orphaned-rec vs. poor-febe

[tcromartie]

Is there a difference? Did we switch the name, or do they actually have two different functions? It's probably not a good idea to continue calling them "orphaned" anyway...

[rossjjennings]

In principle, there is a difference between orphaned-rec and poor-febe, but it's a very minor one. In the 15-year data set, we applied the orphaned-rec cut if there were fewer than three epochs available from that receiver/backend combination, and we created poor-febe later, for cases that technically passed this cut, but where we still didn't have enough data to constrain the white noise parameters well.

Here's Joe Swiggum explaining this, from timing_analysis PR !490 (link requires NANOGrav login):

In completing several other tasks recently, it's become apparent that a handful of pulsars have smatterings of TOAs that do not drop below the 3-epoch threshold where "orphaned" would apply, but they should clearly be cut due to low SNR, or other reasons. In most cases, these appear to be very short test campaigns with a given frontend-backend ("febe") combination, abandoned due to the pulsar's poor performance.

Here, I have implemented a way to explicitly remove such TOAs early on (before outlier runs) and I've added info to yamls where applicable. I also fixed two bugs here -- instances of TOA flags previously being set to integers/floats (now need to be strs).

[mtlam]

I agree with the rationale for both, but one thought is that perhaps we should streamline and just have something like a "bad-febe" flag and call it a day. The rationale for why it is bad could be whatever reason, but then we only have to maintain one thing and not two. Or, if there's some other reason in the future (I can't immediately think of why) then we wouldn't need to maintain three.

[tcromartie]

Thanks for digging that up Ross. I think it could get confusing to merge the two tags without some ability to make comments next to the rx in the yaml (e.g. what we do for manually excised TOAs) because in one case the cut is due to a febe combo not meeting a quantitative threshold, and the other is basically due to "vibes," so it makes it harder to track why something was actually removed. That said, I'd be fine with a new tag and just making sure we get comments working for that yaml field.