SPR1-1009: Hetero beams #57
Conversation
Retrieve the appropriate primary beam models and store as a parameter to pass to the pipeline for use when predicting visibilities
To address SPR1-1009 this commit uses the primary beam models retireved from telstate to predict model visibilities when a complex (multi-component) sky model is provided. The prediction makes a number of simplifying assumptions to spead up the prediction. These are: 1.) Calibrator field sources are unpolarised (Q,U,V = 0) 2.) Antennas of the same design have the same primary beam 3.) H and V primary beams are identical.
|
thanks we need to chat about formats to store these new models in. This can
only be next week though.
I suggest the Tigger Sky LSM format
…On Wed, Nov 10, 2021 at 2:12 PM KimMcAlpine ***@***.***> wrote:
Enable the pipeline to use primary beam models when predicting model
visibilities. This will allow it to use an intrinsic sky model for
calibrator fields, which can then be modified by the appropriate primary
beam model to produce different apparent sky models for MeerKAT and SKA
dishes.
This cannot be properly tested or merged until we have an appropriate
intrinsic sky model for the calibrator fields. The currently used models
are all apparent sky models (for MeerKAT antennas)
------------------------------
You can view, comment on, or merge this pull request online at:
#57
Commit Summary
- Retrieve primary beam models
<9ebb7f8>
- Predict model visibilities using primary beams
<f18ab50>
- Allow simulator to use primary beam models
<2adcd67>
- Modify unit tests to incorporate primary beams
<7efe6c0>
File Changes
(8 files <https://github.com/ska-sa/katsdpcal/pull/57/files>)
- *M* katsdpcal/calprocs.py
<https://github.com/ska-sa/katsdpcal/pull/57/files#diff-905d72f74e05d52e1546f65d00af33c24e3f181dc111bf4ea6739245910c5114>
(62)
- *M* katsdpcal/pipelineprocs.py
<https://github.com/ska-sa/katsdpcal/pull/57/files#diff-7fa31003c0f3d46c2a9e46af4aeab78251488f33d7cff8294c3fd88631c02633>
(40)
- *M* katsdpcal/reduction.py
<https://github.com/ska-sa/katsdpcal/pull/57/files#diff-3959af217a748793a519d49a58be81d5a6797e7207c6a8f6d8aa5db62aefa66a>
(6)
- *M* katsdpcal/scan.py
<https://github.com/ska-sa/katsdpcal/pull/57/files#diff-9dad9efb3ff86695cde5c2d485d1ef3b5cc1fd6c6d8219965227e61d78eeb29c>
(37)
- *M* katsdpcal/simulator.py
<https://github.com/ska-sa/katsdpcal/pull/57/files#diff-c4c69cf933b453d0f9f66a0fa7b1da4ffd7fa82dbe076ff3b09c66e15b9358c6>
(4)
- *M* katsdpcal/test/test_calprocs.py
<https://github.com/ska-sa/katsdpcal/pull/57/files#diff-926b80f88d143933668cbcb850a304e72766f4a24daa605be1bae1d63358dd78>
(9)
- *M* katsdpcal/test/test_control.py
<https://github.com/ska-sa/katsdpcal/pull/57/files#diff-0dbff85efae94800b88b65a25ecdceeb57b45b79886c728c148e443e57e38e3f>
(13)
- *M* katsdpcal/test/test_pipelineprocs.py
<https://github.com/ska-sa/katsdpcal/pull/57/files#diff-4ce1854afb9bc830ab9cfff8f3cdb2e1e704786a0e0e8299a4b996f3b0fcd928>
(13)
Patch Links:
- https://github.com/ska-sa/katsdpcal/pull/57.patch
- https://github.com/ska-sa/katsdpcal/pull/57.diff
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#57>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AB4RE6WF4ABRWABC7C5HRHTULJOSFANCNFSM5HXXTX6Q>
.
Triage notifications on the go with GitHub Mobile for iOS
<https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675>
or Android
<https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>.
--
--
Benjamin Hugo
PhD. student,
Centre for Radio Astronomy Techniques and Technologies
Department of Physics and Electronics
Rhodes University
Junior software developer
Radio Astronomy Research Group
South African Radio Astronomy Observatory
Black River Business Park
Observatory
Cape Town
|
| @@ -239,6 +246,27 @@ def _check_blank_freqwin(parameters): | |||
| return blank_freqwin | |||
|
|
|||
|
|
|||
| def _get_primary_beam_model(telstate_l0, antenna): | |||
| with katsdpmodels.fetch.requests.Fetcher() as fetcher: | |||
There was a problem hiding this comment.
I thought it would be more convenient to use a TelescopeStateFetcher... but then I realised (as you probably also did) that indexed keys are not so nice in this regard. 🤦🏾
I wonder whether we should extend telstate keys in future to make this more convenient 🙂 We could e.g. base keys on pathlib and then treat the index key as a kind of suffix, or just support tuples of strings.
Then we could pass primary_beam_key + antenna.name to TelescopeStateFetcher.get.
Just a thought, not for now...
There was a problem hiding this comment.
Agreed, it is a bit of a clumsy workaround as it stands now :(
katsdpcal/calprocs.py
Outdated
| pol1 = beam_pol_idx[pol_order[1]] | ||
| beam_phand = np.stack([beam_sample[..., pol0, pol0], beam_sample[..., pol1, pol1]], axis=-1) | ||
| # select only the hh and vv beams, order doesn't matter as we average the two pols | ||
| beam_phand = np.stack([beam_sample[..., 0, 0], beam_sample[..., 1, 1]], axis=-1) |
There was a problem hiding this comment.
A small suggestion: maybe it's better to just add the HH and VV parts together, instead of first assembling them into a new array of twice the size and then averaging that down again in a second step. I'm not sure exactly how large beam_sample is, but this seems more memory efficient and also a bit more readable. E.g.:
beam_phand = 0.5 * (beam_sample[..., 0, 0] + beam_sample[..., 1, 1])Another thought: how does this differ from the UNPOLARIZED_POWER OutputType? Why not use that instead?
There was a problem hiding this comment.
Good suggestion, the current setup it probably a hangover of when I was still doing separate polarisations.
I think the difference here from UNPOLARIZED_POWER is that you retain some phase information, while power is a scalar. Ideally you want to retain this phase information so that for the SKA - MKAT baselines you have an approximately correct phase for the combined beam. For MKAT - MKAT baselines and SKA - SKA baselines the phases of beams will cancel each other out and then it should be indistinguishable from using the UNPOLARIZED_POWER type.
katsdpcal/calprocs.py
Outdated
|
|
||
| def _complex_interp(fp, xp, x): | ||
| return complex_interp(xp, x, fp) | ||
| beam = np.apply_along_axis(_complex_interp, 0, beam_phand, |
There was a problem hiding this comment.
Cool trick - I must remember this!
Since you are already wrapping the function, why not populate the x and xp parameters upfront? For example:
def _complex_interp_over_freq(fp):
return complex_interp(chan_freqs.value, chan_freqs.value[::freq_stride], fp)
beam = np.apply_along_axis(_complex_interp_over_freq, 0, beam_phand)| # sample beam at close to its frequency resolution | ||
| channel_width = chan_freqs[1] - chan_freqs[0] | ||
| freq_stride = int(np.floor(beam_model.frequency_resolution() / channel_width)) | ||
|
|
There was a problem hiding this comment.
Maybe add
sampled_freqs = chan_freqs[::freq_stride]for more clarity, and use below (also with complex_interp).
This is to address SPR1-1009
Enable the pipeline to use primary beam models when predicting model visibilities. This will allow it to use an intrinsic sky model for calibrator fields, which can then be modified by the appropriate primary beam model to produce different apparent sky models for MeerKAT and SKA dishes.
This cannot be properly tested or merged until we have an appropriate intrinsic sky model for the calibrator fields. The currently used models are all apparent sky models (for MeerKAT antennas)