Closed piyanatk closed 1 year ago
@piyanatk
I tried to replicate your problem in a test, but it passed:
def test_freq_parser_1freq_channel_width():
input_freq_dict = {
"Nfreqs": 1,
"channel_width": 50e3,
"start_freq": 150e6,
}
output_freq_dict = pyuvsim.parse_frequency_params(input_freq_dict)
expected_freq_dict = {
"Nfreqs": 1,
"freq_array": [150e6],
"channel_width": [50e3],
"Nspws": 1,
}
assert output_freq_dict == expected_freq_dict
Is this similar to what you're setting? What version are you running?
@bhazelton I checked again, and it looks like the issue is actually _complete_uvdata
replacing the channel_width
attribute. This was done in hera_sim
to prepare the uvdata object for visibility simulation.
https://github.com/HERA-Team/hera_sim/blob/9576e78bd8807473c40152f9b6b61745ca9dc879/hera_sim/visibilities/simulators.py#L168-L184
@piyanatk Are you sure you're specifying the channel width in your parameter file?
I expanded my test to include initialize_uvdata_from_params
and _complete_uvdata
and I'm still getting that the channel_width
is what I set it to, not 1.
I'll note that there is no logic related to channel_width
in either initialize_uvdata_from_params
or _complete_uvdata
, so I didn't think they would be related but I checked them anyway.
def test_freq_parser_1freq_channel_width():
input_freq_dict = {
"Nfreqs": 1,
"channel_width": 50e3,
"start_freq": 150e6,
}
output_freq_dict = pyuvsim.parse_frequency_params(input_freq_dict)
expected_freq_dict = {
"Nfreqs": 1,
"freq_array": [150e6],
"channel_width": [50e3],
"Nspws": 1,
}
assert output_freq_dict == expected_freq_dict
# Now make a UVData object to check its channel_width
param_filename = os.path.join(SIM_DATA_PATH, "test_config", "param_10time_10chan_0.yaml")
param_dict = pyuvsim.simsetup._config_str_to_dict(param_filename)
freq_keywords = [
'freq_array', 'start_freq', 'end_freq', 'Nfreqs', 'channel_width', 'bandwidth'
]
for key in freq_keywords:
param_dict["freq"].pop(key, None)
for key, value in input_freq_dict.items():
param_dict["freq"][key] = value
assert param_dict["freq"]["channel_width"] == input_freq_dict["channel_width"]
uvd, _, _ = pyuvsim.initialize_uvdata_from_params(param_dict)
assert uvd.channel_width == np.asarray([input_freq_dict["channel_width"]])
assert uvd.freq_array == expected_freq_dict["freq_array"]
uvd2 = pyuvsim.simsetup._complete_uvdata(uvd)
assert uvd2.channel_width == np.asarray([input_freq_dict["channel_width"]])
assert uvd2.freq_array == expected_freq_dict["freq_array"]
@bhazelton Yes, I am very sure. My test was not very sophisticated, but please see the attached screen shot for what I did.
@piyanatk Thanks, that was helpful. I saw that you were using v1.2.1 (as opposed to the latest main branch), so I tested on that version and I see the error. So I think that if you get the latest main it will fix it, but we should also make a new release.
@bhazelton, using the main branch indeed fixed the problem. Thank you!
Fixed in recent version
initialize_uvdata_from_params
will always return a UVData object withchannel_width=1.0
when the configuration file hasNfreqs
set to 1 regardless of the actualchannel_width
.