incorrect mode identification for main sequence stars

[darthoctopus]

I'm trying pbjam with a main sequence star (16 Cyg A; KIC 12069424), using a power spectrum from the KASOC time series, and with stellar parameters instantiated as follows:

star = pb.star(kic, pg, 
               [2200, 50], [150,50], teff=[5830, 11], bp_rp=[0.8201175, 0.1])

calling star() returns the following l = 0,2 mode identifications:

The l=1 ridge is being misidentified as l=0,2 modes, with a value of the small separation that is likely determined by noise peaks (2 μHz rather than ~8 μHz).

[nielsenmb]

Hi Joel, thanks trying this out. The main-sequence part of the prior is still a bit sparse.

Can you try increasing the error on your Teff, \pm11 K is very small. I don't know if that'll affect anything, but we do have a Gaussian prior on Teff on top of the KDE, so if it's too narrow it might try to put epsilon at a wrong value.

Also, when I was working on the main-sequence targets I found that increasing the bandwidth (to 2-4) of the KDE sometimes helped. This is the bw_fac optional parameter when you call session or star (after initializing it). Once we get enough targets on the main sequence this should hopefully no longer be an issue.

[grd349]

Hey @darthoctopus ,

Thanks for looking at this. I'm trying to recreate this but Lightkurve doesn't like 16 Cyg (complains about multiple targets). I guess you are using a custom lightcurve. If you share the light curve (or even your notebook) I'll have a go.

First out the gate as a suspect here is your starting Dnu. PBjam is really designed to work only if you give a decent estimate of Dnu (& Numax) to the code. I think the results you are seeing are simply a manifestation of your large uncertainty in Dnu. Could you retry with dnu=[103.4, 1.0]?

Ta,

[darthoctopus]

Thanks @nielsenmb and @grd349 for the suggestions. As @grd349 suspects, pbjam is able to identify the modes correctly with a more constrained value of Δν.