, resolution, paddingRad):
"""Creates a new quad-sphere sky pixelisation.
resolution: the number of pixels along the x and y axes of each root
pixel.
paddingRad: the angular separation (rad) by which fiducial sky-pixels
are to be padded.
"""
if not isinstance(resolution, (int, long)):
raise TypeError(
'Quad-sphere resolution parameter must be an int or long')
if resolution < 3 or resolution > 16384:
raise RuntimeError(
'Quad-sphere resolution must be in range [3, 16384]')
if not isinstance(paddingRad, float):
raise TypeError(
'Quad-sphere pixel padding angle must be a float')
if paddingRad < 0.0 or paddingRad >= math.pi * 0.25:
raise RuntimeError(
'Quad-sphere pixel padding radius must be in range [0, 45) deg')
R = resolution
self.resolution = R
self.padding = paddingRad
x, y, z = (1.0, 0.0, 0.0), (0.0, 1.0, 0.0), (0.0, 0.0, 1.0)
nx, ny, nz = (-1.0, 0.0, 0.0), (0.0, -1.0, 0.0), (0.0, 0.0, -1.0)
# center of each root pixel
self.center = [z, x, y, nx, ny, nz]
# x basis vector of each root pixel
self.x = [ny, y, nx, ny, x, ny]
# y basis vector of each root pixel
self.y = [x, z, z, z, z, nx]
# functions for rotating vectors in direction of increasing x (per root pixel)
self.xrot = [_rotX, _rotZ, _rotZ, _rotZ, _rotZ, _rotNX]
# functions for rotating vectors in direction of increasing y (per root pixel)
self.yrot = [_rotY, _rotNY, _rotX, _rotY, _rotNX, _rotY]
# Compute fiducial and splitting x/y planes for each root pixel
self.xplane = []
self.yplane = []
sp = math.sin(self.padding)
for root in xrange(6):
xplanes = []
yplanes = []
c, x, y = self.center[root], self.x[root], self.y[root]
for i in xrange(R + 1):
xfp = self._fiducialXPlane(root, i)
yfp = self._fiducialYPlane(root, i)
f = 2.0 * float(i) / float(R) - 1.0
thetaX = math.radians(0.5 * geom.cartesianAngularSep(
(c[0] + f * x[0] + y[0],
c[1] + f * x[1] + y[1],
c[2] + f * x[2] + y[2]),
(c[0] + f * x[0] - y[0],
c[1] + f * x[1] - y[1],
c[2] + f * x[2] - y[2])))
thetaY = math.radians(0.5 * geom.cartesianAngularSep(
(c[0] + x[0] + f * y[0],
c[1] + x[1] + f * y[1],
c[2] + x[2] + f * y[2]),
(c[0] - x[0] + f * y[0],
c[1] - x[1] + f * y[1],
c[2] - x[2] + f * y[2])))
sinrx = sp / math.cos(thetaX)
sinry = sp / math.cos(thetaY)
cosrx = math.sqrt(1.0 - sinrx * sinrx)
cosry = math.sqrt(1.0 - sinry * sinry)
if i == 0:
xlp = None
ybp = None
else:
xlp = self.xrot[root](xfp, sinrx, cosrx)
ybp = self.yrot[root](yfp, sinry, cosry)
xlp = (-xlp[0], -xlp[1], -xlp[2])
ybp = (-ybp[0], -ybp[1], -ybp[2])
if i == R:
xrp = None
ytp = None
else:
xrp = self.xrot[root](xfp, -sinrx, cosrx)
ytp = self.yrot[root](yfp, -sinry, cosry)
xplanes.append((xfp, xlp, xrp))
yplanes.append((yfp, ybp, ytp))
self.xplane.append(xplanes)
self.yplane.append(yplanes)
# Corner pixel neighbors.
self.cornerNeighbors = (
# root pixel 0
(
# x, y = 0, 0
(self.id(0, 1, 0),
self.id(0, 1, 1),
self.id(0, 0, 1),
self.id(2, R - 1, R - 1),
self.id(2, R - 2, R - 1),
self.id(3, 0, R - 1),
self.id(3, 1, R - 1)),
# x, y = 0, R-1
(self.id(0, 1, R - 1),
self.id(0, 1, R - 2),
self.id(0, 0, R - 2),
self.id(1, R - 1, R - 1),
self.id(1, R - 2, R - 1),
self.id(2, 0, R - 1),
self.id(2, 1, R - 1)),
# x, y = R-1, 0
(self.id(0, R - 2, 0),
self.id(0, R - 2, 1),
self.id(0, R - 1, 1),
self.id(3, R - 2, R - 1),
self.id(3, R - 1, R - 1),
self.id(4, 0, R - 1),
self.id(4, 1, R - 1)),
# x, y = R-1, R-1
(self.id(0, R - 2, R - 1),
self.id(0, R - 2, R - 2),
self.id(0, R - 1, R - 2),
self.id(1, 0, R - 1),
self.id(1, 1, R - 1),
self.id(4, R - 2, R - 1),
self.id(4, R - 1, R - 1))
),
# root pixel 1
(
# x, y = 0, 0
(self.id(1, 1, 0),
self.id(1, 1, 1),
self.id(1, 0, 1),
self.id(4, R - 1, 0),
self.id(4, R - 1, 1),
self.id(5, R - 2, 0),
self.id(5, R - 1, 0)),
# x, y = 0, R-1
(self.id(1, 1, R - 1),
self.id(1, 1, R - 2),
self.id(1, 0, R - 2),
self.id(4, R - 1, R - 2),
self.id(4, R - 1, R - 1),
self.id(0, R - 2, R - 1),
self.id(0, R - 1, R - 1)),
# x, y = R-1, 0
(self.id(1, R - 2, 0),
self.id(1, R - 2, 1),
self.id(1, R - 1, 1),
self.id(2, 0, 0),
self.id(2, 0, 1),
self.id(5, 0, 0),
self.id(5, 1, 0)),
# x, y = R-1, R-1
(self.id(1, R - 2, R - 1),
self.id(1, R - 2, R - 2),
self.id(1, R - 1, R - 2),
self.id(2, 0, R - 2),
self.id(2, 0, R - 1),
self.id(0, 0, R - 1),
self.id(0, 1, R - 1))
),
# root pixel 2
(
# x, y = 0, 0
(self.id(2, 1, 0),
self.id(2, 1, 1),
self.id(2, 0, 1),
self.id(1, R - 1, 0),
self.id(1, R - 1, 1),
self.id(5, 0, 0),
self.id(5, 0, 1)),
# x, y = 0, R-1
(self.id(2, 1, R - 1),
self.id(2, 1, R - 2),
self.id(2, 0, R - 2),
self.id(1, R - 1, R - 2),
self.id(1, R - 1, R - 1),
self.id(0, 0, R - 2),
self.id(0, 0, R - 1)),
# x, y = R-1, 0
(self.id(2, R - 2, 0),
self.id(2, R - 2, 1),
self.id(2, R - 1, 1),
self.id(3, 0, 0),
self.id(3, 0, 1),
self.id(5, 0, R - 2),
self.id(5, 0, R - 1)),
# x, y = R-1, R-1
(self.id(2, R - 2, R - 1),
self.id(2, R - 2, R - 2),
self.id(2, R - 1, R - 2),
self.id(3, 0, R - 2),
self.id(3, 0, R - 1),
self.id(0, 0, 0),
self.id(0, 0, 1))
),
# root pixel 3
(
# x, y = 0, 0
(self.id(3, 1, 0),
self.id(3, 1, 1),
self.id(3, 0, 1),
self.id(2, R - 1, 0),
self.id(2, R - 1, 1),
self.id(5, 0, R - 1),
self.id(5, 1, R - 1)),
# x, y = 0, R-1
(self.id(3, 1, R - 1),
self.id(3, 1, R - 2),
self.id(3, 0, R - 2),
self.id(2, R - 1, R - 2),
self.id(2, R - 1, R - 1),
self.id(0, 0, 0),
self.id(0, 1, 0)),
# x, y = R-1, 0
(self.id(3, R - 2, 0),
self.id(3, R - 2, 1),
self.id(3, R - 1, 1),
self.id(4, 0, 0),
self.id(4, 0, 1),
self.id(5, R - 2, R - 1),
self.id(5, R - 1, R - 1)),
# x, y = R-1, R-1
(self.id(3, R - 2, R - 1),
self.id(3, R - 2, R - 2),
self.id(3, R - 1, R - 2),
self.id(4, 0, R - 2),
self.id(4, 0, R - 1),
self.id(0, R - 2, 0),
self.id(0, R - 1, 0))
),
# root pixel 4
(
# x, y = 0, 0
(self.id(4, 1, 0),
self.id(4, 1, 1),
self.id(4, 0, 1),
self.id(3, R - 1, 0),
self.id(3, R - 1, 1),
self.id(5, R - 1, R - 2),
self.id(5, R - 1, R - 1)),
# x, y = 0, R-1
(self.id(4, 1, R - 1),
self.id(4, 1, R - 2),
self.id(4, 0, R - 2),
self.id(3, R - 1, R - 2),
self.id(3, R - 1, R - 1),
self.id(0, R - 1, 0),
self.id(0, R - 1, 1)),
# x, y = R-1, 0
(self.id(4, R - 2, 0),
self.id(4, R - 2, 1),
self.id(4, R - 1, 1),
self.id(1, 0, 0),
self.id(1, 0, 1),
self.id(5, R - 1, 0),
self.id(5, R - 1, 1)),
# x, y = R-1, R-1
(self.id(4, R - 2, R - 1),
self.id(4, R - 2, R - 2),
self.id(4, R - 1, R - 2),
self.id(1, 0, R - 2),
self.id(1, 0, R - 1),
self.id(0, R - 1, R - 2),
self.id(0, R - 1, R - 1))
),
# root pixel 5
(
# x, y = 0, 0
(self.id(5, 1, 0),
self.id(5, 1, 1),
self.id(5, 0, 1),
self.id(1, R - 2, 0),
self.id(1, R - 1, 0),
self.id(2, 0, 0),
self.id(2, 1, 0)),
# x, y = 0, R-1
(self.id(5, 1, R - 1),
self.id(5, 1, R - 2),
self.id(5, 0, R - 2),
self.id(2, R - 2, 0),
self.id(2, R - 1, 0),
self.id(3, 0, 0),
self.id(3, 1, 0)),
# x, y = R-1, 0
(self.id(5, R - 2, 0),
self.id(5, R - 2, 1),
self.id(5, R - 1, 1),
self.id(1, 0, 0),
self.id(1, 1, 0),
self.id(4, R - 2, 0),
self.id(4, R - 1, 0)),
# x, y = R-1, R-1
(self.id(5, R - 2, R - 1),
self.id(5, R - 2, R - 2),
self.id(5, R - 1, R - 2),
self.id(3, R - 2, 0),
self.id(3, R - 1, 0),
self.id(4, 0, 0),
self.id(4, 1, 0))
),
)
def verifySkyTiles(cursor, qsp, kind, wcsMap, wcsList, inButler):
"""Verifies that the sky-tile to CCD mapping in the registry used by inButler
is correct. Also computes statistics on the angular separation between amp
corners pre and post WCS determination.
"""
if kind == 'imsim':
geomPolicy = cameraGeomUtils.getGeomPolicy(
pexPolicy.Policy.createPolicy(
pexPolicy.DefaultPolicyFile("obs_lsstSim", "Full_STA_geom.paf",
"description")))
else:
geomPolicy = cameraGeomUtils.getGeomPolicy(
pexPolicy.Policy.createPolicy(
pexPolicy.DefaultPolicyFile("obs_cfht",
"Full_Megacam_geom.paf",
"megacam/description")))
camera = cameraGeomUtils.makeCamera(geomPolicy)
ampDiskLayout = {}
for p in geomPolicy.get("CcdDiskLayout").getArray("Amp"):
ampDiskLayout[p.get("serial")] = (p.get("flipLR"), p.get("flipTB"))
actualTiles = set(wcsMap.keys())
processedTiles = set(inButler.queryMetadata("raw", "skyTile"))
emptyTiles = processedTiles - actualTiles
missedTiles = actualTiles - processedTiles
if len(emptyTiles) > 0:
print dedent("""\
%d tiles not overlapping any Science CCDs (calexps) were found. This can
happen if the run didn't span all the input amps in the registry, or
when a raw amp gets shifted off a tile after WCS determination. So this
is normal, but note that empty sky-tiles should not contain any
SourceAssoc output.""" % len(emptyTiles))
for tile in emptyTiles:
print "\t%d" % tile
if len(missedTiles) > 0:
print dedent("""\
%d tiles overlapping Science CCDs (calexps) were not processed.
This indicates the raw amp padding radius used by the registry
creation script is too small!""" % len(missedTiles))
for tile in missedTiles:
print "\t%d" % tile
print "----"
missedAmps = set()
for tile in actualTiles:
tileGeom = qsp.getGeometry(tile)
tileAmps = set()
if kind == "imsim":
for visit, raft, sensor, channel in inButler.queryMetadata(
"raw",
"channel", ("visit", "raft", "sensor", "channel"),
skyTile=tile,
snap=0):
s1, comma, s2 = sensor
c1, comma, c2 = channel
raftNum = lsstSimRafts.index(raft)
ccdNum = int(s1) * 3 + int(s2)
ampNum = (int(c2) << 3) + int(c1)
tileAmps.add((long(visit), raftNum, ccdNum, ampNum))
else:
for visit, ccd, amp in inButler.queryMetadata(
"raw", "amp", ("visit", "ccd", "amp"), skyTile=tile):
tileAmps.add((long(visit), 0, int(ccd), int(amp)))
for ccdData in wcsMap[tile]:
amps = getAmps(ccdData, camera, ampDiskLayout, cursor, kind)
for ampData in amps:
sciCoords = [
ccdData[0].pixelToSky(c[0], c[1]) for c in ampData[2]
]
# find amps that should have been included in tile
ampSerial = ampData[0].getId().getSerial()
if (tile in missedTiles
or not (ccdData[2], ccdData[3], ccdData[4], ampSerial)
in tileAmps):
if tileGeom.intersects(coordsToPolygon(sciCoords)):
print(
"Tile %d missing visit %s raft %s ccd %s amp %d" %
(tile, str(ccdData[1]), str(
ccdData[2]), str(ccdData[3]), ampSerial))
missedAmps.add((ccdData[1], ampSerial))
print "----"
print "Computing statistics on angular separation between"
print "science corners and raw corners/edges..."
nSamples = 0
maxDist, maxEdgeDist, maxMissEdgeDist = 0.0, 0.0, 0.0
meanDist, meanEdgeDist, meanMissEdgeDist = 0.0, 0.0, 0.0
minDist, minEdgeDist, minMissEdgeDist = 180.0, 180.0, 180.0
for ccdData in wcsList:
amps = getAmps(ccdData, camera, ampDiskLayout, cursor, kind)
for ampData in amps:
sciCoords = [ccdData[0].pixelToSky(c[0], c[1]) for c in ampData[2]]
rawCoords = [ampData[1].pixelToSky(c[0], c[1]) for c in ampData[3]]
rawPoly = coordsToPolygon(rawCoords)
v = rawPoly.getVertices()
e = rawPoly.getEdges()
for raw, sci in izip(rawCoords, sciCoords):
sciPos = tuple(sci.getVector())
dist = geom.cartesianAngularSep(tuple(raw.getVector()),
tuple(sciPos))
minDist = min(dist, minDist)
maxDist = max(dist, maxDist)
meanDist += dist
edgeDist = 180.0
for i in xrange(len(v)):
edgeDist = min(
edgeDist, geom.minEdgeSep(sciPos, e[i], v[i - 1],
v[i]))
minEdgeDist = min(edgeDist, minEdgeDist)
maxEdgeDist = max(edgeDist, maxEdgeDist)
meanEdgeDist += edgeDist
nSamples += 1
if (ccdData[1], ampData[0].getId().getSerial()) in missedAmps:
minMissEdgeDist = min(edgeDist, minMissEdgeDist)
maxMissEdgeDist = max(edgeDist, maxMissEdgeDist)
meanMissEdgeDist += edgeDist
meanDist /= nSamples
meanEdgeDist /= nSamples
print "----"
print "Number of amps examined: %d" % nSamples
print "Min raw to science corner distance: %.6f deg" % minDist
print "Max raw to science corner distance: %.6f deg" % maxDist
print "Mean raw to science corner distance: %.6f deg" % meanDist
print "Min min science corner to raw amp edge distance: %.6f deg" % minEdgeDist
print "Max min science corner to raw amp edge distance: %.6f deg" % maxEdgeDist
print "Mean min science corner to raw amp edge distance: %.6f deg" % meanEdgeDist
print "Number of amps missed: %d" % len(missedAmps)
if len(missedAmps) > 0:
meanMissEdgeDist /= len(missedAmps)
print "Min min science corner to missed raw amp edge distance: %.6f deg" % minMissEdgeDist
print "Max min science corner to missed raw amp edge distance: %.6f deg" % maxMissEdgeDist
print "Mean min science corner to missed raw amp edge distance: %.6f deg" % meanMissEdgeDist
def verifySkyTiles(cursor, qsp, kind, wcsMap, wcsList, inButler):
"""Verifies that the sky-tile to CCD mapping in the registry used by inButler
is correct. Also computes statistics on the angular separation between amp
corners pre and post WCS determination.
"""
if kind == 'imsim':
geomPolicy = cameraGeomUtils.getGeomPolicy(pexPolicy.Policy.createPolicy(
pexPolicy.DefaultPolicyFile("obs_lsstSim", "Full_STA_geom.paf", "description")))
else:
geomPolicy = cameraGeomUtils.getGeomPolicy(pexPolicy.Policy.createPolicy(
pexPolicy.DefaultPolicyFile("obs_cfht", "Full_Megacam_geom.paf", "megacam/description")))
camera = cameraGeomUtils.makeCamera(geomPolicy)
ampDiskLayout = {}
for p in geomPolicy.get("CcdDiskLayout").getArray("Amp"):
ampDiskLayout[p.get("serial")] = (p.get("flipLR"), p.get("flipTB"))
actualTiles = set(wcsMap.keys())
processedTiles = set(inButler.queryMetadata("raw", "skyTile"))
emptyTiles = processedTiles - actualTiles
missedTiles = actualTiles - processedTiles
if len(emptyTiles) > 0:
print dedent("""\
%d tiles not overlapping any Science CCDs (calexps) were found. This can
happen if the run didn't span all the input amps in the registry, or
when a raw amp gets shifted off a tile after WCS determination. So this
is normal, but note that empty sky-tiles should not contain any
SourceAssoc output.""" % len(emptyTiles))
for tile in emptyTiles:
print "\t%d" % tile
if len(missedTiles) > 0:
print dedent("""\
%d tiles overlapping Science CCDs (calexps) were not processed.
This indicates the raw amp padding radius used by the registry
creation script is too small!""" % len(missedTiles))
for tile in missedTiles:
print "\t%d" % tile
print "----"
missedAmps = set()
for tile in actualTiles:
tileGeom = qsp.getGeometry(tile)
tileAmps = set()
if kind == "imsim":
for visit, raft, sensor, channel in inButler.queryMetadata(
"raw", "channel", ("visit", "raft", "sensor", "channel"),
skyTile=tile, snap=0):
s1, comma, s2 = sensor
c1, comma, c2 = channel
raftNum = lsstSimRafts.index(raft)
ccdNum = int(s1) * 3 + int(s2)
ampNum = (int(c2) << 3) + int(c1)
tileAmps.add((long(visit), raftNum, ccdNum, ampNum))
else:
for visit, ccd, amp in inButler.queryMetadata(
"raw", "amp", ("visit", "ccd", "amp"), skyTile=tile):
tileAmps.add((long(visit), 0, int(ccd), int(amp)))
for ccdData in wcsMap[tile]:
amps = getAmps(ccdData, camera, ampDiskLayout, cursor, kind)
for ampData in amps:
sciCoords = [ccdData[0].pixelToSky(c[0], c[1]) for c in ampData[2]]
# find amps that should have been included in tile
ampSerial = ampData[0].getId().getSerial()
if (tile in missedTiles or not
(ccdData[2], ccdData[3], ccdData[4], ampSerial) in tileAmps):
if tileGeom.intersects(coordsToPolygon(sciCoords)):
print ("Tile %d missing visit %s raft %s ccd %s amp %d" %
(tile, str(ccdData[1]), str(ccdData[2]), str(ccdData[3]), ampSerial))
missedAmps.add((ccdData[1], ampSerial))
print "----"
print "Computing statistics on angular separation between"
print "science corners and raw corners/edges..."
nSamples = 0
maxDist, maxEdgeDist, maxMissEdgeDist = 0.0, 0.0, 0.0
meanDist, meanEdgeDist, meanMissEdgeDist = 0.0, 0.0, 0.0
minDist, minEdgeDist, minMissEdgeDist = 180.0, 180.0, 180.0
for ccdData in wcsList:
amps = getAmps(ccdData, camera, ampDiskLayout, cursor, kind)
for ampData in amps:
sciCoords = [ccdData[0].pixelToSky(c[0], c[1]) for c in ampData[2]]
rawCoords = [ampData[1].pixelToSky(c[0], c[1]) for c in ampData[3]]
rawPoly = coordsToPolygon(rawCoords)
v = rawPoly.getVertices()
e = rawPoly.getEdges()
for raw, sci in izip(rawCoords, sciCoords):
sciPos = tuple(sci.getVector())
dist = geom.cartesianAngularSep(tuple(raw.getVector()),
tuple(sciPos))
minDist = min(dist, minDist)
maxDist = max(dist, maxDist)
meanDist += dist
edgeDist = 180.0
for i in xrange(len(v)):
edgeDist = min(edgeDist, geom.minEdgeSep(sciPos, e[i], v[i - 1], v[i]))
minEdgeDist = min(edgeDist, minEdgeDist)
maxEdgeDist = max(edgeDist, maxEdgeDist)
meanEdgeDist += edgeDist
nSamples += 1
if (ccdData[1], ampData[0].getId().getSerial()) in missedAmps:
minMissEdgeDist = min(edgeDist, minMissEdgeDist)
maxMissEdgeDist = max(edgeDist, maxMissEdgeDist)
meanMissEdgeDist += edgeDist
meanDist /= nSamples
meanEdgeDist /= nSamples
print "----"
print "Number of amps examined: %d" % nSamples
print "Min raw to science corner distance: %.6f deg" % minDist
print "Max raw to science corner distance: %.6f deg" % maxDist
print "Mean raw to science corner distance: %.6f deg" % meanDist
print "Min min science corner to raw amp edge distance: %.6f deg" % minEdgeDist
print "Max min science corner to raw amp edge distance: %.6f deg" % maxEdgeDist
print "Mean min science corner to raw amp edge distance: %.6f deg" % meanEdgeDist
print "Number of amps missed: %d" % len(missedAmps)
if len(missedAmps) > 0:
meanMissEdgeDist /= len(missedAmps)
print "Min min science corner to missed raw amp edge distance: %.6f deg" % minMissEdgeDist
print "Max min science corner to missed raw amp edge distance: %.6f deg" % maxMissEdgeDist
print "Mean min science corner to missed raw amp edge distance: %.6f deg" % meanMissEdgeDist
def getAllSipWcs(cursor, qsp, kind):
"""Constructs a Wcs object from each entry in the Science_Ccd_Exposure
table. SIP distortion parameters are read in from
Science_Ccd_Exposure_Metadata. Returns a 3-tuple with the following
contents:
- a mapping from sky-tiles to a list of (WCS, filter) tuples for
overlapping science CCDs
- a list of all (WCS, filter) tuples read in
- a bounding circle (lsst.geom.SphericalCircle) for all science CCDs.
"""
wcsMap = {}
wcsList = []
offset = 0
centers = []
radius = 0.0
n = 0
blocksize = 1000
# An alternative would be to create WCSes from FITS metadata or FITS
# files themselves, but these aren't always available.
while True:
cursor.execute(
"""SELECT scienceCcdExposureId, filterId,
raDeSys, equinox, ctype1, ctype2,
crpix1, crpix2, crval1, crval2,
cd1_1, cd1_2, cd2_1, cd2_2
FROM Science_Ccd_Exposure LIMIT %d, %d
""" % (offset, blocksize))
rows = cursor.fetchall()
if len(rows) == 0:
break
# For each CCD in the results, build up a metadata PropertySet
# including SIP distortion terms and use it to obtain a WCS.
for row in rows:
ps = dafBase.PropertySet()
ps.setString("RADESYS", row[2])
ps.setDouble("EQUINOX", row[3])
ps.setString("CTYPE1", row[4])
ps.setString("CTYPE2", row[5])
ps.setString("CUNIT1", "deg")
ps.setString("CUNIT2", "deg")
ps.setDouble("CRPIX1", row[6])
ps.setDouble("CRPIX2", row[7])
ps.setDouble("CRVAL1", row[8])
ps.setDouble("CRVAL2", row[9])
ps.setDouble("CD1_1", row[10])
ps.setDouble("CD1_2", row[11])
ps.setDouble("CD2_1", row[12])
ps.setDouble("CD2_2", row[13])
if row[4].endswith("-SIP") and row[5].endswith("-SIP"):
cursor.execute("""
SELECT metadataKey, intValue
FROM Science_Ccd_Exposure_Metadata
WHERE scienceCcdExposureId = %d AND
intValue IS NOT NULL AND
metadataKey RLIKE '^[AB]P?_ORDER$'
""" % row[0])
for k, v in cursor.fetchall():
ps.setInt(k, v)
cursor.execute("""
SELECT metadataKey, doubleValue
FROM Science_Ccd_Exposure_Metadata
WHERE scienceCcdExposureId = %d AND
doubleValue IS NOT NULL AND
metadataKey RLIKE '^[AB]P?_[0-9]+_[0-9]+$'
""" % row[0])
for k, v in cursor.fetchall():
ps.setDouble(k, v)
wcs = afwImage.makeWcs(ps)
wcsList.append([wcs, row[1]])
if kind == 'imsim':
poly = qs.imageToPolygon(wcs, 4072.0, 4000.0, 0.0)
else:
poly = qs.imageToPolygon(wcs, 2048.0, 4610.0, 0.0)
bc = poly.getBoundingCircle()
radius = max(radius, bc.getRadius())
centers.append(geom.cartesianUnitVector(bc.getCenter()))
pix = qsp.intersect(poly)
for skyTile in pix:
if skyTile in wcsMap:
wcsMap[skyTile].append([wcs, row[1]])
else:
wcsMap[skyTile] = [[wcs, row[1]]]
n = offset + len(rows)
print "... %d" % n
offset += blocksize
x, y, z = 0.0, 0.0, 0.0
for v in centers:
x += v[0]; y += v[1]; z += v[2]
cen = geom.normalize((x, y, z))
r = max(geom.cartesianAngularSep(cen, v) for v in centers) + 2.0*radius
return wcsMap, wcsList, geom.SphericalCircle(cen, r)