[daip] about the task EVAUV {External}

[Makoto Miyoshi]

Dear Eric Greisen,

I understand the meaning of closures in the task of EVAUV.

Thanks,

Miyoshi

________________________________
差出人: Eric Greisen <egreisen at nrao.edu>
送信日時: 2024年8月8日 1:13
宛先: Makoto Miyoshi <makoto.miyoshi at nao.ac.jp>
CC: daip <daip at nrao.edu>
件名: Re: about the task EVAUV {External}

You are mis-reading the help file.  The closure computation is done on the GAIN data, which is the observed divided by the model, not subtracted.  And it is true that closure phase and amplitude of the observed data and of the model data will be different than the same parameters of the GAIN file.  The task is simply trying to report how close the model is to the observations.  The statistics of the GAIN file are used to evaluate how well particular sources meet the requirements of calibration sources.  I am adding the following paragraph to the EVAUV help file:

         The use of closure statistics to evaluate potential calibration
         sources is discussed in detail in the paper Xu, M.H., Anderson,
         J.M., Heinkelmann, R., Lunz, S., Schuh, H., Wang, G.L., 2019,
         "Structure Effects for 3417 Celestial Reference Frame Radio
         Sources", Ap. J. Suppl., 242:5.  Available via:
              https://doi.org/10.3847/1538-4365/ab16ea

Eric Greisen
________________________________
From: Makoto Miyoshi <makoto.miyoshi at nao.ac.jp>
Sent: Wednesday, August 7, 2024 1:16 AM
To: Eric Greisen <egreisen at nrao.edu>
Cc: daip <daip at nrao.edu>
Subject: about the task EVAUV {External}


Dear Sirs,

I am emailing because I have a question about the closure residual calculation in EVAUV.

The description of the task, EVAUV, states the following:
"The task will now compute and report closure phase and　amplitude statistics from the "visibilties" of the gain data set (produced by division of the input data by the model, not subtracting (1,0))."
http://www.aips.nrao.edu/cgi-bin/ZXHLP2.PL?EVAUV

In other words, it can be understood for me that at first make the difference between the observed visibility and the visibility corresponding to the image is calculated, and then the closures are calculated from the difference visibilities.

Wouldn't that give a different result than the case we calculated the closures from each of the observed visibilities, visibilities that corresponding to the model image, and then find the mean and standard deviation from the difference between them?

It is confirmed that this is different from the statistics of the difference between the closure phases output using CLPLT from the observed data and those corresponding to a model image.

One of them seems to be wrong. If my understanding is wrong, I am sorry.


Yours sincerely

Makoto Miyoshi
NAOJ, Mitaka


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listmgr.nrao.edu/pipermail/daip/attachments/20240809/25f1e57f/attachment.html>