search

sot / sparkles

ACA review of star catalogs from proseco
BSD 3-Clause "New" or "Revised" License
0 stars 1 forks source link

Revisit guide star selection rules for close spoilers #139

Open taldcroft opened 4 years ago

taldcroft commented 4 years ago

See discussion below.

I remember looking into this and it wasn't completely trivial to do. This overlaps a bit with #34 since included stars are the only ones where this can happen (apart from outright bugs in proseco).

jeanconn commented 4 years ago

Thanks! And you fixed it a bunch with #133 but if the ASP1 > 0 star in 22634 can sneak through there's still some work to be done. In the meantime, #34 would certainly help.

taldcroft commented 4 years ago

Ahh, I knew this seemed familiar, and better still that I actually solved this already!

The resolution now is that the current guide star selection rules allow up to ASPQ1 < 20, corresponding to an offset of up to 1.0 arcsec. https://github.com/sot/proseco/blob/78611dd7a0d5573f97b4753b4dfaf44ebc1f944f/proseco/guide.py#L469

So maybe this just means you were being quite observant in pre-review to detect ASPQ1=4 but forgot (along with me) that we formally allow this and did not need to have any back/forth discussion? What was your process for noting something wrong about that star?

jeanconn commented 4 years ago

Since I came at this somewhat trying to figure out why the star had not been selected, my process was just to review the report

https://icxc.cfa.harvard.edu/aspect/SFE/2020/MAR3020/prelim2/MAR3020P2_2020_085_10_24_07_831_sparkles/obs22634/guide/index.html#39716680

So I suppose the ASPQ1 warning is a bit of a red herring but relaxing the limit doesn't help to get the star if the other spoiler check(s) are still in effect though the stages. So my subject line to you about ASPQ >0 was based on a poor recollection but fundamentally the manually included star was still one that looks like it wouldn't have been a candidate at any stage?

taldcroft commented 4 years ago

It looks like this rule is not helping out here:

Not selected stage 4: Cand 39716680 spoiled by 39716681, too close (1.0) pix for magdiff (-3.0)

As I computed and as reflected in the ASPQ1 value, the actual offset from this star is around 0.2 to 0.3 arcsec.

@jeanconn - you've spent more time thinking about guide star rules, is there a downside to relying entirely on ASPQ1 for close spoilers? By "close" I mean any spoiler star that is within the radius which is included in the ASPQ1 calculations.