Non-circular/elliptical positional errors

[fcivano]

Hi, we are working with a Chandra X-ray catalog where we have elliptical positional errors rather than circular positional errors. Is there a way we could use them? Thanks

[JohannesBuchner]

The short answer is that NWAY currently only supports circular answers.

I was looking into how hard it would be to include elliptical errors. Reading the CSC information http://cxc.harvard.edu/csc/why/err_ellipse_msc.html, they use local tangential planes, which is defined at the ra/dec position and oriented along the local declination axis. The error ellipse is given with major and minor axis angular size and direction relative to the plane.

In Pineau+17, the necessary formulae are developed which could extend NWAY to elliptical errors, but as far as I understand their formalism, they use a equidistant projection and keep working in RA/Dec, which
means that the formulae would not work at the poles (haversine formula, etc). Something more generic (but similar) would have to be developed, which can digest ingredients like in the CSC, or those ingredients would
have to be transformed into the Pineau+17 formalism.

Either way, we would be interested to include it eventually once it is clear how to do it; it will be needed for eROSITA.

For a short-term solution, a circular error of radius=sqrt(ra_err^2+dec_err^2), should suffice for Chandra.

[fxpineau]

@JohannesBuchner (and @fcivano ) I am discovering this thread: better late than never.
The formalism I am using in the paper also consider a local tangential plane as described in 4.1.
So (RA, Dec) becomes (x, y). It works perfectly well at the poles, and one has to be careful at the non-aligned North directions for nearby sources near to the poles.

[JohannesBuchner]

btw, elliptical errors are implemented in NWAY, but not documented yet.