search

ratt-ru / QuartiCal

CubiCal, but with greater power.
MIT License
8 stars 4 forks source link

TEC solver #105

Closed JSKenyon closed 3 years ago

JSKenyon commented 3 years ago

This PR adds a TEC(-ish) solver. Essentially this solves for a gain parameterised as np.exp(1j*(1/nu)*a + b). The reason I say TEC-ish is because I am not entirely sure what units a will be in. Typical values seem to be in the 1e8 range, but that is way smaller than a TECU. @o-smirnov It definitely runs on your data, so feel free to try it out. Just set solver.type=tec.

bennahugo commented 3 years ago

John it is a Nu^2 effect in rotation measure - I assume that you are trying to fit for vertical TEC based off U and Q here? See BJ Burn 1966 for details

On Fri, Aug 20, 2021 at 11:44 AM JSKenyon @.***> wrote:

This PR adds a TEC(-ish) solver. Essentially this solves for a gain parameterised as np.exp(1j(1/nu)a + b). The reason I say TEC-ish is because I am not entirely sure what units a will be in. Typical values seem to be in the 1e8 range, but that is way smaller than a TECU. @o-smirnov https://github.com/o-smirnov It definitely runs on your data, so feel free to try it out. Just set solver.type=tec.

You can view, comment on, or merge this pull request online at:

https://github.com/JSKenyon/QuartiCal/pull/105 Commit Summary

  • Add basic tec solver which runs but doesn't yet do the correct thing.
  • TEC solver possibly working.

File Changes

Patch Links:

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/JSKenyon/QuartiCal/pull/105, or unsubscribe https://github.com/notifications/unsubscribe-auth/AB4RE6SKK4DIR6LL2UOTYRTT5YPXDANCNFSM5CP7EEPA . 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&utm_campaign=notification-email .

--

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

JSKenyon commented 3 years ago

I leave it to you and @o-smirnov to duke it out. Changing it to 1/nu**2 is trivial, so I am happy to do so once there is a conclusion.

o-smirnov commented 3 years ago

It is ~1/nu in on-diagonal phase, which @JSKenyon is implementing. It is ~1/nu^2 in QU rotation, which we're not implementing yet

bennahugo commented 3 years ago

Sure - the second-order effects are seen on the diagonal - they are almost impossible to model though unless you know the source is compact and have a model beforehand, so the usefulness of such a solver is debatable. I would argue the effect should be fitted for in RM instead where it is much stronger? The previous implementation in cubical was of limited use - the ionospheric effects are definitely there, but you cannot see them on the diagonal that well (unless you look at Q!)

bennahugo commented 3 years ago

But this is just a recommendation