def test_phasing_funcs():
# these tests are based on a notebook where I tested against the mwa_tools phasing code
ra_hrs = 12.1
dec_degs = -42.3
mjd = 55780.1
array_center_xyz = np.array([-2559454.08, 5095372.14, -2849057.18])
lat_lon_alt = uvutils.LatLonAlt_from_XYZ(array_center_xyz)
obs_time = Time(mjd,
format='mjd',
location=(lat_lon_alt[1], lat_lon_alt[0]))
icrs_coord = SkyCoord(ra=Angle(ra_hrs, unit='hr'),
dec=Angle(dec_degs, unit='deg'),
obstime=obs_time)
gcrs_coord = icrs_coord.transform_to('gcrs')
# in east/north/up frame (relative to array center) in meters: (Nants, 3)
ants_enu = np.array([-101.94, 0156.41, 0001.24])
ant_xyz_abs = uvutils.ECEF_from_ENU(ants_enu, lat_lon_alt[0],
lat_lon_alt[1], lat_lon_alt[2])
ant_xyz_rel_itrs = ant_xyz_abs - array_center_xyz
ant_xyz_rel_rot = uvutils.rotECEF_from_ECEF(ant_xyz_rel_itrs,
lat_lon_alt[1])
array_center_coord = SkyCoord(x=array_center_xyz[0] * units.m,
y=array_center_xyz[1] * units.m,
z=array_center_xyz[2] * units.m,
frame='itrs',
obstime=obs_time)
itrs_coord = SkyCoord(x=ant_xyz_abs[0] * units.m,
y=ant_xyz_abs[1] * units.m,
z=ant_xyz_abs[2] * units.m,
frame='itrs',
obstime=obs_time)
gcrs_array_center = array_center_coord.transform_to('gcrs')
gcrs_from_itrs_coord = itrs_coord.transform_to('gcrs')
gcrs_rel = (gcrs_from_itrs_coord.cartesian -
gcrs_array_center.cartesian).get_xyz().T
gcrs_uvw = uvutils.phase_uvw(gcrs_coord.ra.rad, gcrs_coord.dec.rad,
gcrs_rel.value)
mwa_tools_calcuvw_u = -97.122828
mwa_tools_calcuvw_v = 50.388281
mwa_tools_calcuvw_w = -151.27976
assert np.allclose(gcrs_uvw[0, 0], mwa_tools_calcuvw_u, atol=1e-3)
assert np.allclose(gcrs_uvw[0, 1], mwa_tools_calcuvw_v, atol=1e-3)
assert np.allclose(gcrs_uvw[0, 2], mwa_tools_calcuvw_w, atol=1e-3)
# also test unphasing
temp2 = uvutils.unphase_uvw(gcrs_coord.ra.rad, gcrs_coord.dec.rad,
np.squeeze(gcrs_uvw))
assert np.allclose(gcrs_rel.value, temp2)
def apply_bda(
uv, max_decorr, pre_fs_int_time, corr_fov_angle, max_time, corr_int_time=None
):
"""Apply baseline dependent averaging to a UVData object.
For each baseline in the UVData object, the expected decorrelation from
averaging in time is computed. Baselines are averaged together in powers-
of-two until the specified level of decorrelation is reached (rounded down).
Parameters
----------
uv : UVData object
The UVData object to apply BDA to. No changes are made to this object,
and instead a copy is returned.
max_decorr : float
The maximum decorrelation fraction desired in the output object. Must
be between 0 and 1.
pre_fs_int_time : astropy Quantity
The pre-finge-stopping integration time inside of the correlator. The
quantity should be compatible with units of time.
corr_fov_angle : astropy Angle
The opening angle at which the maximum decorrelation is to be
calculated. Because a priori it is not known in which direction the
decorrelation will be largest, the expected decorrelation is computed in
all 4 cardinal directions at `corr_fov_angle` degrees off of zenith,
and the largest one is used. This is a "worst case scenario"
decorrelation.
max_time : astropy Quantity
The maximum amount of time that spectra from different times should be
combined for. The ultimate integration time for a given baseline will be
for max_time or the integration time that is smaller than the specified
decorrelation level, whichever is smaller. The quantity should be
compatible with units of time.
corr_int_time : astropy Quantity, optional
The output time of the correlator. If not specified, the smallest
integration_time in the UVData object is used. If specified, the
quantity should be compatible with units of time.
Returns
-------
uv2 : UVData object
The UVData object with BDA applied.
Raises
------
ValueError
This is raised if the input parameters are not the appropriate type or
in the appropriate range. It is also raised if the input UVData object
is not in drift mode (the BDA code does rephasing within an averaged
set of baselines).
AssertionError
This is raised if the baselines of the UVData object are not time-
ordered.
"""
if not isinstance(uv, UVData):
raise ValueError(
"apply_bda must be passed a UVData object as its first argument"
)
if not isinstance(corr_fov_angle, Angle):
raise ValueError(
"corr_fov_angle must be an Angle object from astropy.coordinates"
)
if not isinstance(pre_fs_int_time, units.Quantity):
raise ValueError("pre_fs_int_time must be an astropy.units.Quantity")
try:
pre_fs_int_time.to(units.s)
except UnitConversionError:
raise ValueError("pre_fs_int_time must be a Quantity with units of time")
if (
corr_fov_angle.to(units.deg).value < 0
or corr_fov_angle.to(units.deg).value > 90
):
raise ValueError("corr_fov_angle must be between 0 and 90 degrees")
if max_decorr < 0 or max_decorr > 1:
raise ValueError("max_decorr must be between 0 and 1")
if not isinstance(max_time, units.Quantity):
raise ValueError("max_time must be an astropy.units.Quantity")
try:
max_time.to(units.s)
except UnitConversionError:
raise ValueError("max_time must be a Quantity with units of time")
if corr_int_time is None:
# assume the correlator integration time is the smallest int_time of the
# UVData object
corr_int_time = np.unique(uv.integration_time)[0] * units.s
else:
if not isinstance(corr_int_time, units.Quantity):
raise ValueError("corr_int_time must be an astropy.units.Quantity")
try:
corr_int_time.to(units.s)
except UnitConversionError:
raise ValueError("corr_int_time must be a Quantity with units of time")
if uv.phase_type != "drift":
raise ValueError("UVData object must be in drift mode to apply BDA")
# get relevant bits of metadata
freq = np.amax(uv.freq_array[0, :]) * units.Hz
chan_width = uv.channel_width * units.Hz
antpos_enu, ants = uv.get_ENU_antpos()
lat, lon, alt = uv.telescope_location_lat_lon_alt
antpos_ecef = uvutils.ECEF_from_ENU(antpos_enu, lat, lon, alt)
telescope_location = EarthLocation.from_geocentric(
uv.telescope_location[0],
uv.telescope_location[1],
uv.telescope_location[2],
unit="m",
)
# make a new UVData object to put BDA baselines in
uv2 = UVData()
# copy over metadata
uv2.Nbls = uv.Nbls
uv2.Nfreqs = uv.Nfreqs
uv2.Npols = uv.Npols
uv2.vis_units = uv.vis_units
uv2.Nspws = uv.Nspws
uv2.spw_array = uv.spw_array
uv2.freq_array = uv.freq_array
uv2.polarization_array = uv.polarization_array
uv2.channel_width = uv.channel_width
uv2.object_name = uv.object_name
uv2.telescope_name = uv.telescope_name
uv2.instrument = uv.instrument
uv2.telescope_location = uv.telescope_location
history = uv.history + " Baseline dependent averaging applied."
uv2.history = history
uv2.Nants_data = uv.Nants_data
uv2.Nants_telescope = uv.Nants_telescope
uv2.antenna_names = uv.antenna_names
uv2.antenna_numbers = uv.antenna_numbers
uv2.x_orientation = uv.x_orientation
uv2.extra_keywords = uv.extra_keywords
uv2.antenna_positions = uv.antenna_positions
uv2.antenna_diameters = uv.antenna_diameters
uv2.gst0 = uv.gst0
uv2.rdate = uv.rdate
uv2.earth_omega = uv.earth_omega
uv2.dut1 = uv.dut1
uv2.timesys = uv.timesys
uv2.uvplane_reference_time = uv.uvplane_reference_time
# initialize place-keeping variables and Nblt-sized metadata
start_index = 0
uv2.Nblts = 0
uv2.uvw_array = np.zeros_like(uv.uvw_array)
uv2.time_array = np.zeros_like(uv.time_array)
uv2.lst_array = np.zeros_like(uv.lst_array)
uv2.ant_1_array = np.zeros_like(uv.ant_1_array)
uv2.ant_2_array = np.zeros_like(uv.ant_2_array)
uv2.baseline_array = np.zeros_like(uv.baseline_array)
uv2.integration_time = np.zeros_like(uv.integration_time)
uv2.data_array = np.zeros_like(uv.data_array)
uv2.flag_array = np.zeros_like(uv.flag_array)
uv2.nsample_array = np.zeros_like(uv.nsample_array)
# iterate over baselines
for key in uv.get_antpairs():
print("averaging baseline ", key)
ind1, ind2, indp = uv._key2inds(key)
if len(ind2) != 0:
raise AssertionError(
"ind2 from _key2inds() is not 0--exiting. This should not happen, "
"please contact the package maintainers."
)
data = uv._smart_slicing(
uv.data_array, ind1, ind2, indp, squeeze="none", force_copy=True
)
flags = uv._smart_slicing(
uv.flag_array, ind1, ind2, indp, squeeze="none", force_copy=True
)
nsamples = uv._smart_slicing(
uv.nsample_array, ind1, ind2, indp, squeeze="none", force_copy=True
)
# get lx and ly for baseline
ant1 = np.where(ants == key[0])[0][0]
ant2 = np.where(ants == key[1])[0][0]
x1, y1, z1 = antpos_ecef[ant1, :]
x2, y2, z2 = antpos_ecef[ant2, :]
lx = np.abs(x2 - x1) * units.m
ly = np.abs(y2 - y1) * units.m
# figure out how many time samples we can combine together
if key[0] == key[1]:
# autocorrelation--don't average
n_two_foldings = 0
else:
n_two_foldings = dc.bda_compression_factor(
max_decorr,
freq,
lx,
ly,
corr_fov_angle,
chan_width,
pre_fs_int_time,
corr_int_time,
)
# convert from max_time to max_samples
max_samples = (max_time / corr_int_time).to(units.dimensionless_unscaled)
max_two_foldings = int(np.floor(np.log2(max_samples)))
n_two_foldings = min(n_two_foldings, max_two_foldings)
n_int = 2 ** (n_two_foldings)
print("averaging {:d} time samples...".format(n_int))
# figure out how many output samples we're going to have
n_in = len(ind1)
n_out = n_in // n_int + min(1, n_in % n_int)
# get relevant metdata
uvw_array = uv.uvw_array[ind1, :]
times = uv.time_array[ind1]
if not np.all(times == np.sort(times)):
raise AssertionError(
"times of uvdata object are not monotonically increasing; "
"throwing our hands up"
)
lsts = uv.lst_array[ind1]
int_time = uv.integration_time[ind1]
# do the averaging
input_shape = data.shape
assert input_shape == (n_in, 1, uv.Nfreqs, uv.Npols)
output_shape = (n_out, 1, uv.Nfreqs, uv.Npols)
data_out = np.empty(output_shape, dtype=np.complex128)
flags_out = np.empty(output_shape, dtype=np.bool_)
nsamples_out = np.empty(output_shape, dtype=np.float32)
uvws_out = np.empty((n_out, 3), dtype=np.float64)
times_out = np.empty((n_out,), dtype=np.float64)
lst_out = np.empty((n_out,), dtype=np.float64)
int_time_out = np.empty((n_out,), dtype=np.float64)
if n_out == n_in:
# we don't need to average
current_index = start_index + n_out
uv2.data_array[start_index:current_index, :, :, :] = data
uv2.flag_array[start_index:current_index, :, :, :] = flags
uv2.nsample_array[start_index:current_index, :, :, :] = nsamples
uv2.uvw_array[start_index:current_index, :] = uvw_array
uv2.time_array[start_index:current_index] = times
uv2.lst_array[start_index:current_index] = lsts
uv2.integration_time[start_index:current_index] = int_time
uv2.ant_1_array[start_index:current_index] = key[0]
uv2.ant_2_array[start_index:current_index] = key[1]
uv2.baseline_array[start_index:current_index] = uvutils.antnums_to_baseline(
ant1, ant2, None
)
start_index = current_index
else:
# rats, we actually have to do work...
# phase up the data along each chunk of times
for i in range(n_out):
# compute zenith of the desired output time
i1 = i * n_int
i2 = min((i + 1) * n_int, n_in)
assert i2 - i1 > 0
t0 = Time((times[i1] + times[i2 - 1]) / 2, scale="utc", format="jd")
zenith_coord = SkyCoord(
alt=Angle(90 * units.deg),
az=Angle(0 * units.deg),
obstime=t0,
frame="altaz",
location=telescope_location,
)
obs_zenith_coord = zenith_coord.transform_to("icrs")
zenith_ra = obs_zenith_coord.ra
zenith_dec = obs_zenith_coord.dec
# get data, flags, and nsamples of slices
data_chunk = data[i1:i2, :, :, :]
flags_chunk = flags[i1:i2, :, :, :]
nsamples_chunk = nsamples[i1:i2, :, :, :]
# actually phase now
# compute new uvw coordinates
icrs_coord = SkyCoord(
ra=zenith_ra, dec=zenith_dec, unit="radian", frame="icrs"
)
uvws = np.float64(uvw_array[i1:i2, :])
itrs_telescope_location = SkyCoord(
x=uv.telescope_location[0] * units.m,
y=uv.telescope_location[1] * units.m,
z=uv.telescope_location[2] * units.m,
representation_type="cartesian",
frame="itrs",
obstime=t0,
)
itrs_lat_lon_alt = uv.telescope_location_lat_lon_alt
frame_telescope_location = itrs_telescope_location.transform_to("icrs")
frame_telescope_location.representation_type = "cartesian"
uvw_ecef = uvutils.ECEF_from_ENU(uvws, *itrs_lat_lon_alt)
itrs_uvw_coord = SkyCoord(
x=uvw_ecef[:, 0] * units.m,
y=uvw_ecef[:, 1] * units.m,
z=uvw_ecef[:, 2] * units.m,
representation_type="cartesian",
frame="itrs",
obstime=t0,
)
frame_uvw_coord = itrs_uvw_coord.transform_to("icrs")
frame_rel_uvw = (
frame_uvw_coord.cartesian.get_xyz().value.T
- frame_telescope_location.cartesian.get_xyz().value
)
new_uvws = uvutils.phase_uvw(
icrs_coord.ra.rad, icrs_coord.dec.rad, frame_rel_uvw
)
# average these uvws together to get the "average" position in
# the uv-plane
avg_uvws = np.average(new_uvws, axis=0)
# calculate and apply phasor
w_lambda = (
new_uvws[:, 2].reshape((i2 - i1), 1)
/ const.c.to("m/s").value
* uv.freq_array.reshape(1, uv.Nfreqs)
)
phs = np.exp(-1j * 2 * np.pi * w_lambda[:, None, :, None])
data_chunk *= phs
# sum data, propagate flag array, and adjusting nsample accordingly
data_slice = np.sum(data_chunk, axis=0)
flag_slice = np.sum(flags_chunk, axis=0)
nsamples_slice = np.sum(nsamples_chunk, axis=0) / (i2 - i1)
data_out[i, :, :, :] = data_slice
flags_out[i, :, :, :] = flag_slice
nsamples_out[i, :, :, :] = nsamples_slice
# update metadata
uvws_out[i, :] = avg_uvws
times_out[i] = (times[i1] + times[i2 - 1]) / 2
lst_out[i] = (lsts[i1] + lsts[i2 - 1]) / 2
int_time_out[i] = np.average(int_time[i1:i2]) * (i2 - i1)
# update data and metadata when we're done with this baseline
current_index = start_index + n_out
uv2.data_array[start_index:current_index, :, :, :] = data_out
uv2.flag_array[start_index:current_index, :, :, :] = flags_out
uv2.nsample_array[start_index:current_index, :, :, :] = nsamples_out
uv2.uvw_array[start_index:current_index, :] = uvws_out
uv2.time_array[start_index:current_index] = times_out
uv2.lst_array[start_index:current_index] = lst_out
uv2.integration_time[start_index:current_index] = int_time_out
uv2.ant_1_array[start_index:current_index] = key[0]
uv2.ant_2_array[start_index:current_index] = key[1]
uv2.baseline_array[start_index:current_index] = uvutils.antnums_to_baseline(
ant1, ant2, None
)
start_index = current_index
# clean up -- shorten all arrays to actually be size nblts
nblts = start_index
uv2.Nblts = nblts
uv2.data_array = uv2.data_array[:nblts, :, :, :]
uv2.flag_array = uv2.flag_array[:nblts, :, :, :]
uv2.nsample_array = uv2.nsample_array[:nblts, :, :, :]
uv2.uvw_array = uv2.uvw_array[:nblts, :]
uv2.time_array = uv2.time_array[:nblts]
uv2.lst_array = uv2.lst_array[:nblts]
uv2.integration_time = uv2.integration_time[:nblts]
uv2.ant_1_array = uv2.ant_1_array[:nblts]
uv2.ant_2_array = uv2.ant_2_array[:nblts]
uv2.baseline_array = uv2.baseline_array[:nblts]
uv2.Ntimes = len(np.unique(uv2.time_array))
# set phasing info
uv2.phase_type = "phased"
uv2.phase_center_ra = zenith_ra.rad
uv2.phase_center_dec = zenith_dec.rad
uv2.phase_center_frame = 2000.0
# fix up to correct old phasing method
uv2.phase(zenith_ra.rad, zenith_dec.rad, epoch="J2000", fix_old_proj=True)
# run a check
uv2.check()
return uv2
def phase(jd, ra, dec, telescope_location, uvws0):
"""
Compute UVWs phased to a given RA/DEC at a particular epoch.
This function was copied from the pyuvdata.UVData.phase method, and modified to be
simpler.
Parameters
----------
jd : float or array_like of float
The Julian date of the observation.
ra : float
The ra to phase to in radians.
dec : float
The dec to phase to in radians.
telescope_location : :class:`astropy.coordinates.EarthLocation`
The location of the reference point of the telescope, in geodetic frame (i.e.
it has lat, lon, height attributes).
uvws0 : array
The UVWs when phased to zenith.
Returns
-------
uvws : array
Array of the same shape as `uvws0`, with entries modified to the new phase
center.
"""
frame_phase_center = SkyCoord(ra=ra, dec=dec, unit="radian", frame="icrs")
obs_time = Time(np.atleast_1d(jd), format="jd")
itrs_telescope_location = telescope_location.get_itrs(obstime=obs_time)
frame_telescope_location = itrs_telescope_location.transform_to(ICRS)
frame_telescope_location.representation_type = "cartesian"
uvw_ecef = uvutils.ECEF_from_ENU(
uvws0,
telescope_location.lat.rad,
telescope_location.lon.rad,
telescope_location.height,
)
unique_times, r_inds = np.unique(obs_time, return_inverse=True)
uvws = np.zeros((uvw_ecef.shape[0], unique_times.size, 3),
dtype=np.float64)
for ind, jd in tqdm.tqdm(
enumerate(unique_times),
desc="computing UVWs",
total=len(unique_times),
unit="times",
disable=not config.PROGRESS or unique_times.size == 1,
):
itrs_uvw_coord = SkyCoord(
x=uvw_ecef[:, 0] * un.m,
y=uvw_ecef[:, 1] * un.m,
z=uvw_ecef[:, 2] * un.m,
frame="itrs",
obstime=jd,
)
frame_uvw_coord = itrs_uvw_coord.transform_to("icrs")
# this takes out the telescope location in the new frame,
# so these are vectors again
frame_rel_uvw = (
frame_uvw_coord.cartesian.get_xyz().value.T -
frame_telescope_location[ind].cartesian.get_xyz().value)
uvws[:, ind, :] = uvutils.phase_uvw(frame_phase_center.ra.rad,
frame_phase_center.dec.rad,
frame_rel_uvw)
return uvws[:, r_inds, :]
frame_telescope_location = frame_telescope_locations[i]
itrs_ant_coord = SkyCoord(
x=antpos_xyz[:, 0] * units.m,
y=antpos_xyz[:, 1] * units.m,
z=antpos_xyz[:, 2] * units.m,
frame="itrs",
obstime=time,
)
frame_ant_coord = itrs_ant_coord.transform_to("icrs")
frame_ant_rel = (
(frame_ant_coord.cartesian - frame_telescope_location.cartesian)
.get_xyz()
.T.value
)
frame_ant_uvw = uvutils.phase_uvw(
frame_phase_center.ra.rad, frame_phase_center.dec.rad, frame_ant_rel
)
delay_values[:, i] = frame_ant_uvw[:, 2] / constants.c.to("m/s").value
dtaus = delay_values[:, -1] - delay_values[:, 0]
assert len(dtaus) == N_ants
print("max delay: ", np.amax(np.abs(dtaus)))
# convert to a dictionary
channel_width = 250e6 / 8192
delay_dict = {}
for i, ant_name in enumerate(ant_names):
# convert to form required by FPGA block
fringe_phase = 2 * np.pi * delay_values[i, :] * channel_width * (180.0 / np.pi)
# make sure angle is between -pi and pi
np.where(fringe_phase > 180, fringe_phase - 360, fringe_phase)