Question about "input.apply_p_jones" and "output.apply_p_jones_inv" functionality #328
Replies: 4 comments 17 replies
-
|
Hi Andrew!
I think I can see why this may be misleading. This is actually referring to the case where you have both columns and skymodels present in the model e.g.
This can get a little complicated so I will try and explain the logic and then you can decide what applies to your use case. The uncalibrated data is in the telescope frame but our model will typically be in the sky frame. Applying parallactic angle to the model moves it into the telescope frame (where we want to do our calibration). This corresponds to I believe that you are on the right track, but hopefully the above makes it a little clearer. Don't hesitate to ask if you need further clarification. |
Beta Was this translation helpful? Give feedback.
-
|
Hi, Ok, that makes sense. For more details about this use case. The current approach is to use CASA to perform the standard "1GC" calibration (i.e., K, Gp, Ga, BP, Df, Xf). But, to my knowledge, CASA cannot explicitly solve for off-diagonal terms when self-calibrating and thus isn't the best for polarized self-calibration. I've had much success using Quartical for polarization self-calibration of VLA data, where I performed 1GC with Casa and split out the CORRECTED_DATA. into a new MS file, then self-calibrated the new MS file with the model produced from initial imaging. But, from what I understand, if I split out CORRECTED_DATA with a parallactic angle correction applied (parang=True in CASA) for a linear-feed instrument (like MeerKAT), I need to be very careful because the P-matrix doesn't commute, as it does for circular feeds. Therefore, there is a risk of "over-correcting" parallactic angle rotation (corrupting EVPA measurements) when performing self-calibration with Quartical on CASA-calibrated 1GC data. I'd imagine I could:
But I'm not 100% sure. If I didn't screw up the explanation (which may have occurred), is there a recommended use case for a scenario like this? Obviously, this is hybridizing CASA and Quartical, and the answer may be to convert to Quartical fully, so I apologize for bringing external software into the problem; I appreciate any help/advice. Andrew |
Beta Was this translation helpful? Give feedback.
-
|
@JSKenyon, it looks like it didn't fix the issue. Here is some extra information,
Andrew |
Beta Was this translation helpful? Give feedback.
-
|
Moreover, If I run
However, this only goes up to 2023-12-31, so I might be S.O.L, given the recency of my observations. Andrew |
Beta Was this translation helpful? Give feedback.
-
|
That is a bit strange - I believe I can reproduce your problem by renaming my .casarc file: Can you please give Ben's solution a try? I think that you just need .casarc to point at something appropriate. Edit: Tried this myself and made .casarc point to |
Beta Was this translation helpful? Give feedback.
-
|
In case it was unclear, |
Beta Was this translation helpful? Give feedback.
-
|
See #330. |
Beta Was this translation helpful? Give feedback.
-
|
My 2 cents on the matter is that you may get away with ignoring parallactic angle rotation during selfcal if you have corrected for it during 1GC (I think getting the order of the terms wrong at that stage will end up being a second or higher order effect). But you may also be able to do the full 1GC chain with QuartiCal now (I'm not sure what Df and Xf refer to above) thereby avoiding the need to calibrate corrected data, which is almost surely always dodgy |
Beta Was this translation helpful? Give feedback.
-
|
For context Df = Leakage, and Xf = cross-hand phase |
Beta Was this translation helpful? Give feedback.
-
|
If you are running Ubuntu you can install casacore-data and set .casarc
accordingly (as you would need to do for CASA installations).
Alternatively you can grab ephemeris directly from the NRAO:
mkdir -p /usr/share/casacore/data/
/usr/share/casacore/data/
rsync -avz rsync://casa-rsync.nrao.edu/casa-data .
…On Thu, Feb 22, 2024 at 10:46 PM AKHughes1994 ***@***.***> wrote:
Thanks for this; I agree your method would make sense, but there is a
caveat.
It looks like if casacore has an outdated LEAP table the p_jones terms
cause QC to fail,
image.png (view on web)
<https://github.com/ratt-ru/QuartiCal/assets/49698839/c0f45e4f-74b7-44b9-84d6-8551e863b5cd>
Still trying to figure out how to get around that. Any suggestions?
For context, I am just using the default version that gets installed using
pip (v0.2.2)
—
Reply to this email directly, view it on GitHub
<#328 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AB4RE6RJGN2LRHPME4LVX2TYU6VCTAVCNFSM6AAAAABDRZNJICVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4DKNRRGUZDQ>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
--
--
Benjamin Hugo
|
Beta Was this translation helpful? Give feedback.
-
|
Just FYI, the casadata wheel is ~350MB and uncompresses to 833MB ~/venv/lib/python3.10/site-packages/casadata/__data__$ du -hd 1
80K ./gui
415M ./ephemerides
25M ./nrao
9.1M ./geodetic
45M ./demo
40M ./catalogs
301M ./alma
833M .by contrast $ wget ftp://ftp.astron.nl/outgoing/Measures/WSRT_Measures.ztar
--2024-02-23 09:22:38-- ftp://ftp.astron.nl/outgoing/Measures/WSRT_Measures.ztar
=> ‘WSRT_Measures.ztar’
Resolving ftp.astron.nl (ftp.astron.nl)... 192.87.1.38
Connecting to ftp.astron.nl (ftp.astron.nl)|192.87.1.38|:21... connected.
Logging in as anonymous ... Logged in!
==> SYST ... done. ==> PWD ... done.
==> TYPE I ... done. ==> CWD (1) /outgoing/Measures ... done.
==> SIZE WSRT_Measures.ztar ... 20488360
==> PASV ... done. ==> RETR WSRT_Measures.ztar ... done.
Length: 20488360 (20M) (unauthoritative)
WSRT_Measures.ztar 100%[=====================================================================================================================================================================================================================================================================================>] 19.54M 5.52MB/s in 4.3s
2024-02-23 09:22:45 (4.52 MB/s) - ‘WSRT_Measures.ztar’ saved [20488360] |
Beta Was this translation helpful? Give feedback.
-
|
It seems that there is substantially more stuff in the |
Beta Was this translation helpful? Give feedback.
-
|
I'm intending to auto download the smaller measures in arcae at some point. |
Beta Was this translation helpful? Give feedback.
-
|
As a follow-up, I grabbed ephemeris directly using the method from @bennahugo, and it solved my problem! Sorry for the delayed reply and Thanks! |
Beta Was this translation helpful? Give feedback.
-
|
Glad that it fixed the problem! I am still thinking about the best way to handle this in general without requiring additional setup from users. |
Beta Was this translation helpful? Give feedback.
-
Hi,
I have a question about the terminology of the title parameters. Namely the help text,
input_model.apply_p_jones Determines whether P-Jones (parallactic angle rotation) is applied to the model. This affects both measurement set columns and predicted components. Care must taken when using this option and output.apply_p_jones_inv. Default: FalseIf I understand this correctly, is this saying that it will apply the parallactic angle rotation to the model AND the input data column visibilities?
For a fuller context, I am trying to perform full-polarization self-calibration.
I have a DATA column that has been split from a larger ms file and has all the Gain terms applied (i.e., K, Gp, Ga, BP, Df, Xf) with parallactic angle correction applied (This is done in casa). I then image the DATA column, using the IQUV model to population MODEL. From here, if I want to self-calibrate.
Would I set
input_model.apply_p_jones = Trueto re-rotate according to the parallactic angle? Then, also setoutput.apply_p_jones_inv = Trueto re-apply the parallactic angle correction to the now self-calibrated corrected visibilities?Am I on the right track here, or have I misunderstood the purpose of these flags?
Thanks for the help,
Andrew
Beta Was this translation helpful? Give feedback.
All reactions